Comment 2 for bug 1834475

We've changed some of those timings in 3.0.4 which will make it in Ubuntu in the next month or so, but those tests can still be slightly flaky even in our CI as we're testing cluster recovery during random node losses, sometimes things take a bit longer than the 30s timeout to recover, especially on busy systems like those adt test VMs.

It's not an actual problem that users will ever see, so if a retry gets you through, don't worry about that particular failure.