|
On further thought, Im now thinking this is telling me which rank is stopped (2), not that two ranks are stopped. I guess I am still curious about why this information is retained here and can rank 2 be made active again? If so, would this be cleaned up out
of "stopped"?
The state diagram here: http://docs.ceph.com/docs/master/cephfs/mds-states/
seems to indicate that once a rank is "Stopped" it has no path to move out of that state. Perhaps I am reading it wrong.
We have updated multi-active-mds clusters and pushed down max_mds to 1 then back to 2 during those upgrades and on all of those clusters we do not have any listed in "stopped." so I am guessing those ranks go back to active.
Thanks for the clarity.
From: ceph-users <ceph-users-bounces@xxxxxxxxxxxxxx> on behalf of Wesley Dillingham <wdillingham@xxxxxxxxxxx>
Sent: Tuesday, May 28, 2019 5:15 PM To: ceph-users@xxxxxxxxxxxxxx Subject: Meaning of Ceph MDS / Rank in "Stopped" state.
Notice:
This email is from an external sender.
I am working to develop some monitoring for our File clusters and as part of the check I inspect `ceph mds stat` for damaged,failed,stopped MDS/Ranks. Initially I set my check to Alarm if any of these states was discovered but as I distributed it out I noticed
that one of our clusters had the following:
"failed": [],
"damaged": [],
"stopped": [
2
],
However the cluster health is good and the mds state is: cephfs-2/2/2 up {0=p3plcephmds001=up:active,1=p3plcephmds002=up:active}, 1 up:standby
A little further digging and I found that a stopped state doesnt apply to an MDS but rather a rank and may indicate that max_mds was previously set higher than its current setting of 2, and the "Stopped" ranks are simply ranks which were active
and simply offloaded their state to other ranks.
My question is, how can I inspect further which ranks are "stopped" and would it be appropriate to "clear" those stopped ranks if possible or should I modify my check to ignore stopped ranks and only focus on damaged/failed ranks.
The cluster is running 12.2.12
Thanks.
Respectfully,
Wes Dillingham
wdillingham@xxxxxxxxxxx
Site Reliability Engineer IV - Platform Storage
|
_______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
