Description of problem: Monitor marked as down after tcmalloc: large alloc" despite service and container are running retrying handle_auth_request Version-Release number of selected component (if applicable): 14.2.8-35 Upgraded from luminous to nautilus 3.3.z4 to 4.1 latest build How reproducible: (Tried only once) Steps followed: 1. Upgrade cluster from luminous to nautilus 2. Convert all filestore OSDs to bluestore OSDs; one host at a time Actual results: Monitor service is down Expected results: Monitor must be up and running Additional info:
The steps are somewhat too simplistic. - does it happen all the time? - is there a workaround? - full logs would be great, not just the journal?
Is there a crash here? If the monitor was just marked down due to some slowness but kept running, that does not meet the blocker criteria: https://mojo.redhat.com/docs/DOC-1146159
Hi Josh, (In reply to Josh Durgin from comment #3) > Is there a crash here? If the monitor was just marked down due to some > slowness but kept running, that does not meet the blocker criteria: > https://mojo.redhat.com/docs/DOC-1146159 No, I coulsn't observe any crash, clearing blocker? (In reply to Yaniv Kaul from comment #2) > The steps are somewhat too simplistic. > - does it happen all the time? > - is there a workaround? > - full logs would be great, not just the journal? Hi Yaniv, Will try to reproduce in couple of days. We couldn't procure logs as logs aren't stored in by default in containerized scenario. Ref - https://bugzilla.redhat.com/show_bug.cgi?id=1794409 . Will get full logs if reproduced. Regards, Vasishta Shastry QE, Ceph
(In reply to Vasishta from comment #4) > Hi Josh, > > (In reply to Josh Durgin from comment #3) > > Is there a crash here? If the monitor was just marked down due to some > > slowness but kept running, that does not meet the blocker criteria: > > https://mojo.redhat.com/docs/DOC-1146159 > > No, I coulsn't observe any crash, clearing blocker? Thanks, moving out of 4.1 to the backlog to clarify that it isn't a blocker.