Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Switches to Allocatable from Capacity and makes status quantities more readable. #154

Merged
merged 1 commit into from
Nov 18, 2020
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
32 changes: 16 additions & 16 deletions pkg/controllers/metricsproducer/v1alpha1/suite_test.go
Original file line number Diff line number Diff line change
Expand Up @@ -66,28 +66,28 @@ var _ = Describe("Examples", func() {
Expect(ns.ParseResources("docs/examples/reserved-capacity-utilization.yaml", mp)).To(Succeed())
mp.Spec.ReservedCapacity.NodeSelector = map[string]string{"k8s.io/nodegroup": ns.Name}

capacity := v1.ResourceList{
v1.ResourceCPU: resource.MustParse("16"),
v1.ResourceMemory: resource.MustParse("128Gi"),
allocatable := v1.ResourceList{
v1.ResourceCPU: resource.MustParse("16300m"),
v1.ResourceMemory: resource.MustParse("128500Mi"),
v1.ResourcePods: resource.MustParse("50"),
}

nodes := []client.Object{
test.NodeWith(test.NodeOptions{Labels: mp.Spec.ReservedCapacity.NodeSelector, Capacity: capacity}),
test.NodeWith(test.NodeOptions{Labels: mp.Spec.ReservedCapacity.NodeSelector, Capacity: capacity}),
test.NodeWith(test.NodeOptions{Labels: map[string]string{"unknown": "label"}, Capacity: capacity}),
test.NodeWith(test.NodeOptions{Labels: mp.Spec.ReservedCapacity.NodeSelector, Capacity: capacity}),
test.NodeWith(test.NodeOptions{Labels: mp.Spec.ReservedCapacity.NodeSelector, Capacity: capacity, ReadyStatus: v1.ConditionFalse}),
test.NodeWith(test.NodeOptions{Labels: mp.Spec.ReservedCapacity.NodeSelector, Capacity: capacity, Unschedulable: true}),
test.NodeWith(test.NodeOptions{Labels: mp.Spec.ReservedCapacity.NodeSelector, Allocatable: allocatable}),
test.NodeWith(test.NodeOptions{Labels: mp.Spec.ReservedCapacity.NodeSelector, Allocatable: allocatable}),
test.NodeWith(test.NodeOptions{Labels: map[string]string{"unknown": "label"}, Allocatable: allocatable}),
test.NodeWith(test.NodeOptions{Labels: mp.Spec.ReservedCapacity.NodeSelector, Allocatable: allocatable}),
test.NodeWith(test.NodeOptions{Labels: mp.Spec.ReservedCapacity.NodeSelector, Allocatable: allocatable, ReadyStatus: v1.ConditionFalse}),
test.NodeWith(test.NodeOptions{Labels: mp.Spec.ReservedCapacity.NodeSelector, Allocatable: allocatable, Unschedulable: true}),
}

pods := []client.Object{
// node[0] 6/16 cores, 76/128 gig allocated
test.Pod(nodes[0].GetName(), ns.Name, v1.ResourceList{v1.ResourceCPU: resource.MustParse("1"), v1.ResourceMemory: resource.MustParse("1Gi")}),
test.Pod(nodes[0].GetName(), ns.Name, v1.ResourceList{v1.ResourceCPU: resource.MustParse("2"), v1.ResourceMemory: resource.MustParse("25Gi")}),
test.Pod(nodes[0].GetName(), ns.Name, v1.ResourceList{v1.ResourceCPU: resource.MustParse("3"), v1.ResourceMemory: resource.MustParse("50Gi")}),
test.Pod(nodes[0].GetName(), ns.Name, v1.ResourceList{v1.ResourceCPU: resource.MustParse("1100m"), v1.ResourceMemory: resource.MustParse("1Gi")}),
test.Pod(nodes[0].GetName(), ns.Name, v1.ResourceList{v1.ResourceCPU: resource.MustParse("2100m"), v1.ResourceMemory: resource.MustParse("25Gi")}),
test.Pod(nodes[0].GetName(), ns.Name, v1.ResourceList{v1.ResourceCPU: resource.MustParse("3300m"), v1.ResourceMemory: resource.MustParse("50Gi")}),
// node[1] 1/16 cores, 76/128 gig allocated
test.Pod(nodes[1].GetName(), ns.Name, v1.ResourceList{v1.ResourceCPU: resource.MustParse("1"), v1.ResourceMemory: resource.MustParse("1Gi")}),
test.Pod(nodes[1].GetName(), ns.Name, v1.ResourceList{v1.ResourceCPU: resource.MustParse("1100m"), v1.ResourceMemory: resource.MustParse("1Gi")}),
// node[2] is ignored
test.Pod(nodes[2].GetName(), ns.Name, v1.ResourceList{v1.ResourceCPU: resource.MustParse("99"), v1.ResourceMemory: resource.MustParse("99Gi")}),
// node[3] is unallocated
Expand All @@ -100,9 +100,9 @@ var _ = Describe("Examples", func() {
ExpectCreated(ns.Client, mp)

