From b365ad692ebcb74ea5807e9385163b2dfad9aab5 Mon Sep 17 00:00:00 2001 From: Love Billingskog Nyberg Date: Tue, 15 Mar 2022 10:23:11 +0000 Subject: [PATCH] Adding new schemas --- master-local/_definitions.json | 9 +++++++-- master-local/topologyspreadconstraint-v1.json | 12 ++++++++++-- master-local/topologyspreadconstraint.json | 12 ++++++++++-- master-standalone-strict/_definitions.json | 9 +++++++-- master-standalone-strict/cronjob-batch-v1.json | 12 ++++++++++-- master-standalone-strict/cronjob-batch-v1beta1.json | 12 ++++++++++-- master-standalone-strict/cronjob.json | 12 ++++++++++-- master-standalone-strict/cronjoblist-batch-v1.json | 12 ++++++++++-- .../cronjoblist-batch-v1beta1.json | 12 ++++++++++-- master-standalone-strict/cronjoblist.json | 12 ++++++++++-- master-standalone-strict/cronjobspec-batch-v1.json | 12 ++++++++++-- .../cronjobspec-batch-v1beta1.json | 12 ++++++++++-- master-standalone-strict/cronjobspec.json | 12 ++++++++++-- master-standalone-strict/daemonset-apps-v1.json | 12 ++++++++++-- master-standalone-strict/daemonset.json | 12 ++++++++++-- master-standalone-strict/daemonsetlist-apps-v1.json | 12 ++++++++++-- master-standalone-strict/daemonsetlist.json | 12 ++++++++++-- master-standalone-strict/daemonsetspec-apps-v1.json | 12 ++++++++++-- master-standalone-strict/daemonsetspec.json | 12 ++++++++++-- master-standalone-strict/deployment-apps-v1.json | 12 ++++++++++-- master-standalone-strict/deployment.json | 12 ++++++++++-- master-standalone-strict/deploymentlist-apps-v1.json | 12 ++++++++++-- master-standalone-strict/deploymentlist.json | 12 ++++++++++-- master-standalone-strict/deploymentspec-apps-v1.json | 12 ++++++++++-- master-standalone-strict/deploymentspec.json | 12 ++++++++++-- master-standalone-strict/job-batch-v1.json | 12 ++++++++++-- master-standalone-strict/job.json | 12 ++++++++++-- master-standalone-strict/joblist-batch-v1.json | 12 ++++++++++-- master-standalone-strict/joblist.json | 12 ++++++++++-- master-standalone-strict/jobspec-batch-v1.json | 12 ++++++++++-- master-standalone-strict/jobspec.json | 12 ++++++++++-- .../jobtemplatespec-batch-v1.json | 12 ++++++++++-- .../jobtemplatespec-batch-v1beta1.json | 12 ++++++++++-- master-standalone-strict/jobtemplatespec.json | 12 ++++++++++-- master-standalone-strict/pod-v1.json | 12 ++++++++++-- master-standalone-strict/pod.json | 12 ++++++++++-- master-standalone-strict/podlist-v1.json | 12 ++++++++++-- master-standalone-strict/podlist.json | 12 ++++++++++-- master-standalone-strict/podspec-v1.json | 12 ++++++++++-- master-standalone-strict/podspec.json | 12 ++++++++++-- master-standalone-strict/podtemplate-v1.json | 12 ++++++++++-- master-standalone-strict/podtemplate.json | 12 ++++++++++-- master-standalone-strict/podtemplatelist-v1.json | 12 ++++++++++-- master-standalone-strict/podtemplatelist.json | 12 ++++++++++-- master-standalone-strict/podtemplatespec-v1.json | 12 ++++++++++-- master-standalone-strict/podtemplatespec.json | 12 ++++++++++-- master-standalone-strict/replicaset-apps-v1.json | 12 ++++++++++-- master-standalone-strict/replicaset.json | 12 ++++++++++-- master-standalone-strict/replicasetlist-apps-v1.json | 12 ++++++++++-- master-standalone-strict/replicasetlist.json | 12 ++++++++++-- master-standalone-strict/replicasetspec-apps-v1.json | 12 ++++++++++-- master-standalone-strict/replicasetspec.json | 12 ++++++++++-- .../replicationcontroller-v1.json | 12 ++++++++++-- master-standalone-strict/replicationcontroller.json | 12 ++++++++++-- .../replicationcontrollerlist-v1.json | 12 ++++++++++-- .../replicationcontrollerlist.json | 12 ++++++++++-- .../replicationcontrollerspec-v1.json | 12 ++++++++++-- .../replicationcontrollerspec.json | 12 ++++++++++-- master-standalone-strict/statefulset-apps-v1.json | 12 ++++++++++-- master-standalone-strict/statefulset.json | 12 ++++++++++-- .../statefulsetlist-apps-v1.json | 12 ++++++++++-- master-standalone-strict/statefulsetlist.json | 12 ++++++++++-- .../statefulsetspec-apps-v1.json | 12 ++++++++++-- master-standalone-strict/statefulsetspec.json | 12 ++++++++++-- .../topologyspreadconstraint-v1.json | 12 ++++++++++-- .../topologyspreadconstraint.json | 12 ++++++++++-- master-standalone/_definitions.json | 9 +++++++-- master-standalone/cronjob-batch-v1.json | 12 ++++++++++-- master-standalone/cronjob-batch-v1beta1.json | 12 ++++++++++-- master-standalone/cronjob.json | 12 ++++++++++-- master-standalone/cronjoblist-batch-v1.json | 12 ++++++++++-- master-standalone/cronjoblist-batch-v1beta1.json | 12 ++++++++++-- master-standalone/cronjoblist.json | 12 ++++++++++-- master-standalone/cronjobspec-batch-v1.json | 12 ++++++++++-- master-standalone/cronjobspec-batch-v1beta1.json | 12 ++++++++++-- master-standalone/cronjobspec.json | 12 ++++++++++-- master-standalone/daemonset-apps-v1.json | 12 ++++++++++-- master-standalone/daemonset.json | 12 ++++++++++-- master-standalone/daemonsetlist-apps-v1.json | 12 ++++++++++-- master-standalone/daemonsetlist.json | 12 ++++++++++-- master-standalone/daemonsetspec-apps-v1.json | 12 ++++++++++-- master-standalone/daemonsetspec.json | 12 ++++++++++-- master-standalone/deployment-apps-v1.json | 12 ++++++++++-- master-standalone/deployment.json | 12 ++++++++++-- master-standalone/deploymentlist-apps-v1.json | 12 ++++++++++-- master-standalone/deploymentlist.json | 12 ++++++++++-- master-standalone/deploymentspec-apps-v1.json | 12 ++++++++++-- master-standalone/deploymentspec.json | 12 ++++++++++-- master-standalone/job-batch-v1.json | 12 ++++++++++-- master-standalone/job.json | 12 ++++++++++-- master-standalone/joblist-batch-v1.json | 12 ++++++++++-- master-standalone/joblist.json | 12 ++++++++++-- master-standalone/jobspec-batch-v1.json | 12 ++++++++++-- master-standalone/jobspec.json | 12 ++++++++++-- master-standalone/jobtemplatespec-batch-v1.json | 12 ++++++++++-- master-standalone/jobtemplatespec-batch-v1beta1.json | 12 ++++++++++-- master-standalone/jobtemplatespec.json | 12 ++++++++++-- master-standalone/pod-v1.json | 12 ++++++++++-- master-standalone/pod.json | 12 ++++++++++-- master-standalone/podlist-v1.json | 12 ++++++++++-- master-standalone/podlist.json | 12 ++++++++++-- master-standalone/podspec-v1.json | 12 ++++++++++-- master-standalone/podspec.json | 12 ++++++++++-- master-standalone/podtemplate-v1.json | 12 ++++++++++-- master-standalone/podtemplate.json | 12 ++++++++++-- master-standalone/podtemplatelist-v1.json | 12 ++++++++++-- master-standalone/podtemplatelist.json | 12 ++++++++++-- master-standalone/podtemplatespec-v1.json | 12 ++++++++++-- master-standalone/podtemplatespec.json | 12 ++++++++++-- master-standalone/replicaset-apps-v1.json | 12 ++++++++++-- master-standalone/replicaset.json | 12 ++++++++++-- master-standalone/replicasetlist-apps-v1.json | 12 ++++++++++-- master-standalone/replicasetlist.json | 12 ++++++++++-- master-standalone/replicasetspec-apps-v1.json | 12 ++++++++++-- master-standalone/replicasetspec.json | 12 ++++++++++-- master-standalone/replicationcontroller-v1.json | 12 ++++++++++-- master-standalone/replicationcontroller.json | 12 ++++++++++-- master-standalone/replicationcontrollerlist-v1.json | 12 ++++++++++-- master-standalone/replicationcontrollerlist.json | 12 ++++++++++-- master-standalone/replicationcontrollerspec-v1.json | 12 ++++++++++-- master-standalone/replicationcontrollerspec.json | 12 ++++++++++-- master-standalone/statefulset-apps-v1.json | 12 ++++++++++-- master-standalone/statefulset.json | 12 ++++++++++-- master-standalone/statefulsetlist-apps-v1.json | 12 ++++++++++-- master-standalone/statefulsetlist.json | 12 ++++++++++-- master-standalone/statefulsetspec-apps-v1.json | 12 ++++++++++-- master-standalone/statefulsetspec.json | 12 ++++++++++-- master-standalone/topologyspreadconstraint-v1.json | 12 ++++++++++-- master-standalone/topologyspreadconstraint.json | 12 ++++++++++-- master/_definitions.json | 9 +++++++-- master/topologyspreadconstraint-v1.json | 12 ++++++++++-- master/topologyspreadconstraint.json | 12 ++++++++++-- 132 files changed, 1308 insertions(+), 264 deletions(-) diff --git a/master-local/_definitions.json b/master-local/_definitions.json index 88039ea242..a90419e785 100644 --- a/master-local/_definitions.json +++ b/master-local/_definitions.json @@ -10746,12 +10746,17 @@ "description": "LabelSelector is used to find matching pods. Pods that match this label selector are counted to determine the number of pods in their corresponding topology domain." }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "format": "int32", + "type": "integer" + }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", "format": "int32", "type": "integer" }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-local/topologyspreadconstraint-v1.json b/master-local/topologyspreadconstraint-v1.json index 3356e40599..3674681640 100644 --- a/master-local/topologyspreadconstraint-v1.json +++ b/master-local/topologyspreadconstraint-v1.json @@ -6,7 +6,15 @@ "description": "LabelSelector is used to find matching pods. Pods that match this label selector are counted to determine the number of pods in their corresponding topology domain." }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", "format": "int32", "type": [ "integer", @@ -14,7 +22,7 @@ ] }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": [ "string", "null" diff --git a/master-local/topologyspreadconstraint.json b/master-local/topologyspreadconstraint.json index 3356e40599..3674681640 100644 --- a/master-local/topologyspreadconstraint.json +++ b/master-local/topologyspreadconstraint.json @@ -6,7 +6,15 @@ "description": "LabelSelector is used to find matching pods. Pods that match this label selector are counted to determine the number of pods in their corresponding topology domain." }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", "format": "int32", "type": [ "integer", @@ -14,7 +22,7 @@ ] }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": [ "string", "null" diff --git a/master-standalone-strict/_definitions.json b/master-standalone-strict/_definitions.json index 8e4248924f..f4aabf3b46 100644 --- a/master-standalone-strict/_definitions.json +++ b/master-standalone-strict/_definitions.json @@ -11088,12 +11088,17 @@ "description": "LabelSelector is used to find matching pods. Pods that match this label selector are counted to determine the number of pods in their corresponding topology domain." }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "format": "int32", + "type": "integer" + }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", "format": "int32", "type": "integer" }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/cronjob-batch-v1.json b/master-standalone-strict/cronjob-batch-v1.json index 53159f3894..12248445e9 100644 --- a/master-standalone-strict/cronjob-batch-v1.json +++ b/master-standalone-strict/cronjob-batch-v1.json @@ -7997,12 +7997,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/cronjob-batch-v1beta1.json b/master-standalone-strict/cronjob-batch-v1beta1.json index c409e9897f..2bc304b90f 100644 --- a/master-standalone-strict/cronjob-batch-v1beta1.json +++ b/master-standalone-strict/cronjob-batch-v1beta1.json @@ -7992,12 +7992,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/cronjob.json b/master-standalone-strict/cronjob.json index 287a370960..d66e820528 100644 --- a/master-standalone-strict/cronjob.json +++ b/master-standalone-strict/cronjob.json @@ -7989,12 +7989,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/cronjoblist-batch-v1.json b/master-standalone-strict/cronjoblist-batch-v1.json index 3840157bb2..cd577d1db9 100644 --- a/master-standalone-strict/cronjoblist-batch-v1.json +++ b/master-standalone-strict/cronjoblist-batch-v1.json @@ -8012,12 +8012,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/cronjoblist-batch-v1beta1.json b/master-standalone-strict/cronjoblist-batch-v1beta1.json index 9fae0a9755..7e4b9bebc4 100644 --- a/master-standalone-strict/cronjoblist-batch-v1beta1.json +++ b/master-standalone-strict/cronjoblist-batch-v1beta1.json @@ -8007,12 +8007,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/cronjoblist.json b/master-standalone-strict/cronjoblist.json index b341341327..f2fd11b8fe 100644 --- a/master-standalone-strict/cronjoblist.json +++ b/master-standalone-strict/cronjoblist.json @@ -8001,12 +8001,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/cronjobspec-batch-v1.json b/master-standalone-strict/cronjobspec-batch-v1.json index 01a4b8e6ef..187d90e5bb 100644 --- a/master-standalone-strict/cronjobspec-batch-v1.json +++ b/master-standalone-strict/cronjobspec-batch-v1.json @@ -7720,12 +7720,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/cronjobspec-batch-v1beta1.json b/master-standalone-strict/cronjobspec-batch-v1beta1.json index 383eda5e30..6f426c4705 100644 --- a/master-standalone-strict/cronjobspec-batch-v1beta1.json +++ b/master-standalone-strict/cronjobspec-batch-v1beta1.json @@ -7715,12 +7715,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/cronjobspec.json b/master-standalone-strict/cronjobspec.json index 383eda5e30..6f426c4705 100644 --- a/master-standalone-strict/cronjobspec.json +++ b/master-standalone-strict/cronjobspec.json @@ -7715,12 +7715,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/daemonset-apps-v1.json b/master-standalone-strict/daemonset-apps-v1.json index 7684535ae3..723a052cf8 100644 --- a/master-standalone-strict/daemonset-apps-v1.json +++ b/master-standalone-strict/daemonset-apps-v1.json @@ -7677,12 +7677,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/daemonset.json b/master-standalone-strict/daemonset.json index 8694b9e559..03e88c79a4 100644 --- a/master-standalone-strict/daemonset.json +++ b/master-standalone-strict/daemonset.json @@ -7674,12 +7674,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/daemonsetlist-apps-v1.json b/master-standalone-strict/daemonsetlist-apps-v1.json index 91a9526b10..efc04bc383 100644 --- a/master-standalone-strict/daemonsetlist-apps-v1.json +++ b/master-standalone-strict/daemonsetlist-apps-v1.json @@ -7692,12 +7692,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/daemonsetlist.json b/master-standalone-strict/daemonsetlist.json index a85c3186f7..c32878beb3 100644 --- a/master-standalone-strict/daemonsetlist.json +++ b/master-standalone-strict/daemonsetlist.json @@ -7686,12 +7686,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/daemonsetspec-apps-v1.json b/master-standalone-strict/daemonsetspec-apps-v1.json index 776f829467..3ab62a2061 100644 --- a/master-standalone-strict/daemonsetspec-apps-v1.json +++ b/master-standalone-strict/daemonsetspec-apps-v1.json @@ -7403,12 +7403,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/daemonsetspec.json b/master-standalone-strict/daemonsetspec.json index 776f829467..3ab62a2061 100644 --- a/master-standalone-strict/daemonsetspec.json +++ b/master-standalone-strict/daemonsetspec.json @@ -7403,12 +7403,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/deployment-apps-v1.json b/master-standalone-strict/deployment-apps-v1.json index 0eb17abdc4..1055f1cf49 100644 --- a/master-standalone-strict/deployment-apps-v1.json +++ b/master-standalone-strict/deployment-apps-v1.json @@ -7763,12 +7763,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/deployment.json b/master-standalone-strict/deployment.json index 95167ebd3e..57786031cc 100644 --- a/master-standalone-strict/deployment.json +++ b/master-standalone-strict/deployment.json @@ -7760,12 +7760,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/deploymentlist-apps-v1.json b/master-standalone-strict/deploymentlist-apps-v1.json index 5628fd92ae..2071f0e6af 100644 --- a/master-standalone-strict/deploymentlist-apps-v1.json +++ b/master-standalone-strict/deploymentlist-apps-v1.json @@ -7778,12 +7778,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/deploymentlist.json b/master-standalone-strict/deploymentlist.json index a87fa3bab7..463e5ddb7e 100644 --- a/master-standalone-strict/deploymentlist.json +++ b/master-standalone-strict/deploymentlist.json @@ -7772,12 +7772,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/deploymentspec-apps-v1.json b/master-standalone-strict/deploymentspec-apps-v1.json index ba6810e539..3d9a4d0a6e 100644 --- a/master-standalone-strict/deploymentspec-apps-v1.json +++ b/master-standalone-strict/deploymentspec-apps-v1.json @@ -7489,12 +7489,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/deploymentspec.json b/master-standalone-strict/deploymentspec.json index ba6810e539..3d9a4d0a6e 100644 --- a/master-standalone-strict/deploymentspec.json +++ b/master-standalone-strict/deploymentspec.json @@ -7489,12 +7489,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/job-batch-v1.json b/master-standalone-strict/job-batch-v1.json index 8e346facaa..c4d29b4142 100644 --- a/master-standalone-strict/job-batch-v1.json +++ b/master-standalone-strict/job-batch-v1.json @@ -7717,12 +7717,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/job.json b/master-standalone-strict/job.json index 93581db04b..45e1e4734f 100644 --- a/master-standalone-strict/job.json +++ b/master-standalone-strict/job.json @@ -7714,12 +7714,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/joblist-batch-v1.json b/master-standalone-strict/joblist-batch-v1.json index 18c462ee09..dd60d35057 100644 --- a/master-standalone-strict/joblist-batch-v1.json +++ b/master-standalone-strict/joblist-batch-v1.json @@ -7732,12 +7732,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/joblist.json b/master-standalone-strict/joblist.json index d182352e89..01efe79544 100644 --- a/master-standalone-strict/joblist.json +++ b/master-standalone-strict/joblist.json @@ -7726,12 +7726,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/jobspec-batch-v1.json b/master-standalone-strict/jobspec-batch-v1.json index 4f21ee17be..19cbe5de50 100644 --- a/master-standalone-strict/jobspec-batch-v1.json +++ b/master-standalone-strict/jobspec-batch-v1.json @@ -7440,12 +7440,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/jobspec.json b/master-standalone-strict/jobspec.json index 4f21ee17be..19cbe5de50 100644 --- a/master-standalone-strict/jobspec.json +++ b/master-standalone-strict/jobspec.json @@ -7440,12 +7440,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/jobtemplatespec-batch-v1.json b/master-standalone-strict/jobtemplatespec-batch-v1.json index bf0db5d70e..2d91dec2e2 100644 --- a/master-standalone-strict/jobtemplatespec-batch-v1.json +++ b/master-standalone-strict/jobtemplatespec-batch-v1.json @@ -7697,12 +7697,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/jobtemplatespec-batch-v1beta1.json b/master-standalone-strict/jobtemplatespec-batch-v1beta1.json index bf0db5d70e..2d91dec2e2 100644 --- a/master-standalone-strict/jobtemplatespec-batch-v1beta1.json +++ b/master-standalone-strict/jobtemplatespec-batch-v1beta1.json @@ -7697,12 +7697,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/jobtemplatespec.json b/master-standalone-strict/jobtemplatespec.json index bf0db5d70e..2d91dec2e2 100644 --- a/master-standalone-strict/jobtemplatespec.json +++ b/master-standalone-strict/jobtemplatespec.json @@ -7697,12 +7697,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/pod-v1.json b/master-standalone-strict/pod-v1.json index 418723e1e7..4621723bb3 100644 --- a/master-standalone-strict/pod-v1.json +++ b/master-standalone-strict/pod-v1.json @@ -7336,12 +7336,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/pod.json b/master-standalone-strict/pod.json index b72161e4b8..c28408c18a 100644 --- a/master-standalone-strict/pod.json +++ b/master-standalone-strict/pod.json @@ -7333,12 +7333,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/podlist-v1.json b/master-standalone-strict/podlist-v1.json index 29edafee79..5ce77362b7 100644 --- a/master-standalone-strict/podlist-v1.json +++ b/master-standalone-strict/podlist-v1.json @@ -7351,12 +7351,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/podlist.json b/master-standalone-strict/podlist.json index b3d4ef5cc5..51b7b05db3 100644 --- a/master-standalone-strict/podlist.json +++ b/master-standalone-strict/podlist.json @@ -7345,12 +7345,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/podspec-v1.json b/master-standalone-strict/podspec-v1.json index 1f866ebd9b..78d82eb251 100644 --- a/master-standalone-strict/podspec-v1.json +++ b/master-standalone-strict/podspec-v1.json @@ -7062,12 +7062,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/podspec.json b/master-standalone-strict/podspec.json index 1f866ebd9b..78d82eb251 100644 --- a/master-standalone-strict/podspec.json +++ b/master-standalone-strict/podspec.json @@ -7062,12 +7062,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/podtemplate-v1.json b/master-standalone-strict/podtemplate-v1.json index 0d12e21838..4b65571cb2 100644 --- a/master-standalone-strict/podtemplate-v1.json +++ b/master-standalone-strict/podtemplate-v1.json @@ -7593,12 +7593,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/podtemplate.json b/master-standalone-strict/podtemplate.json index cb8822fc15..9237e820ed 100644 --- a/master-standalone-strict/podtemplate.json +++ b/master-standalone-strict/podtemplate.json @@ -7590,12 +7590,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/podtemplatelist-v1.json b/master-standalone-strict/podtemplatelist-v1.json index ff304eef7b..aa9aef0f7b 100644 --- a/master-standalone-strict/podtemplatelist-v1.json +++ b/master-standalone-strict/podtemplatelist-v1.json @@ -7608,12 +7608,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/podtemplatelist.json b/master-standalone-strict/podtemplatelist.json index fea92f9b1f..0f466f377c 100644 --- a/master-standalone-strict/podtemplatelist.json +++ b/master-standalone-strict/podtemplatelist.json @@ -7602,12 +7602,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/podtemplatespec-v1.json b/master-standalone-strict/podtemplatespec-v1.json index 27f5eb145a..c18577c63a 100644 --- a/master-standalone-strict/podtemplatespec-v1.json +++ b/master-standalone-strict/podtemplatespec-v1.json @@ -7316,12 +7316,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/podtemplatespec.json b/master-standalone-strict/podtemplatespec.json index 27f5eb145a..c18577c63a 100644 --- a/master-standalone-strict/podtemplatespec.json +++ b/master-standalone-strict/podtemplatespec.json @@ -7316,12 +7316,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/replicaset-apps-v1.json b/master-standalone-strict/replicaset-apps-v1.json index 501be9e7e4..460b2506ba 100644 --- a/master-standalone-strict/replicaset-apps-v1.json +++ b/master-standalone-strict/replicaset-apps-v1.json @@ -7677,12 +7677,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/replicaset.json b/master-standalone-strict/replicaset.json index a28fa8c459..d1c75fc5fa 100644 --- a/master-standalone-strict/replicaset.json +++ b/master-standalone-strict/replicaset.json @@ -7674,12 +7674,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/replicasetlist-apps-v1.json b/master-standalone-strict/replicasetlist-apps-v1.json index 3a9d80b3d0..0222a96d29 100644 --- a/master-standalone-strict/replicasetlist-apps-v1.json +++ b/master-standalone-strict/replicasetlist-apps-v1.json @@ -7692,12 +7692,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/replicasetlist.json b/master-standalone-strict/replicasetlist.json index bf1a18a2ff..6ab26baa28 100644 --- a/master-standalone-strict/replicasetlist.json +++ b/master-standalone-strict/replicasetlist.json @@ -7686,12 +7686,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/replicasetspec-apps-v1.json b/master-standalone-strict/replicasetspec-apps-v1.json index d7fe79915e..021b60d8d0 100644 --- a/master-standalone-strict/replicasetspec-apps-v1.json +++ b/master-standalone-strict/replicasetspec-apps-v1.json @@ -7403,12 +7403,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/replicasetspec.json b/master-standalone-strict/replicasetspec.json index d7fe79915e..021b60d8d0 100644 --- a/master-standalone-strict/replicasetspec.json +++ b/master-standalone-strict/replicasetspec.json @@ -7403,12 +7403,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/replicationcontroller-v1.json b/master-standalone-strict/replicationcontroller-v1.json index 40732ace9c..3b78594a4b 100644 --- a/master-standalone-strict/replicationcontroller-v1.json +++ b/master-standalone-strict/replicationcontroller-v1.json @@ -7626,12 +7626,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/replicationcontroller.json b/master-standalone-strict/replicationcontroller.json index 9f8be53c1d..aea11c6538 100644 --- a/master-standalone-strict/replicationcontroller.json +++ b/master-standalone-strict/replicationcontroller.json @@ -7623,12 +7623,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/replicationcontrollerlist-v1.json b/master-standalone-strict/replicationcontrollerlist-v1.json index 1b421dc6ed..8d5da1fb3d 100644 --- a/master-standalone-strict/replicationcontrollerlist-v1.json +++ b/master-standalone-strict/replicationcontrollerlist-v1.json @@ -7641,12 +7641,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/replicationcontrollerlist.json b/master-standalone-strict/replicationcontrollerlist.json index 90ff32cda6..65ff6b38f0 100644 --- a/master-standalone-strict/replicationcontrollerlist.json +++ b/master-standalone-strict/replicationcontrollerlist.json @@ -7635,12 +7635,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/replicationcontrollerspec-v1.json b/master-standalone-strict/replicationcontrollerspec-v1.json index e275172109..b1e55f138e 100644 --- a/master-standalone-strict/replicationcontrollerspec-v1.json +++ b/master-standalone-strict/replicationcontrollerspec-v1.json @@ -7349,12 +7349,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/replicationcontrollerspec.json