It’s important to also share kumod --version along with that stack trace so that we can interpret it correctly.
the /metric is only showing connection stuff.
curl -i 'http://localhost:8000/metrics' | wc -l
11022
grep on memory will not show anything
ii kumomta-dev 2024.03.28.065631.1b451e3a amd64 A high performance, modern MTA
last time you looked at the config I stripped some snippets.. and the rest should be the same…
I will create a new config tar though
this is the config, without my dkim keys 
I just did an ‘apt install lldb’ so next time I can do that command
I am fine with the correct version.. does it make a difference that I created the binary on a vanilla Debian 12 machine?
we need to know the commit hash that was used to build it so that we can interpret the line numbers from the stack trace
that’s the purpose behind asking for the version
I am nearly done setting up a dev version of the production setup with KumoMTA and logstash. So probably at the end of the week I should be able to synthetically run tests. And with that maybe I can replay the issue.
hmm now tried to reproduce even with a sample email from a customer and on my test setup it just works. I have a setup with elasticsearch and a kumoserver with logstash and a dedicated smtp server to pickup the emails and dump them. I currently use 2 cpu AWS hosts in production.. but I am testing on an i5 11th gen Intel laptop with 32Gb and 8 virtual cores. The vhosts do have 3 cores atm. So I can reduce that to see if it makes a difference.
So test platform done. INitial tests with sending mail to an internal blacklist@domain.com address is working.. with like 100 messages per minute. (which is above what production was doing)
I tested with the 2024.04.03 and with a 2024.03.08 version
I will try to do some more testing tommorrow. I am pretty stumped atm. but maybe I have a bright idea tommorrow.
It always works when it knows it is being watched, but now you have the monitoring.
Update: without the webhooks it survived so far… only one ‘bigger’ sendout. I managed to create a script to have the logs in elastic afterwards using the same logstash http input.
Well today a client started to sendout with 3500 per 10 minutes and kumomta choked again. No response anymore … Added a kumo-bt.txt and the kumo version.
kumod --version
kumod 2024.03.28-1b451e3a
kumo-bt.txt (11.6 KB)
I uploaded my current config where I disabled the webhook.
What is the current instance size and specs?