ExpectEventuallyHappy(ns.Client, mp)
Expect(mp.Status.ReservedCapacity[v1.ResourceCPU]).To(BeEquivalentTo("14%, 7/48"))
Expect(mp.Status.ReservedCapacity[v1.ResourceMemory]).To(BeEquivalentTo("20%, 82678120448/412316860416"))
Expect(mp.Status.ReservedCapacity[v1.ResourcePods]).To(BeEquivalentTo("2%, 4/150"))
Expect(mp.Status.ReservedCapacity[v1.ResourceCPU]).To(BeEquivalentTo("15.54%, 7600m/48900m"))
Expect(mp.Status.ReservedCapacity[v1.ResourceMemory]).To(BeEquivalentTo("20.45%, 77Gi/385500Mi"))
Expect(mp.Status.ReservedCapacity[v1.ResourcePods]).To(BeEquivalentTo("2.67%, 4/150"))

ExpectDeleted(ns.Client, mp)
ExpectDeleted(ns.Client, nodes...)
Expand Down
20 changes: 14 additions & 6 deletions pkg/metrics/producers/reservedcapacity/gauges.go
Original file line number Diff line number Diff line change
Expand Up @@ -16,21 +16,29 @@ package reservedcapacity

import (
"fmt"

"github.com/awslabs/karpenter/pkg/metrics"
"github.com/prometheus/client_golang/prometheus"
v1 "k8s.io/api/core/v1"
)

type MetricType string

const (
Subsystem = "reserved_capacity"
Reserved = "reserved"
Capacity = "capacity"
Utilization = "utilization"
Reserved = MetricType("reserved")
Capacity = MetricType("capacity")
Utilization = MetricType("utilization")
)

func init() {
for _, resource := range []v1.ResourceName{v1.ResourcePods, v1.ResourceCPU, v1.ResourceMemory} {
for _, name := range []string{Reserved, Capacity, Utilization} {
metrics.RegisterNewGauge(Subsystem, fmt.Sprintf("%s_%s", resource, name))
for _, resourceName := range []v1.ResourceName{v1.ResourcePods, v1.ResourceCPU, v1.ResourceMemory} {
for _, metricType := range []MetricType{Reserved, Capacity, Utilization} {
metrics.RegisterNewGauge(Subsystem, fmt.Sprintf("%s_%s", resourceName, metricType))
}
}
}

func GaugeFor(resourceName v1.ResourceName, metricType MetricType) *prometheus.GaugeVec {
return metrics.Gauges[Subsystem][fmt.Sprintf("%s_%s", resourceName, metricType)]
}
25 changes: 15 additions & 10 deletions pkg/metrics/producers/reservedcapacity/producer.go
Original file line number Diff line number Diff line change
Expand Up @@ -18,8 +18,7 @@ import (
"context"
"fmt"
"math"

"github.com/awslabs/karpenter/pkg/metrics"
"strconv"

