Always include volumes in statefulsets

[dzsibi]

What problem are we solving?

When both master.data.type and master.logs.type is persistentVolumeClaim, deployment of the Helm chart fails:

create Pod seaweedfs-master-0 in StatefulSet seaweedfs-master failed error: Pod "seaweedfs-master-0" is invalid: spec.containers[0].volumeMounts[1].name: Not found: "master-config"

This is due to a condition that omits the entire volumes section in this case:

https://github.com/seaweedfs/seaweedfs/blob/d5d9fbb8aa5e86f2b866e25f58bf36a7a989478d/k8s/charts/seaweedfs/templates/master-statefulset.yaml#L217C20-L217C20

The issue was introduced by #4797, as it did not take into account this condition.

How are we solving the problem?

I removed the conditions entirely, including the ones for volume-statefulset. They also interfere with a bunch of other options (extraVolumes, enableSecurity).

How is the PR tested?

For the default configuration (using hostPath volumes), deployment works before and after my changes. For a configuration that uses PVCs instead, deployments fail before and work after.

An example values.yaml to trigger the bug:

seaweedfs:
  master:
    data:
      type: persistentVolumeClaim
    logs:
      type: persistentVolumeClaim

Checks

I am unsure if these are relevant for this PR.

• I have added unit tests if possible.
• I will add related wiki document changes and link to this PR after merging.

[chrislusf]

Thanks! Merging.

cc @aviau let me know if you have any concern.

[dzsibi]

Hmm, I didn't know it's going to auto-publish immediately, I think it may have overwritten the previous version since I did not include a version bump in the PR.

[jessebot]

@dzsibi yes, can you please submit a PR to bump the helm chart version?

@chrislusf merge this after the helm chart is bumped to prevent this in the future: #5106 If you don't want till after, you'll have to force merge, because the lint might complain that nothing has changed for the version bump.

[jessebot]

nevermind, #5108 should take care of this :)