I've created a custom service file which looks like this:

cat /etc/systemd/system/aaa.service

Description=aaa main application
After=syslog.target network-online.target



When checking status with:

systemctl status aaa

Everything looks ok, except for:

... systemd[1]: aaa.service: Supervising process 18285 which is not our child. We'll most likely not notice when it exits.

Wondering why this message is showing and if it is anything to be concerned about?

my ExecStart is a bash script named run_aaa.sh; What it does is prepare some environment variables, run some required maintainace scripts, and finally runs the aaa binary (as a daemon) with some dynamic command line options, in the background - and quits (the run_aaa.sh script quits, the aaa daemon binary keeps running in the background, and maintains its own pid file that systemd knows about)

I know systemd is running run_aaa.sh and expects it to fork (because I specified type=forking) and it does actually "fork" by running the "aaa" binary daemon in the background and exiting afterwards. Plus I do specify PIDFile path and my aaa binary maintains this PIDFile as it should, so why would systemd complain that it might fail to supervise my daemon?

Is there a better way to configure the service for this use case? (I must use run_aaa.sh script in order to execute my binary)

  • You may try to remove PIDFile, which is not required. Or run your binary with exec in the script. If you start your service, what is the parent PID for the aaa binary? Is it PID 1 or your shell script, which has not exited? Oct 27, 2017 at 9:19
  • Stop calling the shell script and let your systemd unit handle all that prerequisite stuff directly. Dec 6, 2017 at 21:00

1 Answer 1


If you need to run shell script only once, then just add ExecStartPre=/path/to/script and run your daemon with ExecStart=/path/to/daemon. The script will not run on restart, maybe this is not what you need.

If you need to run the script on reboots as well, then set Type=simple and run the daemon in foreground mode from shell script, so the script will not quit. But you still get that message and stop/restart actions are waiting 30 seconds before State 'stop-sigterm' timed out. Killing, very unpleasant. If you then add ExecStop=/bin/kill $MAINPID, the timeout problem goes away. Don't remove PIDFile. You will still see that message, but every action start/stop/restart/status work correctly. Maybe there is another better way to achieve the same. At least this works for me and the message may be safely ignored. When daemon quits, shell script will also quit, so systemd will know that.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .