Ceph

+1

David Orman also suggested requiring a manual override to install a "pulled" version once that version has been flagged.

There's going to be a new mgr module called 'feedback' for users to report Ceph issues from the CLI and UI.

Given that's basically a wrapper around Redmine, we could use it to regularly (daily) poll Redmine and trigger a Healthcheck error if an issue with "severity = critital" affects the version the cluster is running. The main downside of this would be DDoS (it seems that Redmine has some built-in cache).

Sage suggested adding a field in https://github.com/ceph/ceph/blob/master/doc/releases/releases.yml to list critical bugs against each release. We could use this information to raise health warnings about critical bugs, end of life of releases etc. The critical bugs bit can also be reflected in our releases table in the docs.

I think this needs to have an "acknowledge" functionality so it doesn't persist to cause warnings.

Given the motivating case, we want to warn users not to enable the bluestone omap fsck, but once they have that information they should not continue getting warned.