cluster: fix metrics-server deployment on CI jobs#103713
cluster: fix metrics-server deployment on CI jobs#103713k8s-ci-robot merged 2 commits intokubernetes:masterfrom
Conversation
|
/assign @liggitt |
|
/test |
|
@aojea: The
Use
DetailsIn 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/test-infra repository. |
|
/hold |
|
/test pull-kubernetes-e2e-gce-iscsi |
|
/triage accepted |
|
/retest |
38092b9 to
fb2f0d2
Compare
|
/test pull-kubernetes-e2e-gce-ubuntu |
use deployment based on upstream manifest - add liveness and readiness probes - use kubelet port secure port https://github.com/kubernetes-sigs/metrics-server/blob/release-0.4/manifests/base/deployment.yaml#L32-L38
3c3b0a1 to
0610968
Compare
yangjunmyfm192085
left a comment
There was a problem hiding this comment.
/test pull-kubernetes-integration
|
/test pull-kubernetes-integration |
|
it works using the latest image , however, ti seems that the setcap fix was backported to the 0.4 branch and promoted correctly @serathius does it makes any sense to you? |
|
/approve @aojea if you need this for 1.22, please add a milestone and give a description/justification for @kubernetes/release-team as a comment. |
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: aojea, dims The full list of commands accepted by this bot can be found here. The pull request process is described here DetailsNeeds approval from an approver in each of these files:Approvers can indicate their approval by writing |
|
@aojea To my knowledge, there are no differences between v0.4.4 and v0.5.0 that would change the behavior that we were seeing before. So there should be no risks with moving forward with v0.5.0 with the exception of maybe missing a real bug with dockershim. In case I double-checked the capabilities of the official images and they are the same: $ docker cp $(docker create --rm k8s.gcr.io/metrics-server/metrics-server:v0.4.4):metrics-server /tmp/metrics-server && getcap /tmp/metrics-server
/tmp/metrics-server cap_net_bind_service=ep
$ docker cp $(docker create --rm k8s.gcr.io/metrics-server/metrics-server:v0.5.0):metrics-server /tmp/metrics-server && getcap /tmp/metrics-server
/tmp/metrics-server cap_net_bind_service=ep |
Yes, please backport to |
|
/lgtm |
|
looks like there is consensus to apply this in 1.22 and fail a cherry-pick. /milestone v1.22 @aojea please file a cherry-pick to 1.22 branch |
…3-upstream-release-1.22 Automated cherry pick of #103713: cluster: fix CI metrics-server deployment
/kind failing-test
/kind flake
Fixes #103708
The metrics-server deployed wasn't working correctly due to the recent changes in kubernetes.
Surprisingly, the metrics-server was constantly crash looping but the tests were passing.
This PR fixes the metrics-server deployment: