Description of problem: With cluster config 1 Mon+1 Mgr,4 MDSs,4 OSDs and 4 Clients,3 MDSs are getting active after fresh cluster setup with 1 as standby, which used to be 1 active mds and 3 standbys Version-Release number of selected component (if applicable): ceph : ceph-12.2.5-12.el7cp os: Red Hat Enterprise Linux Server release 7.5 (Maipo) How reproducible: Always Steps to Reproduce: 1.Set up a ceph 3.1 cluster with configs above mentioned. 2.Check ceph health ,should see 1 active mds with 3 standby Actual results: id: 897b8555-f20a-48d5-9d8a-4cb03ea26ee3 health: HEALTH_WARN too few PGs per OSD (20 < min 30) services: mon: 1 daemons, quorum ceph-sshreeka-run125-node1-monmgrinstaller mgr: ceph-sshreeka-run125-node1-monmgrinstaller(active) mds: cephfs-3/3/3 up {0=ceph-sshreeka-run125-node6-mds=up:active,1=ceph-sshreeka-run125-node4-mds=up:active,2=ceph-sshreeka-run125-node3-mds=up:active}, 1 up:standby osd: 12 osds: 12 up, 12 in data: pools: 3 pools, 80 pgs objects: 57 objects, 4870 bytes usage: 1298 MB used, 346 GB / 347 GB avail pgs: 80 active+clean Expected results: Should have 1 active-mds Additional info: Log of ansible: http://magna002.ceph.redhat.com/cephci-jenkins/cephci-run-1527140513806/ceph_ansible_0.log
Shreekara would you please provide the ceph.conf in the nodes hosting the MDS services ? Looking at the docs it seems that this is multimds config http://docs.ceph.com/docs/master/cephfs/multimds/ Either way I think that Patrick would be best suited to advise and configure the default behavior. It appears as though this is expected for luminous(RHCS3) and was not for jewel(RHCS2) see https://github.com/ceph/ceph-ansible/commit/683bec9eb231cfcd97e93eaee25a4f5b0e9f76ab#diff-f840ae8f0b27425aa50c964de6b8a46fR23
Hi Gregory, Here is the links of ceph.conf file's contents pasted : http://pastebin.test.redhat.com/597636 http://pastebin.test.redhat.com/597639 http://pastebin.test.redhat.com/597640 http://pastebin.test.redhat.com/597641
https://github.com/ceph/ceph-ansible/pull/2719 was tagged back in ceph-ansible v3.1.0rc7. I'm setting Fixed In Version to the latest available ceph-ansible NVR.
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2018:2819