mirror of
https://github.com/fosrl/pangolin.git
synced 2026-07-18 07:34:42 -05:00
Traefik shutting down randomly in Azure #82
Reference in New Issue
Block a user
Delete Branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Originally created by @jwhite on GitHub (Feb 12, 2025).
At random times (so far over night in our region) the Traefik server stops functioning and is not restarting.
Any ideas on how to debug?
@hhftechnology commented on GitHub (Feb 13, 2025):
Check if Traefik is being killed by the OOM killer
dmesg | grep -i killjournalctl -u docker.service --since "1 hour ago"@jwhite commented on GitHub (Mar 20, 2025):
@hhftechnology - Any ideas here?
Seems to crash every 7 to 10 days now, might be less frequent with the latest version.
dmesg reports nothing (with kill.)
It went down sometime last night, so here are the last 24 hours from journalctl. I didn't request the 'stopping' so I'm wondering if something in Azure told the containers to go down. For upgrade maybe? The more interesting thing is they don't come back up.
Cheers!
(Sorry for the ugly logs... not sure how to format for github.)
@oschwartz10612 commented on GitHub (Mar 20, 2025):
The logs show Docker was stopped and restarted at 14:03:08 on March 20. This appears to be a system-initiated action (systemd[1]: Stopping Docker Application Container Engine...).
Container restart failures: After Docker restarts, it tries to bring up the containers but encounters errors:
time="2025-03-20T14:03:15.561317851Z" level=error msg="failed to start container"
This could be related to Azure's automatic maintenance operations. Check your Azure portal for scheduled maintenance events or set up notifications for them. Do you have a machine type that gets stopped and started like AWS Spot instances?
Make sure you are not running out of resources triggering your machine to restart - memory or CPU.
@jwhite commented on GitHub (Mar 21, 2025):
I'll check the resources.
No, this is not on spot instances, but is a low resource vm.
I'll try to see if there are scheduled events. Either way --- it should come back up.
How do I find out why?
@hhftechnology commented on GitHub (Mar 21, 2025):
this is definitely oom issues
Mar 20 14:03:08 pangolin systemd[1]: Stopping Docker Application Container Engine...indicate Docker is being stopped by systemd
Try to increase resources a bit.
Last resort if you don't increase resources. check your compose path. otherwise, script will not run.
@jwhite commented on GitHub (Mar 24, 2025):
It was a memory constraint issue. I need at least 2gb on the host, or it will sit around 70% consumed, and then eventually get shut down. That seems a bit high, should I be adjusting my docker compose to set some better limits or has that already been taken into consideration?