cinder-csi-plugin does not have options for leader election

[bgagnon]

Of all the containers in the pod template of the csi-cinder-controllerplugin StatefulSet, docker.io/k8scloudprovider/cinder-csi-plugin is the only one that does not appear to have options for leader election.

Isn't this a requirement to run in HA with a multi-replica Deployment?

[jichenjc]

/assign
I am not expert, but I can take a look, if I will not able to fix it, I will unassign myself :)

[adisky]

/kind feature

[jichenjc]

ok, I mostly figured this out, will submit a PR soon..

[jichenjc]

@adisky it shows to me that we are using RPC server here, while some other containers like external-provisioner is using controller, my search shows the leader election only applied to controller, so do we have any comments on whether we still can use leader election here?

[bgagnon]

@jichenjc if leader election is not necessary for this particular container, then the StatefulSet with replicas=1 should be replaced with a Deployment with appropriate leader election flags on the containers that support it.

The comments in the docs suggest that the StatefulSet is being used precisely to avoid having more than one pod at all times... this led me to believe that leader election is not yet supported.

[jichenjc]

ok, I will take another look at this and paste some comments if I have . thanks for the info~ @bgagnon

[jichenjc]

just for record, I think using CSI-lib https://github.com/kubernetes-csi/csi-lib-utils will be a good choice to include leader election stuff as manila CSI did, then in turn encounter some issue and need fix kubernetes-csi/csi-lib-utils#32 first

[fejta-bot]

Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle stale

[bgagnon]

@jichenjc any updates on this issue?

[davidovich]

/remove-lifecycle stale

[jichenjc]

sorry, busy in other stuffs recently, I unassign myself and will continue to work on it later if someone is not going to work on it

[fejta-bot]

Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle stale

[ramineni]

/remove-lifecycle stale

[fejta-bot]

Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle stale

[ramineni]

/remove-lifecycle stale

[ramineni]

/assign

[fejta-bot]

Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle stale

[ramineni]

/remove-lifecycle stale

[fejta-bot]

Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.

If this issue is safe to close now please do so with /close.

Send feedback to sig-contributor-experience at kubernetes/community.
/lifecycle stale

[ramineni]

/remove-lifecycle stale

[fejta-bot]

Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.

If this issue is safe to close now please do so with /close.

Send feedback to sig-contributor-experience at kubernetes/community.
/lifecycle stale

[fejta-bot]

Stale issues rot after 30d of inactivity.
Mark the issue as fresh with /remove-lifecycle rotten.
Rotten issues close after an additional 30d of inactivity.

If this issue is safe to close now please do so with /close.

Send feedback to sig-contributor-experience at kubernetes/community.
/lifecycle rotten

[ramineni]

/remove-lifecycle rotten

[m-yosefpor]

I don't understand why do we need to add leader election flag to driver. Leader election (lease) in case of csi are only for controllers to avoid race condition when watching/updating resources on api server. However csi driver act only as a gRPC server and which serves requests by other csi controllers (the ones that already support leader election). Only the one controller which is leader calls its side-car driver.