Comment 43 for bug 1642966

Revision history for this message
Martin Pitt (pitti) wrote :

Till's log shows that cups.path triggers after cups got stopped:

Dec 02 10:59:15 till-x1carbon systemd[1]: cups.path: Got notified about unit deactivation.
Dec 02 10:59:15 till-x1carbon systemd[1]: cups.path: Changed running -> waiting
Dec 02 10:59:15 till-x1carbon systemd[1]: cups.socket: Changed running -> listening
[...]
Dec 02 10:59:19 till-x1carbon systemd[1]: cups.path: Got triggered.
Dec 02 10:59:19 till-x1carbon systemd[1]: cups.service: Trying to enqueue job cups.service/start/replace

and soon afterwards something stops it again:

Dec 02 10:59:29 till-x1carbon systemd[1]: cups.service: Trying to enqueue job cups.service/stop/replace
Dec 02 10:59:29 till-x1carbon systemd[1]: cups.service: Installed new job cups.service/stop as 6213
[...]
Dec 02 10:59:29 till-x1carbon systemd[1]: cups.service: Changed running -> stop-sigterm
Dec 02 10:59:29 till-x1carbon systemd[1]: Stopping CUPS Scheduler...

and then starts again:

Dec 02 10:59:30 till-x1carbon systemd[1]: cups.path: Got triggered.
Dec 02 10:59:30 till-x1carbon systemd[1]: cups.service: Trying to enqueue job cups.service/start/replace

but cups shuts down again:

Dec 02 10:59:30 till-x1carbon systemd[1]: Child 9764 (cupsd) died (code=exited, status=0/SUCCESS
)
Dec 02 10:59:30 till-x1carbon systemd[1]: cups.service: Child 9764 belongs to cups.service
Dec 02 10:59:30 till-x1carbon systemd[1]: cups.service: Main process exited, code=exited, status=0/SUCCESS
Dec 02 10:59:30 till-x1carbon systemd[1]: cups.service: Changed stop-sigterm -> dead
Dec 02 10:59:30 till-x1carbon systemd[1]: cups.path: Got notified about unit deactivation.
Dec 02 10:59:30 till-x1carbon systemd[1]: cups.path: Changed running -> waiting

and so on, and I suppose eventually it runs into the restart limit.

Unfortunately the description of all cups.* unit is identical ("CUPS Scheduler" so on a non-debug log it is a bit hard to see which of the three it is when it says "Stopped". But nowhere in Till's log I see that cups.path itself gets stopped (and neither cups.service) -- this is clearly missing for a reliable stopping.