Ceph monitor process kill failed to result in host degrade or ceph health warn
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
StarlingX |
Fix Released
|
Medium
|
Wei Zhou |
Bug Description
Brief Description
-----------------
Ceph monitor process kill failed to result in degrade condition
Severity
--------
Major
Steps to Reproduce
------------------
1. Remove the ceph monitor and kill the ceph monitor process. This leaves two ceph monitors in the quorum.
2. No degrade condition is reported and ceph health remains okay with 2 members up in the quorum.
[2018-12-09 13:57:27,619] 389 DEBUG MainThread ssh.expect :: Output:
cluster f6f37d71-
health HEALTH_OK
monmap e2: 2 mons at {controller-
osdmap e181: 4 osds: 3 up, 3 in
flags sortbitwise,
pgmap v8390: 1728 pgs, 6 pools, 1786 MB data, 238 objects
3707 MB used, 1762 GB / 1766 GB avail
controller-1:~$
Need confirmation if this is the new behaviour resulting from going upstream for ceph. If so, test case needs to be modified accordingly. Or if this is a software issue.
Expected Behavior
------------------
Monitor where process was killed goes into degrade and ceph reports a health warn.
Actual Behavior
----------------
All nodes available and ceph reported okay
Reproducibility
---------------
Seems reproducible.
System Configuration
-------
Storage
Branch/Pull Time/Commit
-------
master as of 2018-12-07_20-18-00
Timestamp/Logs
--------------
2018-12-09 13:57:27,619
tags: |
added: stx.2019.05 removed: stx.2019.03 |
tags: |
added: stx.2.0 removed: stx.2019.05 |
Assigning to Ovidiu to comment on whether this is a bug or expected behavior given the recent ceph changes for AIO