Support #35694
closedCephFS stops working after upgrade from 12.2.7 to 12.2.8
0%
Description
Hi !
We have done an upgrade from 12.2.7 to 12.2.8 on Ubuntu 14.04.5 LTS (amd64).
In the order MON-OSD-MDS-MGR-RADOSGW
There are approx. 70 Servers connected via CephFS.
The Upgrade itself went very well, but after a while we recognized,
that CephFS is not working anymore.
There is one active and two standby MDS configured.
After a restart of the mds services, we could see, that CephFS
is working 10-20 seconds.
The mds process is still running, but in the cluster status we can see that there
is a switchover to the next standby.
This is reproduceable an all 3 mds servers.
/var/log/ceph/ceph-mds.ID.log contains nothing except the normal messages from the switchover.
To bring CephFS back to life again, we installed the package from 12.2.7 again like this :
wget https://download.ceph.com/debian-luminous/pool/main/c/ceph/ceph-mds_12.2.7-1trusty_amd64.deb
dpkg -i --ignore-depends=ceph-base ceph-mds_12.2.7-1trusty_amd64.deb
Is this a known issue ?
How can we further debug this ?
Updated by John Spray over 5 years ago
- Project changed from Ceph to CephFS
Suggest setting "debug mds = 10" and gathering logs from the period when an MDS daemon goes active to the point that it appears to be replaced.
Updated by Patrick Donnelly over 5 years ago
- Tracker changed from Bug to Support
- Status changed from New to Rejected
The appropriate forum for these questions/support is ceph-users. Please reopen an issue when you're certain you've found a bug and have necessary diagnostic information (like debug logs as John outlined).