b/master-standalone-strict/replicationcontrollerspec.json index e275172109..b1e55f138e 100644 --- a/master-standalone-strict/replicationcontrollerspec.json +++ b/master-standalone-strict/replicationcontrollerspec.json @@ -7349,12 +7349,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/statefulset-apps-v1.json b/master-standalone-strict/statefulset-apps-v1.json index 336fdb393e..c62326808a 100644 --- a/master-standalone-strict/statefulset-apps-v1.json +++ b/master-standalone-strict/statefulset-apps-v1.json @@ -7724,12 +7724,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/statefulset.json b/master-standalone-strict/statefulset.json index 83e251b6be..11e78dc5fa 100644 --- a/master-standalone-strict/statefulset.json +++ b/master-standalone-strict/statefulset.json @@ -7721,12 +7721,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/statefulsetlist-apps-v1.json b/master-standalone-strict/statefulsetlist-apps-v1.json index e12da51b66..037f3e0dd5 100644 --- a/master-standalone-strict/statefulsetlist-apps-v1.json +++ b/master-standalone-strict/statefulsetlist-apps-v1.json @@ -7739,12 +7739,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/statefulsetlist.json b/master-standalone-strict/statefulsetlist.json index a604c0af66..2f131a7f1e 100644 --- a/master-standalone-strict/statefulsetlist.json +++ b/master-standalone-strict/statefulsetlist.json @@ -7733,12 +7733,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/statefulsetspec-apps-v1.json b/master-standalone-strict/statefulsetspec-apps-v1.json index 8693b4dc88..a20bfbc737 100644 --- a/master-standalone-strict/statefulsetspec-apps-v1.json +++ b/master-standalone-strict/statefulsetspec-apps-v1.json @@ -7453,12 +7453,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/statefulsetspec.json b/master-standalone-strict/statefulsetspec.json index 52a81654be..269a0d6c11 100644 --- a/master-standalone-strict/statefulsetspec.json +++ b/master-standalone-strict/statefulsetspec.json @@ -7453,12 +7453,20 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone-strict/topologyspreadconstraint-v1.json b/master-standalone-strict/topologyspreadconstraint-v1.json index e73c105c58..5f72a56b1e 100644 --- a/master-standalone-strict/topologyspreadconstraint-v1.json +++ b/master-standalone-strict/topologyspreadconstraint-v1.json @@ -70,7 +70,15 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", "format": "int32", "type": [ "integer", @@ -78,7 +86,7 @@ ] }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": [ "string", "null" diff --git a/master-standalone-strict/topologyspreadconstraint.json b/master-standalone-strict/topologyspreadconstraint.json index e73c105c58..5f72a56b1e 100644 --- a/master-standalone-strict/topologyspreadconstraint.json +++ b/master-standalone-strict/topologyspreadconstraint.json @@ -70,7 +70,15 @@ "additionalProperties": false }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", "format": "int32", "type": [ "integer", @@ -78,7 +86,7 @@ ] }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": [ "string", "null" diff --git a/master-standalone/_definitions.json b/master-standalone/_definitions.json index 88039ea242..a90419e785 100644 --- a/master-standalone/_definitions.json +++ b/master-standalone/_definitions.json @@ -10746,12 +10746,17 @@ "description": "LabelSelector is used to find matching pods. Pods that match this label selector are counted to determine the number of pods in their corresponding topology domain." }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "format": "int32", + "type": "integer" + }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", "format": "int32", "type": "integer" }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/cronjob-batch-v1.json b/master-standalone/cronjob-batch-v1.json index 5b591225b7..89ae4afe14 100644 --- a/master-standalone/cronjob-batch-v1.json +++ b/master-standalone/cronjob-batch-v1.json @@ -7794,12 +7794,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/cronjob-batch-v1beta1.json b/master-standalone/cronjob-batch-v1beta1.json index adb1f4c835..9a5d139066 100644 --- a/master-standalone/cronjob-batch-v1beta1.json +++ b/master-standalone/cronjob-batch-v1beta1.json @@ -7789,12 +7789,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/cronjob.json b/master-standalone/cronjob.json index a203872182..f36e146c66 100644 --- a/master-standalone/cronjob.json +++ b/master-standalone/cronjob.json @@ -7786,12 +7786,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/cronjoblist-batch-v1.json b/master-standalone/cronjoblist-batch-v1.json index 3b37fb1971..71e720ea6f 100644 --- a/master-standalone/cronjoblist-batch-v1.json +++ b/master-standalone/cronjoblist-batch-v1.json @@ -7809,12 +7809,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/cronjoblist-batch-v1beta1.json b/master-standalone/cronjoblist-batch-v1beta1.json index be976ea5f1..103296e13c 100644 --- a/master-standalone/cronjoblist-batch-v1beta1.json +++ b/master-standalone/cronjoblist-batch-v1beta1.json @@ -7804,12 +7804,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/cronjoblist.json b/master-standalone/cronjoblist.json index 2ab01832b3..9d8e602642 100644 --- a/master-standalone/cronjoblist.json +++ b/master-standalone/cronjoblist.json @@ -7798,12 +7798,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/cronjobspec-batch-v1.json b/master-standalone/cronjobspec-batch-v1.json index 06d47c42cb..e4b2e62465 100644 --- a/master-standalone/cronjobspec-batch-v1.json +++ b/master-standalone/cronjobspec-batch-v1.json @@ -7520,12 +7520,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/cronjobspec-batch-v1beta1.json b/master-standalone/cronjobspec-batch-v1beta1.json index ea9dbe012e..7bbf3bf97e 100644 --- a/master-standalone/cronjobspec-batch-v1beta1.json +++ b/master-standalone/cronjobspec-batch-v1beta1.json @@ -7515,12 +7515,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/cronjobspec.json b/master-standalone/cronjobspec.json index ea9dbe012e..7bbf3bf97e 100644 --- a/master-standalone/cronjobspec.json +++ b/master-standalone/cronjobspec.json @@ -7515,12 +7515,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/daemonset-apps-v1.json b/master-standalone/daemonset-apps-v1.json index 9c6a8c9d10..85cb237ba8 100644 --- a/master-standalone/daemonset-apps-v1.json +++ b/master-standalone/daemonset-apps-v1.json @@ -7477,12 +7477,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/daemonset.json b/master-standalone/daemonset.json index 