tl;dr - I'm adding a fourth MR to this issue, to hide the Prometheus integration settings page entirely
I reached out to Environments group to coordinate removal efforts in %16.0
Here are their removals planned for 16.0
We're removing some of the cluster integration features in 16.0. They are related to the deprecated certificate based cluster integration (predecessor of GitLab Agent).
You can see some of the removed features in #388696 (closed) and will remove more in #366573 (closed)
Their answer regarding the setting page we are partially editing:
The Prometheus integration in Project Integration page were made for the certificate based cluster integration, which later was taken over by Prometheus App, which was removed in !105803 (merged). So it would make sense to remove it as well.
@psjakubowska I've enabled the flag on staging to start verifying the different MRs. So far I've noticed a couple of extra things that we don't have MRs for yet:
Navigating here directly still displays the metrics page - we should force it to render a 404 instead.
My proposal - either of us make MRs for these when we get a chance in the next day or so, and post the MR links here once created to make sure we don't double-up on work.
I assume there's a group (and possibly admin) equivalent.
It would be good to remove these too, but in my view they're not a blocker for flipping the flag - you can only get to this page from the integration settings, which are handled in !118218 (merged).
@tristan.read I wasn't able to guard these views, but as you mentions, with the merge of your MR for integrations settings they can't be reached anyways.
I'm signing of for today. All the MRs we worked on today are merged, but haven't reached the staging yet. So I wasn't able to test them. I will try in a couple of hours, but it might be too early anyways. So I probably leave that and flipping the flag on gitlab.com to you.
@syasonik I can see you've added some MRs that will use the flag as well. Is it a must have to merge them all before we proceed with enabling the flag?
As long as the UI is functioning well with the flag enabled on staging, I think we can enable the flag by default @psjakubowska! Just as long as we're not showing users partially broken states
I moved the backend code behind the same flag so that the frontend will have any backend APIs it needs regardless of flag state, but I don't think they strictly need to be connected.
Thanks @syasonik and @psjakubowska! I'll test the frontend changes on staging today, and will flip the flag on production when I've confirmed they're working as expected.
Current deployment status - most changes are deployed to workflowcanary but at least one (!118218 (merged)) is not deployed anywhere yet. Will continue to monitor throughout the day
For the reference, the Cluster Integration tab was removed as part of this issue (!120600 (merged))
Following the Monitor Metrics removal, I have noticed that by hiding the Cluster Health tab, the Integrations tab became redundant. It previously stored checks to enable Elastic Stack and Prometheus integrations. Elastic Stack was removed in early 2022. Now enabling Prometheus for Metrics will no longer be possible. Hence adding the Cluster Integrations tab to Metrics removal.
Link to Slack thread, where we discussed it with Environments group (internal only)
@psjakubowska@syasonik Now that !119989 (merged) is merged, the metrics removal flag is now default-enabled. So the metrics frontend is now removed for self-hosted instances as well as on .com.
We don't know if the change will make it into the %16.0 release yet, we'll have to monitor the release channels over the next day or two to see which commits get picked in to the release.