Rsyslog constantly restarts after upgrade to 1.25.11
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
juju-core |
Confirmed
|
Undecided
|
Unassigned |
Bug Description
Trusty, Mitaka, Juju 1.25.11
Juju is restarting rsyslog as often as every 1 minute. This seems to have started approximately when juju was upgraded from 1.25.10 to .11, on 2017-03-31.
When we first began investigating, units were falling over with the message https:/
All units returned from 'lost' or 'failed' to 'idle'.
syslog shows frequent queue errors, queue and queue index files are left on disks and never processed. Contents of queue files at times at least appear uncorrupted, afaict. Clearing the spool only temporarily alleviates the problem.
sosreports are uploading to https:/
Changed in juju-core: | |
status: | New → Incomplete |
Changed in juju-core: | |
status: | Incomplete → Confirmed |
I'm downloading the SOS report (taking a long time).
In the mean time can you take a look in /var/log/syslog on the controller(s) and check if rsyslog is falling over? We have had situations in 1.25 where rsyslog got into a bad state and kept crashing.