[Debian-ha-maintainers] corosync gets stopped during the qdevice autopkgtest

[Valentin Vidic]

On Sat, Feb 02, 2019 at 07:14:09PM +0100, Ferenc Wágner wrote:
> According to the discussion in
> https://github.com/corosync/corosync/pull/420 this shouldn't happen, and
> if I start with the corosync autopkgtest, it really does not.  At the
> moment I have no idea why corosync is stopped during the qdevice test...

systemd debug suggests that corosync unit is not needed anymore and starts
the stop action. Adding pacemaker to the depends list of the autopkgtest
seems to help as it will mark the corosync service as needed.

Feb  2 21:39:01 node1 systemd[1]: corosync.service: Trying to enqueue job corosync.service/start/replace
Feb  2 21:39:01 node1 systemd[1]: corosync.service: Installed new job corosync.service/start as 28331
Feb  2 21:39:01 node1 systemd[1]: corosync.service: Enqueued job corosync.service/start as 28331
Feb  2 21:39:01 node1 systemd[1]: corosync.service: ConditionKernelCommandLine=!nocluster succeeded.
Feb  2 21:39:01 node1 systemd[1]: corosync.service: Passing 0 fds to service
Feb  2 21:39:01 node1 systemd[1]: corosync.service: About to execute: /usr/sbin/corosync -f $COROSYNC_OPTIONS
Feb  2 21:39:01 node1 systemd[1]: corosync.service: Forked /usr/sbin/corosync as 23133
Feb  2 21:39:01 node1 systemd[1]: corosync.service: Changed dead -> start
Feb  2 21:39:01 node1 systemd[1]: Starting Corosync Cluster Engine...
Feb  2 21:39:01 node1 systemd[23133]: Successfully mounted /tmp/systemd-private-d2f8ac7954584c6384b6d0be34c3f78c-corosync.service-oN9bh3/tmp to /run/systemd/unit-root/tmp
Feb  2 21:39:01 node1 systemd[23133]: Successfully mounted /var/tmp/systemd-private-d2f8ac7954584c6384b6d0be34c3f78c-corosync.service-Eo32Fm/tmp to /run/systemd/unit-root/var/tmp
Feb  2 21:39:01 node1 systemd[23133]: corosync.service: Executing: /usr/sbin/corosync -f
Feb  2 21:39:01 node1 corosync[23133]:   [MAIN  ] Corosync built-in features: dbus monitoring watchdog augeas systemd xmlconf snmp pie relro bindnow
Feb  2 21:39:01 node1 systemd[1]: corosync.service: Got notification message from PID 23133 (READY=1)
Feb  2 21:39:01 node1 systemd[1]: corosync.service: Changed start -> running
Feb  2 21:39:01 node1 systemd[1]: corosync.service: Job 28331 corosync.service/start finished, result=done
Feb  2 21:39:01 node1 systemd[1]: Started Corosync Cluster Engine.

Feb  2 21:39:01 node1 systemd[1]: corosync.service: Unit is not needed anymore.
Feb  2 21:39:01 node1 systemd[1]: corosync.service: Trying to enqueue job corosync.service/stop/fail
Feb  2 21:39:01 node1 systemd[1]: corosync.service: Installed new job corosync.service/stop as 28404
Feb  2 21:39:01 node1 systemd[1]: corosync.service: Enqueued job corosync.service/stop as 28404
Feb  2 21:39:01 node1 systemd[1]: corosync.service: Failed to send unit change signal for corosync.service: Connection reset by peer
Feb  2 21:39:01 node1 systemd[1]: corosync.service: Changed running -> stop-sigterm
Feb  2 21:39:01 node1 systemd[1]: Stopping Corosync Cluster Engine...
Feb  2 21:39:02 node1 systemd[1]: Preset files don't specify rule for corosync.service. Enabling.
Feb  2 21:39:02 node1 systemd[1]: corosync.service: Reinstalled deserialized job corosync.service/stop as 28404
Feb  2 21:39:02 node1 systemd[1]: corosync.service: Changed dead -> stop-sigterm
Feb  2 21:39:02 node1 systemd[1]: Received SIGCHLD from PID 23133 (corosync).
Feb  2 21:39:02 node1 systemd[1]: Child 23133 (corosync) died (code=exited, status=0/SUCCESS)
Feb  2 21:39:02 node1 systemd[1]: corosync.service: Child 23133 belongs to corosync.service.
Feb  2 21:39:02 node1 systemd[1]: corosync.service: Main process exited, code=exited, status=0/SUCCESS
Feb  2 21:39:02 node1 systemd[1]: corosync.service: Succeeded.
Feb  2 21:39:02 node1 systemd[1]: corosync.service: Changed stop-sigterm -> dead
Feb  2 21:39:02 node1 systemd[1]: corosync.service: Job 28404 corosync.service/stop finished, result=done
Feb  2 21:39:02 node1 systemd[1]: Stopped Corosync Cluster Engine.
Feb  2 21:39:02 node1 systemd[1]: Spawning thread to nuke /tmp/systemd-private-d2f8ac7954584c6384b6d0be34c3f78c-corosync.service-oN9bh3
Feb  2 21:39:02 node1 systemd[1]: Spawning thread to nuke /var/tmp/systemd-private-d2f8ac7954584c6384b6d0be34c3f78c-corosync.service-Eo32Fm

-- 
Valentin