Server gets immediately nuked on restart/stop, no chance to run shutdown events/hooks

[foosel]

It looks like we have a behaviour change in that regard from from OctoPi < 1.0.0 to OctoPi >= 1.0.0.

Example:

OctoPi 0.18.0:

2023-11-28 17:52:47,473 - octoprint.server - INFO - Shutting down...
2023-11-28 17:52:47,485 - octoprint.server - INFO - Calling on_shutdown on plugins
2023-11-28 17:52:47,489 - octoprint.events - INFO - Processing shutdown event, this will be our last event
2023-11-28 17:52:47,672 - octoprint.events - INFO - Event loop shut down
2023-11-28 17:52:47,921 - octoprint.plugins.tracking - INFO - Sent tracking event shutdown, payload: {}
2023-11-28 17:52:48,111 - octoprint.server - INFO - Goodbye!
2023-11-28 17:52:55,660 - octoprint.startup - INFO - ******************************************************************************
2023-11-28 17:52:55,663 - octoprint.startup - INFO - Starting OctoPrint 1.9.3
2023-11-28 17:52:55,665 - octoprint.startup - INFO - ******************************************************************************

OctoPi 1.0:

2023-11-28 17:58:10,696 - octoprint.plugins.github_release_patcher - INFO - Started dummy release server on http://127.0.0.1:50672
2023-11-28 17:58:10,696 - octoprint.plugins.github_release_patcher - INFO - Set Github release URL to http://127.0.0.1:50672/releases/{user}/{repo}
2023-11-28 18:01:21,363 - octoprint.startup - INFO - ******************************************************************************
2023-11-28 18:01:21,366 - octoprint.startup - INFO - Starting OctoPrint 1.9.3
2023-11-28 18:01:21,367 - octoprint.startup - INFO - ******************************************************************************

No shutdown events triggered. It looks like the systemd unit on 1.0 simply nukes away the server directly, whereas on OctoPi < 1.0 it still got a chance to cleanly shut down. That's something that needs to get looked at, but that's rather something for OctoPrint/OctoPi-UpToDate or guysoft/OctoPi.

Originally posted by @foosel in OctoPrint/OctoPrint#4882 (comment)