🌱 Cache Pods for KCP #11453
🌱 Cache Pods for KCP #11453
Conversation
k8s-ci-robot
added
do-not-merge/work-in-progress
|
@fabriziopandini: The
The following commands are available to trigger optional jobs:
Use
In response to this:
Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes-sigs/prow repository. |
k8s-ci-robot
added
the
size/S
|
/test? |
|
@fabriziopandini: The following commands are available to trigger required jobs:
The following commands are available to trigger optional jobs:
Use
In response to this:
Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes-sigs/prow repository. |
|
/test pull-cluster-api-e2e-main |
fabriziopandini
force-pushed
the
improve-control-plane-rollout
branch
from
f784a78
to
4e9cb6f
Compare
fabriziopandini
changed the title
k8s-ci-robot
removed
the
do-not-merge/work-in-progress
controlplane/kubeadm/main.go
Outdated
| req1, _ := labels.NewRequirement("tier", selection.Equals, []string{"control-plane"}) | ||
| req2, _ := labels.NewRequirement("component", selection.In, []string{"kube-apiserver", "kube-controller-manager", "kube-scheduler", "etcd"}) | ||
| podSelector := labels.NewSelector().Add(*req1, *req2) |
There was a problem hiding this comment.
We did elsewhere:
| req1, _ := labels.NewRequirement("tier", selection.Equals, []string{"control-plane"}) | |
| req2, _ := labels.NewRequirement("component", selection.In, []string{"kube-apiserver", "kube-controller-manager", "kube-scheduler", "etcd"}) | |
| podSelector := labels.NewSelector().Add(*req1, *req2) | |
| must := func(r *labels.Requirement, err error) labels.Requirement { | |
| if err != nil { | |
| panic(err) | |
| } | |
| return *r | |
| } | |
| podSelector := labels.NewSelector().Add( | |
| must(labels.NewRequirement("tier", selection.Equals, []string{"control-plane"})), | |
| must(labels.NewRequirement("component", selection.In, []string{"kube-apiserver", "kube-controller-manager", "kube-scheduler", "etcd"})), | |
| ) |
So it also ensures on changes we keep it right :-)
There was a problem hiding this comment.
Fixed,
But we have still 4 places where we are using "req, _ := labels.NewRequirement", I will follow up
fabriziopandini
added
the
area/provider/control-plane-kubeadm
fabriziopandini
force-pushed
the
improve-control-plane-rollout
branch
from
4e9cb6f
to
6e059a2
Compare
k8s-ci-robot
added
the
lgtm
|
LGTM label has been added. Git tree hash: b68b8ad8e833d3f737e76112ec301ed4bc762580
|
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: chrischdi The full list of commands accepted by this bot can be found here. The pull request process is described here
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
k8s-ci-robot
added
the
approved
What this PR does / why we need it:
This PR enable caching for kubeadm static pods only in KCP.
The main reason while we identified this improvement was to improve KCP rollouts when temporary connection issue happens.
With the previous setup, pod where not cached, and thus every KCP reconcile a series of API calls where issues to gather status of the controlplane components in the workload cluster.
However, in case of temporary connection issues, kcp was "freezing" because stuck in a series of get pod--> wait 10s second timeout.
While looking at options to fix this issue, we figured out that by caching selectively the kubeadm Pods we can achieve a good trade off between:
/area provider/control-plane-kubeadm