You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
To allow users to conform with this recommendation I propose adding the ability for users to modify their calico config to make use of Typha via editing their cluster and adding a typhaReplicas option to their networking: calico: ... stanza.
Setting typhaReplicas to a non zero positive integer would then have a number of effects:
Create a service for calico-typha
Set the typha_service_name to calico-typha in the Calico configmap to configure the calico-node pods to connect to Typha rather than directly to the API Server.
Set the number of replicas for the calico-typha deployment to the same value as typhaReplicas
The intention of this would be to act as an intermediate step between the complexity allowed for by: #6872 where people can scale their clusters to a massive degree using BGP Route Reflectors and the current 0 Typha, API server datastore default setup. 2. Feel free to provide a design supporting your feature request.
The text was updated successfully, but these errors were encountered:
gjtempleton
changed the title
Ability to Configure Typha for Calico
Ability to Configure Typha for Calico CNI
Jun 18, 2019
1. Describe IN DETAIL the feature/behavior/change you would like to see.
The Calico deployment in k8s 1.12 currently sets the
typha_replicas
to 0 andtypha_service_name: "none"
in https://github.com/kubernetes/kops/blob/master/upup/models/cloudup/resources/addons/networking.projectcalico.org/k8s-1.12.yaml.templateThe recommendation from the Calico docs when using the k8s APIServer as the datastore as kops does is that once a cluster goes above 50 nodes then Typha is used to lessen the impact on the k8s API Servers from Felix: https://docs.projectcalico.org/v3.4/getting-started/kubernetes/installation/calico#installing-with-the-kubernetes-api-datastoremore-than-50-nodes
I've seen #7051 and this doesn't change this behaviour and the recommendation is still the same with Calico v3.7 : https://docs.projectcalico.org/v3.7/getting-started/kubernetes/installation/calico#installing-with-the-kubernetes-api-datastoremore-than-50-nodes
To allow users to conform with this recommendation I propose adding the ability for users to modify their calico config to make use of Typha via editing their cluster and adding a
typhaReplicas
option to theirnetworking: calico: ...
stanza.e.g.
Setting
typhaReplicas
to a non zero positive integer would then have a number of effects:calico-typha
typha_service_name
tocalico-typha
in the Calico configmap to configure the calico-node pods to connect to Typha rather than directly to the API Server.typhaReplicas
The intention of this would be to act as an intermediate step between the complexity allowed for by: #6872 where people can scale their clusters to a massive degree using BGP Route Reflectors and the current 0 Typha, API server datastore default setup.
2. Feel free to provide a design supporting your feature request.
The text was updated successfully, but these errors were encountered: