[Bug]: stop_grace_period not honoured

[hanjo]

Operating system

Debian GNU/Linux 12 (bookworm)

Description

Hi,

I noticed that my DSM complains about unclean shutdown whenever I stop it. To fix this, I tried to increase the timeout and set stop_grace_period: 5m in my compose file. However, the VM is getting killed after exactly 50 seconds, which I believe is the threshold hardcoded in power.sh.

Now I don't fully understand how these two values are interacting, but I captured the log file for reference below.

Is there any way to increase those 50 seconds to the 5 minutes I want?

Thanks!

Docker compose

stop_grace_period: 5m

Docker log

May 20 17:48:37 server c6dc96d5f0f6[7140]: ❯ Received SIGTERM signal, sending shutdown command...
May 20 17:48:37 server c6dc96d5f0f6[7140]: ❯ Virtual DSM is now ready to shutdown...

[...]

May 20 17:49:27 server c6dc96d5f0f6[7140]: ❯ ERROR: Shutdown timeout reached, aborting...
May 20 17:49:27 server c6dc96d5f0f6[7140]:
May 20 17:49:27 server c6dc96d5f0f6[7140]: ❯ ERROR: Forcefully terminating QEMU process, reason: 143...
May 20 17:49:27 server c6dc96d5f0f6[7140]:
May 20 17:49:27 server c6dc96d5f0f6[7140]: qemu-system-x86_64: terminating on signal 15 from pid 7 (bash)
May 20 17:49:29 server c6dc96d5f0f6[7140]:
May 20 17:49:29 server c6dc96d5f0f6[7140]: ❯ Shutdown completed!
May 20 17:49:34 server 8fcb86069ecc[7140]: ❯ Starting Virtual DSM for Docker v7.36...

[...]

Screenshots (optional)

No response

[xrh0905]

Looks like a duplication of #439 to me.
The current version of power.sh indeed hard-coded a 50 second timeout on shutdown request. Maybe convert that into a ENV?

[hanjo]

Not really a duplicate, but related, I agree. This parameter configurable through an environment variable would be a good solution, I think. Another option - or maybe in addition - but not sure if possible, would be to just make the script kill the process after stop_grace_period - 10s (or whatever), instead of the hardcoded 50s.