v1.19.0 changelog is missing some info

[dmaes]

Describe the bug

We recently updated from v1.18.3 to v1.19.0. We read the changelog, changed the registry we're pulling from and figured we should be fine, bumped the tag and figured we should be fine.

But we weren't:

• The operator now requires additional RBAC (events stuff)
• New probe endpoints were introduced
• The metrics endpoint, which we were using for probes, switched from HTTP to HTTPS

These seem to be part of #1105

• Some additonal info in the release would have been nice
• Even if someone did go through the PR's mentioned in the release, the above changes aren't mentioned in the PR description.

Reproduction steps

Update operator, notice the undocumented changes

Expected behavior

For the changes to be correctly documented

Additional context

No response

[Zerpet]

The operator now requires additional RBAC (events stuff)

The new RBAC is a bug we fixed in #1145, we completely missed this bit. You were not expected to update any RBAC manually.

New probe endpoints were introduced

What problems did this cause? The Deployment manifest includes everything it needs to utilise those probes.

The metrics endpoint, which we were using for probes, switched from HTTP to HTTPS

This was the old probe. The new manifest has dedicated probes, already configured in the manifest in the release. We have to document the new authorisation process to query the metrics, I simply haven't found the time to PR the docs. If you are not scraping the metrics to e.g. Prometheus, this is a no-op for you.

We read the changelog, changed the registry we're pulling from and figured we should be fine, bumped the tag and figured we should be fine.

We don't expect the "upgrade" to be performed as you did. Just updating the image and tag. The YAML manifest attached in the release contains all the updated objects (except the events bit).

The release notes are not a substitute for the docs, nor for testing upgrades in a staging environment. There's also a big warning in the release notes, which I hoped would hint everyone to tread this new version carefully.

[dmaes]

We're deploying the operator via the bitnami chart, not via the 1 big manifest. Which is another can of worms that has had his own share of discussion already, not going to continue on that.

I'm sorry if this came of as you broke our production because the release notes were insufficient that was not my intent.
It was more of a this upgrade turned out to be more of a hassle than we expected it to be, based on the release notes. It could have some more details in it to save others some time.
We do have proper procedures and a dev setup and all that in place to test upgrades before pushing them to prod.

As for the big warning, that's only about the containers being served from a different registry now, not about any possible functionality changes. Something like There has been some major refactorings (migrating to Kubebuilder V4), expect changes on how the operator has to be deployed (probes, metrics endpoints, RBAC, etc...) would already be a big help.

I get that you missed some things and fixed them afterwards (like the RBAC), but I personally would have edited the existing release notes to tell users to either manually apply them, or wait for the next release.

Anyways, love your work on the operator, just wanted to suggest some improvements on the release notes, that's all.