hardcoded deployment name in v11.0.0

[jekader]

What steps did you take and what happened:

• Install chart version 10.1.3 as a non-default name (i.e. test-velero)
• Note that resources (deployment, service, role and bindings) are called test-velero from the value of velero.fullname
• Upgrade to 11.0.0
• The deployment called test-velero gets deleted and replaced with one called velero for no apparent reason. Everything else is still called test-velero.

What did you expect to happen: Deployment gets updated but not re-created under a hardcoded inconsistent name.

Anything else you would like to add: This commit hardcoded the deployment name: https://github.com/vmware-tanzu/helm-charts/commit/4a179e6c698e8d29fe36469eb996136baa3b3a35

Everything else including documentation still uses velero.fullname so the reason behind this is unclear: https://github.com/search?q=repo%3Avmware-tanzu%2Fhelm-charts+fullname&type=code

This causes unneeded resource re-creation, makes naming inconsistent and also makes it impossible to deploy several velero instances in one namespace as that would cause name collisions for the deployment name.

[Kajot-dev]

Hi, I;m the guy that made the change. This is not a bug - the deployment name must be velero and the node-agent daemonset must be named node-agent - these names are hardcoded across velero codebase.

If the deployment name is not velero, pod volume restores and repo maintenances break. For more detailed explanation, see: https://github.com/vmware-tanzu/helm-charts/pull/709#discussion_r2366189320

Relevant issue: https://github.com/vmware-tanzu/helm-charts/issues/679

[ederevea]

Thanks for sharing the details. Basically, the chart is working around a bug in velero itself.

Here's the related velero issue: https://github.com/vmware-tanzu/velero/issues/9023

Once that gets fixed we could revert it but for now I guess there's not much the chart can do.