"github.com/awslabs/karpenter/pkg/apis/autoscaling/v1alpha1"
"github.com/awslabs/karpenter/pkg/utils/node"
Expand Down Expand Up @@ -66,16 +65,22 @@ func (p *Producer) record(reservations *Reservations) {
p.Status.ReservedCapacity = map[v1.ResourceName]string{}
}
for resource, reservation := range reservations.Resources {
computed := reservation.Compute()
for metricType, value := range computed {
metrics.Gauges[Subsystem][fmt.Sprintf("%s_%s", resource, metricType)].
WithLabelValues(p.Name, p.Namespace).Set(value)
reserved, _ := strconv.ParseFloat(reservation.Reserved.AsDec().String(), 64)
capacity, _ := strconv.ParseFloat(reservation.Capacity.AsDec().String(), 64)
utilization := math.NaN()
if capacity != 0 {
utilization = reserved / capacity
}

GaugeFor(resource, Utilization).WithLabelValues(p.Name, p.Namespace).Set(utilization)
GaugeFor(resource, Reserved).WithLabelValues(p.Name, p.Namespace).Set(reserved)
GaugeFor(resource, Capacity).WithLabelValues(p.Name, p.Namespace).Set(capacity)

p.Status.ReservedCapacity[resource] = fmt.Sprintf(
"%v%%, %d/%d",
math.Floor(computed[Utilization]*100),
int64(computed[Reserved]),
int64(computed[Capacity]),
"%.2f%%, %v/%v",
reserved/capacity*100,
reservation.Reserved,
reservation.Capacity,
)
}
}
22 changes: 3 additions & 19 deletions pkg/metrics/producers/reservedcapacity/reservations.go
Original file line number Diff line number Diff line change
Expand Up @@ -15,9 +15,6 @@ limitations under the License.
package reservedcapacity

import (
"math"
"math/big"

v1 "k8s.io/api/core/v1"
"k8s.io/apimachinery/pkg/api/resource"
)
Expand Down Expand Up @@ -53,25 +50,12 @@ func (r *Reservations) Add(node *v1.Node, pods *v1.PodList) {
r.Resources[v1.ResourceMemory].Reserved.Add(*container.Resources.Requests.Memory())
}
}
r.Resources[v1.ResourcePods].Capacity.Add(*node.Status.Capacity.Pods())
r.Resources[v1.ResourceCPU].Capacity.Add(*node.Status.Capacity.Cpu())
r.Resources[v1.ResourceMemory].Capacity.Add(*node.Status.Capacity.Memory())
r.Resources[v1.ResourcePods].Capacity.Add(*node.Status.Allocatable.Pods())
r.Resources[v1.ResourceCPU].Capacity.Add(*node.Status.Allocatable.Cpu())
r.Resources[v1.ResourceMemory].Capacity.Add(*node.Status.Allocatable.Memory())
}

type Reservation struct {
Reserved *resource.Quantity
Capacity *resource.Quantity
}

func (r *Reservation) Compute() map[string]float64 {
var utilization = math.NaN()
if r.Capacity.Value() != 0 {
utilization, _ = big.NewRat(r.Reserved.Value(), r.Capacity.Value()).Float64()
}

return map[string]float64{
Reserved: float64(r.Reserved.Value()),
Capacity: float64(r.Capacity.Value()),
Utilization: utilization,
}
}
6 changes: 3 additions & 3 deletions pkg/test/objects.go
Original file line number Diff line number Diff line change
Expand Up @@ -46,7 +46,7 @@ type NodeOptions struct {
Labels map[string]string
ReadyStatus v1.ConditionStatus
Unschedulable bool
Capacity v1.ResourceList
Allocatable v1.ResourceList
}

func NodeWith(options NodeOptions) *v1.Node {
Expand All @@ -69,8 +69,8 @@ func NodeWith(options NodeOptions) *v1.Node {
Unschedulable: options.Unschedulable,
},
Status: v1.NodeStatus{
Capacity: options.Capacity,
Conditions: []v1.NodeCondition{{Type: v1.NodeReady, Status: options.ReadyStatus}},
Allocatable: options.Allocatable,
Conditions: []v1.NodeCondition{{Type: v1.NodeReady, Status: options.ReadyStatus}},
},
}
}