How to set a delay between service exiting and service restarting?

Product support
1

Hello!
We have a python application that exits with a runtime error, then it is immediately restarted automatically. Is there a way to put a delay between the application exit and when the supervisor restarts the service?
See logs for timestamps:

[Logs]    [2023-08-30T05:59:21.447Z] Service exited 'main sha256:d32dcf1d7d5b77dfe39c2d67298689918443b7c077dcc43508e3892199e0d0c7'
[Logs]    [2023-08-30T05:59:22.170Z] Restarting service 'main sha256:d32dcf1d7d5b77dfe39c2d67298689918443b7c077dcc43508e3892199e0d0c7'
4

Hi there,

Unfortunately I don’t believe that there is a way to do this right now. Could you describe the issue that automatically attempting to restart the container causes for you, maybe there’s other workarounds that might help you

7

Hi,

I’ve tried something similar a while back.
My solution was to simply write a wrapper script/entrypoint.
This wrapper

  • Does preparations needed for my application, like fixing certificates;
  • Calls the actual application;
  • Does cleanup after my application exits

I use an environment variable to tell the wrapper what to do when my application returns.
This variable translates to:

  • do nothing (sleep forever)
  • wait X seconds before exiting the container (causing supervisor to restart it)
  • request a reboot from supervisor
1 Like
8

If you are using Systemd as your supervisor on a Linux system, you can modify the service unit configuration to introduce a delay before restarting the service. You can do this by editing the service unit file (typically found in /etc/systemd/system/) and adding a delay using the RestartSec option. For example:

[Service]
ExecStart=/path/to/your/python_script.py
Restart=on-failure
RestartSec=10 # Add a 10-second delay before restarting