What should happen when using "--dry-run=server"

[o4rz3l]

Output of helm version: 3.13.1

Output of kubectl version: v1.25.2

Hi, was trying the new dry-run=server feature but i'm not quite sure, what it does. In my point of view the difference between client und server should be:

• --dry-run=client: try to render the template
• --dry-run=server: try to render the template and do kubectl apply -f template.yaml --dry-run=server
  If the apply succeeds, status should be "green", if not, it should notify me, that an helm upgrade will fail.

I was trying it with a PVC resource. The default value for the size is 1GB. I manually increased the size to > 1GB. Now i ran helm upgrade --dry-run=server and from my point of view, it should fail - but it did not.
Running kubectl apply -f .. --dry-run=server returns exit code 1 due to the PVC size, which is the expected output since the PVC could not be downgraded in size,

So, what is --dry-run=server doing? Any hints to clearify this is appreciated!

Best Regards,
Oliver

[z4ce]

--dry-run=server will enable lookup functions. This table pretty much sums it up.. we should probably add it to the docs somewhere:

#9426 (comment)

[gjenkins8]

We thought about his a little when implementing --dry-run=server. The first/initial goal was to get lookup to work. The next step might be to get actual server-side validation enabled. But that would need further work/design. At the moment, agree with @z4ce that docs should explain this.

[o4rz3l]

Ok, thanks for the clarification and agree, that should be in the docs ;)