The reason I ask is because my service takes a significant amount of time to start up and so I’d like to know when the service will be automatically restarting

@Chaoyu it also restarted after getting this error:

Traceback (most recent call last): File "/home/giants/.local/lib/python3.7/site-packages/tornado/ioloop.py", line 921, in _run val = self.callback() File "/home/giants/.local/lib/python3.7/site-packages/circus/util.py", line 1039, in wrapper % arbiter._exclusive_running_command) circus.exc.ConflictError: arbiter is already running arbiter_stop command

I’ve always gotten this error, and when I tried to see how to fix it, it seems there’s not a good fix for it. It’s less about fixing this specific error and more about the fact that the service restarts at certain times and then while it’s restarting, my request lags.

I’m not sure if it’s related but I have (purposefully) cpu-throttling set to false.

Also, when I say my service restarts, I’m referring to the service.py script being run again. I know this is happening because some of the time consuming tasks which I have logs for are showing up in the logs. It just got re-run for no apparent reason (even after only a handful of 200 requests). So my question is: for what reasons / when does service.py get run?

@Chaoyu would it be possible to get 30 minutes on the calendar with you or someone else on your team? I wonder if my question is just misguided to begin with, and maybe if I walk you through the code that something else will jump out. Monday, Tuesday, or Wednesday would be particularly good days for me. 🙏

I've received the service code, but it seems I don't have the models so I can not reproduce it. Would you mind sharing a exported bento? You can get it by calling

bentoml export <your_bento_name>