[systemd-devel] Odd status after core dump

[Ulrich Windl]

Hi!

For systemd-228-157.12.5.x86_64 (SLES12 SP5) I had an odd status:
Three services were re-started in parallel, and one of those failed (stop
routine killed itself with signal 6 due to a timeout):

Now the status systemctl displays looks like this (some details omitted):
● iotwatch at NFS1.service - iotwatch I/O performance monitor instance "NFS1"
   Loaded: loaded (/etc/iotwatch.conf; enabled; vendor preset: disabled)
   Active: active (running) since Fri 2020-07-31 11:31:57 CEST; 6min ago
...
 Main PID: 12246 (iotwatch-NFS1)
    Tasks: 4 (limit: 512)
   CGroup: /system.slice/system-iotwatch.slice/iotwatch at NFS1.service 
           └─12246 iotwatch-NFS1 -l /var/log/iotwatch/NFS1/iotwatch-NFS1.log
...

Jul 31 11:31:57 v04 systemd[1]: Starting iotwatch I/O performance moni.....
Jul 31 11:31:57 v04 systemd[1]: Started iotwatch I/O performance monit...".
Jul 31 11:31:59 v04 systemd-coredump[12190]: Process 7123 (iotwatch-NFS1...

                                                  Stack trace of thread 7123:
Hint: Some lines were ellipsized, use -l to show in full.

Now the odd thing is the sequence of journal entries:
The service started and then dumped core, but still is active?

Actually what happened was: A client told the server process to stop, but when
it didn't in time, it killed it with signal 6. So the old process stopped.
THEN the new process was started (PID 12246).

Is the ordering due to the fact that coredump completed after the new service
started? Or is it just systemd's archiving of the core dump file? I thought the
pid is visible until core-dump had completed....

Regards,
Ulrich