[MM3-users] [Solved] Strange behaviour when SSL (https) is activated

Le 18996ième jour après Epoch, François TOURDE écrivait:

But I need to investigate a bit more about 2 or 3 issues I just noticed. I got oom-kill on uwsgi (argh !), and some timeouts (but not clearly related to) on haproxy side.

Finally I found a bad timeout value (500ms on a slow and old Xen host supporting too much VMs) which caused the issue.

I moved 2 VMs on another Xen, freeing enough memory to increase mailman VM size. No more oom-kill now.

My haproxy config had global timeouts, backends timeouts ... plus a specific one on my frontend https config. Removing this one solves the problem. Due to Ansible deployment, and because my frontend is global for all URLs, I didn't notice it :(

Thanks for your support !