Kumo locking up not responding at all

so I will checkout the git repo anyway and see if I can be of any help.. my Failure logs will help out to complete the list over time.

Thank you!

Hmmm after running the kumomta-dev for a while the same issue is popping up again. I now run a self-build Debian 12 package of kumomta-dev 2024.03.08.133648.3b539b3c . And now with debug stuff enabled it logs this:
ar 20 15:22:03 ip-172-31-17-90 kumod[10649]: 2024-03-20T15:22:03.934017Z DEBUG logger kumod::logging: deadline reached, running expiration for this segment
Mar 20 15:22:03 ip-172-31-17-90 kumod[10649]: 2024-03-20T15:22:03.934286Z DEBUG logger kumod::logging: Flushed “/var/log/kumomta/20240320-152103”
Mar 20 15:22:03 ip-172-31-17-90 kumod[10649]: 2024-03-20T15:22:03.934326Z DEBUG logger kumod::logging: waiting until deadline=None for a log record

At first I thought it was the injection of more mails causing issues, but now I disabled the injection port and restarted the Kumomta to only do the sendout. The odd thing is that kumomta is not responding at all… kcli doesn’t do anything. SMTP port injection does not respond at all.

After some more testing it looks like it breaks on sending webhooks too quickly to logstash. So I added a queue_config to it… And after that it looked to be working again. Still not 100% convinced though.

I had issues stopping kumomta… but when it did I could start it and it would do some webhooks again and then it would hang again. Really weird.
message.txt (8.24 KB)

Just to clarify, it is writing logs during the period where it does not respond to port 25 or the HTTP calls?

no…

it does absolutley nothing

well actually when ops tracing on the binary it is memory checking all the time

I did trace -smtp and that disconnected with websocket error

I just found something else too.. but not sure if that is related or that this is about the max retry time of 20 minutes in the webhook:
requeue_message{increment_attempts=true delay=None}: kumod::ready_queue: Error while calling requeue_message event: invalid return type for message_requeued event. will reuse current queue

That was fixed in the -dev from March 17, how recent is your build?

2024.03.08.133648.3b539b3c

oh well..

I need to build a new one then :slightly_smiling_face:

Yeah, time to update and rebuild and let’s see what happens.

but not sure if that is related to the hanging

Always a good step to make sure the latest is in, it may help with the hanging, it may not, but at least we’ll eliminate some variables.

I already had a pretty new one I thought :slightly_smiling_face:

We move very quickly around here. :wink:

well I injected some new mail.. since the customer was still mailing and it crashed again

Ok, need to see your configs so we can further investigate.