7390c0f7c8..2854161410 100644 --- a/master-standalone/daemonset.json +++ b/master-standalone/daemonset.json @@ -7474,12 +7474,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/daemonsetlist-apps-v1.json b/master-standalone/daemonsetlist-apps-v1.json index e8ea83b3ce..ece22e2322 100644 --- a/master-standalone/daemonsetlist-apps-v1.json +++ b/master-standalone/daemonsetlist-apps-v1.json @@ -7492,12 +7492,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/daemonsetlist.json b/master-standalone/daemonsetlist.json index 5b0b5b0f0f..857a922d06 100644 --- a/master-standalone/daemonsetlist.json +++ b/master-standalone/daemonsetlist.json @@ -7486,12 +7486,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/daemonsetspec-apps-v1.json b/master-standalone/daemonsetspec-apps-v1.json index 8aafb15ab5..42a10184ae 100644 --- a/master-standalone/daemonsetspec-apps-v1.json +++ b/master-standalone/daemonsetspec-apps-v1.json @@ -7206,12 +7206,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/daemonsetspec.json b/master-standalone/daemonsetspec.json index 8aafb15ab5..42a10184ae 100644 --- a/master-standalone/daemonsetspec.json +++ b/master-standalone/daemonsetspec.json @@ -7206,12 +7206,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/deployment-apps-v1.json b/master-standalone/deployment-apps-v1.json index a4d3e17229..cde1681679 100644 --- a/master-standalone/deployment-apps-v1.json +++ b/master-standalone/deployment-apps-v1.json @@ -7561,12 +7561,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/deployment.json b/master-standalone/deployment.json index 22db6519f3..4c4238a180 100644 --- a/master-standalone/deployment.json +++ b/master-standalone/deployment.json @@ -7558,12 +7558,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/deploymentlist-apps-v1.json b/master-standalone/deploymentlist-apps-v1.json index 7a3ba1683c..21408be0f7 100644 --- a/master-standalone/deploymentlist-apps-v1.json +++ b/master-standalone/deploymentlist-apps-v1.json @@ -7576,12 +7576,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/deploymentlist.json b/master-standalone/deploymentlist.json index 3d7d6b1d50..ab0c69a782 100644 --- a/master-standalone/deploymentlist.json +++ b/master-standalone/deploymentlist.json @@ -7570,12 +7570,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/deploymentspec-apps-v1.json b/master-standalone/deploymentspec-apps-v1.json index aa8d450a87..115784477e 100644 --- a/master-standalone/deploymentspec-apps-v1.json +++ b/master-standalone/deploymentspec-apps-v1.json @@ -7290,12 +7290,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/deploymentspec.json b/master-standalone/deploymentspec.json index aa8d450a87..115784477e 100644 --- a/master-standalone/deploymentspec.json +++ b/master-standalone/deploymentspec.json @@ -7290,12 +7290,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/job-batch-v1.json b/master-standalone/job-batch-v1.json index 46935b7d58..d571df143d 100644 --- a/master-standalone/job-batch-v1.json +++ b/master-standalone/job-batch-v1.json @@ -7517,12 +7517,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/job.json b/master-standalone/job.json index 629689485e..06d9406bbf 100644 --- a/master-standalone/job.json +++ b/master-standalone/job.json @@ -7514,12 +7514,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/joblist-batch-v1.json b/master-standalone/joblist-batch-v1.json index 5a6590e091..53c95f1a05 100644 --- a/master-standalone/joblist-batch-v1.json +++ b/master-standalone/joblist-batch-v1.json @@ -7532,12 +7532,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/joblist.json b/master-standalone/joblist.json index 206a5483fd..21f862a849 100644 --- a/master-standalone/joblist.json +++ b/master-standalone/joblist.json @@ -7526,12 +7526,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/jobspec-batch-v1.json b/master-standalone/jobspec-batch-v1.json index e34c5c55fa..2c2ba19ff0 100644 --- a/master-standalone/jobspec-batch-v1.json +++ b/master-standalone/jobspec-batch-v1.json @@ -7243,12 +7243,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/jobspec.json b/master-standalone/jobspec.json index e34c5c55fa..2c2ba19ff0 100644 --- a/master-standalone/jobspec.json +++ b/master-standalone/jobspec.json @@ -7243,12 +7243,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/jobtemplatespec-batch-v1.json b/master-standalone/jobtemplatespec-batch-v1.json index 03a99df1b5..1341a8b6a3 100644 --- a/master-standalone/jobtemplatespec-batch-v1.json +++ b/master-standalone/jobtemplatespec-batch-v1.json @@ -7497,12 +7497,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/jobtemplatespec-batch-v1beta1.json b/master-standalone/jobtemplatespec-batch-v1beta1.json index 03a99df1b5..1341a8b6a3 100644 --- a/master-standalone/jobtemplatespec-batch-v1beta1.json +++ b/master-standalone/jobtemplatespec-batch-v1beta1.json @@ -7497,12 +7497,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/jobtemplatespec.json b/master-standalone/jobtemplatespec.json index 03a99df1b5..1341a8b6a3 100644 --- a/master-standalone/jobtemplatespec.json +++ b/master-standalone/jobtemplatespec.json @@ -7497,12 +7497,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/pod-v1.json b/master-standalone/pod-v1.json index 883b62fe0e..fcaf13588f 100644 --- a/master-standalone/pod-v1.json +++ b/master-standalone/pod-v1.json @@ -7141,12 +7141,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/pod.json b/master-standalone/pod.json index 9f54ec22d6..6262a22cc0 100644 --- a/master-standalone/pod.json +++ b/master-standalone/pod.json @@ -7138,12 +7138,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/podlist-v1.json b/master-standalone/podlist-v1.json index 5aa885f8fd..865d973feb 100644 --- a/master-standalone/podlist-v1.json +++ b/master-standalone/podlist-v1.json @@ -7156,12 +7156,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/podlist.json b/master-standalone/podlist.json index 48c35856b1..48af892ebb 100644 --- a/master-standalone/podlist.json +++ b/master-standalone/podlist.json @@ -7150,12 +7150,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/podspec-v1.json b/master-standalone/podspec-v1.json index ea886ad6d6..9c2fdc6ad5 100644 --- a/master-standalone/podspec-v1.json +++ b/master-standalone/podspec-v1.json @@ -6870,12 +6870,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/podspec.json b/master-standalone/podspec.json index ea886ad6d6..9c2fdc6ad5 100644 --- a/master-standalone/podspec.json +++ b/master-standalone/podspec.json @@ -6870,12 +6870,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/podtemplate-v1.json b/master-standalone/podtemplate-v1.json index 0ab7d5a6e4..394e73c2f8 100644 --- a/master-standalone/podtemplate-v1.json +++ b/master-standalone/podtemplate-v1.json @@ -7395,12 +7395,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/podtemplate.json b/master-standalone/podtemplate.json index 440b09a634..2b974e032e 100644 --- a/master-standalone/podtemplate.json +++ b/master-standalone/podtemplate.json @@ -7392,12 +7392,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/podtemplatelist-v1.json b/master-standalone/podtemplatelist-v1.json index 6536483a38..88a05ed4ac 100644 --- a/master-standalone/podtemplatelist-v1.json +++ b/master-standalone/podtemplatelist-v1.json @@ -7410,12 +7410,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/podtemplatelist.json b/master-standalone/podtemplatelist.json index 2e7c997af8..78aeaaeb0c 100644 --- a/master-standalone/podtemplatelist.json +++ b/master-standalone/podtemplatelist.json @@ -7404,12 +7404,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/podtemplatespec-v1.json b/master-standalone/podtemplatespec-v1.json index 193d014bed..7c598ac470 100644 --- a/master-standalone/podtemplatespec-v1.json +++ b/master-standalone/podtemplatespec-v1.json @@ -7121,12 +7121,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/podtemplatespec.json b/master-standalone/podtemplatespec.json index 193d014bed..7c598ac470 100644 --- a/master-standalone/podtemplatespec.json +++ b/master-standalone/podtemplatespec.json @@ -7121,12 +7121,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/replicaset-apps-v1.json b/master-standalone/replicaset-apps-v1.json index 3ae39ada6c..9c2ff8003a 100644 --- a/master-standalone/replicaset-apps-v1.json +++ b/master-standalone/replicaset-apps-v1.json @@ -7477,12 +7477,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/replicaset.json b/master-standalone/replicaset.json index 13d2b6bf47..dd022e07fc 100644 --- a/master-standalone/replicaset.json +++ b/master-standalone/replicaset.json @@ -7474,12 +7474,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/replicasetlist-apps-v1.json b/master-standalone/replicasetlist-apps-v1.json index 31b5ae4b39..7981f8a04c 100644 --- a/master-standalone/replicasetlist-apps-v1.json +++ b/master-standalone/replicasetlist-apps-v1.json @@ -7492,12 +7492,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/replicasetlist.json b/master-standalone/replicasetlist.json index 550787728b..10f2b3ae2e 100644 --- a/master-standalone/replicasetlist.json +++ b/master-standalone/replicasetlist.json @@ -7486,12 +7486,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/replicasetspec-apps-v1.json b/master-standalone/replicasetspec-apps-v1.json index cd93e5da3c..d47a6f97b3 100644 --- a/master-standalone/replicasetspec-apps-v1.json +++ b/master-standalone/replicasetspec-apps-v1.json @@ -7206,12 +7206,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/replicasetspec.json b/master-standalone/replicasetspec.json index cd93e5da3c..d47a6f97b3 100644 --- a/master-standalone/replicasetspec.json +++ b/master-standalone/replicasetspec.json @@ -7206,12 +7206,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/replicationcontroller-v1.json b/master-standalone/replicationcontroller-v1.json index f125c3ee9a..924ee07e0c 100644 --- a/master-standalone/replicationcontroller-v1.json +++ b/master-standalone/replicationcontroller-v1.json @@ -7428,12 +7428,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/replicationcontroller.json b/master-standalone/replicationcontroller.json index 63c1f62f93..8f0fd9cf36 100644 --- a/master-standalone/replicationcontroller.json +++ b/master-standalone/replicationcontroller.json @@ -7425,12 +7425,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/replicationcontrollerlist-v1.json b/master-standalone/replicationcontrollerlist-v1.json index 337ca0b037..673441c95d 100644 --- a/master-standalone/replicationcontrollerlist-v1.json +++ b/master-standalone/replicationcontrollerlist-v1.json @@ -7443,12 +7443,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/replicationcontrollerlist.json b/master-standalone/replicationcontrollerlist.json index 8ba1deffad..7e5c143140 100644 --- a/master-standalone/replicationcontrollerlist.json +++ b/master-standalone/replicationcontrollerlist.json @@ -7437,12 +7437,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/replicationcontrollerspec-v1.json b/master-standalone/replicationcontrollerspec-v1.json index f3d419a71e..c2b904b3b4 100644 --- a/master-standalone/replicationcontrollerspec-v1.json +++ b/master-standalone/replicationcontrollerspec-v1.json @@ -7154,12 +7154,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/replicationcontrollerspec.json b/master-standalone/replicationcontrollerspec.json index f3d419a71e..c2b904b3b4 100644 --- a/master-standalone/replicationcontrollerspec.json +++ b/master-standalone/replicationcontrollerspec.json @@ -7154,12 +7154,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/statefulset-apps-v1.json b/master-standalone/statefulset-apps-v1.json index 3da9ea013a..592757304d 100644 --- a/master-standalone/statefulset-apps-v1.json +++ b/master-standalone/statefulset-apps-v1.json @@ -7523,12 +7523,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/statefulset.json b/master-standalone/statefulset.json index d0d5b654d5..536badb640 100644 --- a/master-standalone/statefulset.json +++ b/master-standalone/statefulset.json @@ -7520,12 +7520,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/statefulsetlist-apps-v1.json b/master-standalone/statefulsetlist-apps-v1.json index 075bf91a20..27fcb6fd3c 100644 --- a/master-standalone/statefulsetlist-apps-v1.json +++ b/master-standalone/statefulsetlist-apps-v1.json @@ -7538,12 +7538,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/statefulsetlist.json b/master-standalone/statefulsetlist.json index a23106d25f..285840ef3f 100644 --- a/master-standalone/statefulsetlist.json +++ b/master-standalone/statefulsetlist.json @@ -7532,12 +7532,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/statefulsetspec-apps-v1.json b/master-standalone/statefulsetspec-apps-v1.json index dac81096f4..a65236487a 100644 --- a/master-standalone/statefulsetspec-apps-v1.json +++ b/master-standalone/statefulsetspec-apps-v1.json @@ -7255,12 +7255,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/statefulsetspec.json