Tectonic 1.8.9-tectonic.1 subPath unmount issue

[yuko11]

Issue Report Template

Tectonic Version

Tectonic 1.8.9-tectonic.1 and Tectonic 1.8.7-tectonic.2

Environment

Bare metal What hardware/cloud provider/hypervisor is being used with Tectonic? VMware

Expected Behavior

Pods with subPath successfully removed

Actual Behavior

Pods with subPath are not removed and stuck in Terminating state, logs said can't get container status

Reproduction Steps

1. In Tectonic 1.8.9-tectonic.1, use subPath for volumeMounts in Deployment.
2. Try to remove Pod and see it stuck in terminating state.
3. For version Tectonic 1.8.7 it was working fine, also I have manage to replicate the issue updating Tectonic from 1.8.7 to 1.8.9 on other cluster.
4. Try to remove subPath from volumeMounts and try to delete Pod, see that it fixes the issue and Pod removed successfully.
5. Pod force delete is working fine in both versions.

Other Information

Seems to be the issue with deleting Pod is due to unmount failure.

[HBO2]

@yuko11 i have exactly the same issue. From my point of view it also influences the creation time of a pod.

I am using a trick to get them out of terminating loop by setting the grace-period to ¨0¨

[rendhalver]

I am having the same problem. Any chance this will get fixed any time soon?

[yuko11]

I saw reply to the issue with similar description, saying with next release early April.

[rendhalver]

Which issue @yuko11 ?

[robszumski]

Same issue as https://github.com/coreos/tectonic-forum/issues/290

[rendhalver]

Define Early April.

[robszumski]

Hi folks, apologies for the delay but Tectonic 1.9.6 has been released to the pre-production channel and should fix this issue. Please report back if you do or don't see a change in behavior.

1.9.6 will be pushed to the production channel after baking in pre-prod.

[rendhalver]

Thanks @robszumski