lock is held by gke**** cluster-autoscaler on GCP

[eyltl]

Which component are you using?:
cluster-autoscaler

Which autoscaling component hosted in this repository (cluster-autoscaler, vertical-pod-autoscaler, addon-resizer, helm charts) is the bug in?
cluster-autoscaler

What version of the component are you using?:
appVersion: 1.21.0

chart version 9.10.7

What k8s version are you using (kubectl version)?:
1.21

What environment is this in?:
gke

What did you expect to happen?:
the held should be by cluster-autoscaler , and not by gke cluster autoscaler.

What happened instead?:
lock is held by gke-******

How to reproduce it (as minimally and precisely as possible):
restart cluster autoscaler pods.

[x13n]

Right, when GKE autoscaling is disabled, it should be possible to run OSS Cluster Autoscaler instead. One workaround would be to use --namespace flag and run CA in any namespace other than kube-system, so the leader election resource lock is not conflicting with GKE CA.

[eyltl]

thank you, seems to solve the issue.
i also had to update the role and the rolebinding to the same namespace specified in the "--namespace" so the serviceaccount will have permission to get and update the configmap

[gjtempleton]

@eyltl Sounds like your issue is resolved, are you happy for us to close this issue?

[eyltl]

Yes
Thank you very much

[ujjwal]

thank you, seems to solve the issue. i also had to update the role and the rolebinding to the same namespace specified in the "--namespace" so the serviceaccount will have permission to get and update the configmap

I am running into the same issue, do you have steps on the updates to the role & role bindings?