b/master-standalone/statefulsetspec.json index 762473fb2c..bc8069abc6 100644 --- a/master-standalone/statefulsetspec.json +++ b/master-standalone/statefulsetspec.json @@ -7255,12 +7255,20 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", "format": "int32", "type": "integer" }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master-standalone/topologyspreadconstraint-v1.json b/master-standalone/topologyspreadconstraint-v1.json index d7a9578234..040cb57bb7 100644 --- a/master-standalone/topologyspreadconstraint-v1.json +++ b/master-standalone/topologyspreadconstraint-v1.json @@ -68,7 +68,15 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", "format": "int32", "type": [ "integer", @@ -76,7 +84,7 @@ ] }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": [ "string", "null" diff --git a/master-standalone/topologyspreadconstraint.json b/master-standalone/topologyspreadconstraint.json index d7a9578234..040cb57bb7 100644 --- a/master-standalone/topologyspreadconstraint.json +++ b/master-standalone/topologyspreadconstraint.json @@ -68,7 +68,15 @@ "x-kubernetes-map-type": "atomic" }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", "format": "int32", "type": [ "integer", @@ -76,7 +84,7 @@ ] }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": [ "string", "null" diff --git a/master/_definitions.json b/master/_definitions.json index 88039ea242..a90419e785 100644 --- a/master/_definitions.json +++ b/master/_definitions.json @@ -10746,12 +10746,17 @@ "description": "LabelSelector is used to find matching pods. Pods that match this label selector are counted to determine the number of pods in their corresponding topology domain." }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "format": "int32", + "type": "integer" + }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", "format": "int32", "type": "integer" }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": "string" }, "whenUnsatisfiable": { diff --git a/master/topologyspreadconstraint-v1.json b/master/topologyspreadconstraint-v1.json index 28692468f3..48ef67ad32 100644 --- a/master/topologyspreadconstraint-v1.json +++ b/master/topologyspreadconstraint-v1.json @@ -6,7 +6,15 @@ "description": "LabelSelector is used to find matching pods. Pods that match this label selector are counted to determine the number of pods in their corresponding topology domain." }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", "format": "int32", "type": [ "integer", @@ -14,7 +22,7 @@ ] }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": [ "string", "null" diff --git a/master/topologyspreadconstraint.json b/master/topologyspreadconstraint.json index 28692468f3..48ef67ad32 100644 --- a/master/topologyspreadconstraint.json +++ b/master/topologyspreadconstraint.json @@ -6,7 +6,15 @@ "description": "LabelSelector is used to find matching pods. Pods that match this label selector are counted to determine the number of pods in their corresponding topology domain." }, "maxSkew": { - "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 1/1/0: | zone1 | zone2 | zone3 | | P | P | | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 1/1/1; scheduling it onto zone1(zone2) would make the ActualSkew(2-0) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "description": "MaxSkew describes the degree to which pods may be unevenly distributed. When `whenUnsatisfiable=DoNotSchedule`, it is the maximum permitted difference between the number of matching pods in the target topology and the global minimum. The global minimum is the minimum number of matching pods in an eligible domain or zero if the number of eligible domains is less than MinDomains. For example, in a 3-zone cluster, MaxSkew is set to 1, and pods with the same labelSelector spread as 2/2/1: In this case, the global minimum is 1. | zone1 | zone2 | zone3 | | P P | P P | P | - if MaxSkew is 1, incoming pod can only be scheduled to zone3 to become 2/2/2; scheduling it onto zone1(zone2) would make the ActualSkew(3-1) on zone1(zone2) violate MaxSkew(1). - if MaxSkew is 2, incoming pod can be scheduled onto any zone. When `whenUnsatisfiable=ScheduleAnyway`, it is used to give higher precedence to topologies that satisfy it. It's a required field. Default value is 1 and 0 is not allowed.", + "format": "int32", + "type": [ + "integer", + "null" + ] + }, + "minDomains": { + "description": "MinDomains indicates a minimum number of eligible domains. When the number of eligible domains with matching topology keys is less than minDomains, Pod Topology Spread treats \"global minimum\" as 0, and then the calculation of Skew is performed. And when the number of eligible domains with matching topology keys equals or greater than minDomains, this value has no effect on scheduling. As a result, when the number of eligible domains is less than minDomains, scheduler won't schedule more than maxSkew Pods to those domains. If value is nil, the constraint behaves as if MinDomains is equal to 1. Valid values are integers greater than 0. When value is not nil, WhenUnsatisfiable must be DoNotSchedule.\n\nFor example, in a 3-zone cluster, MaxSkew is set to 2, MinDomains is set to 5 and pods with the same labelSelector spread as 2/2/2: | zone1 | zone2 | zone3 | | P P | P P | P P | The number of domains is less than 5(MinDomains), so \"global minimum\" is treated as 0. In this situation, new pod with the same labelSelector cannot be scheduled, because computed skew will be 3(3 - 0) if new Pod is scheduled to any of the three zones, it will violate MaxSkew.\n\nThis is an alpha field and requires enabling MinDomainsInPodTopologySpread feature gate.", "format": "int32", "type": [ "integer", @@ -14,7 +22,7 @@ ] }, "topologyKey": { - "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. It's a required field.", + "description": "TopologyKey is the key of node labels. Nodes that have a label with this key and identical values are considered to be in the same topology. We consider each as a \"bucket\", and try to put balanced number of pods into each bucket. We define a domain as a particular instance of a topology. Also, we define an eligible domain as a domain whose nodes match the node selector. e.g. If TopologyKey is \"kubernetes.io/hostname\", each Node is a domain of that topology. And, if TopologyKey is \"topology.kubernetes.io/zone\", each zone is a domain of that topology. It's a required field.", "type": [ "string", "null"