Hi Stephen,
Thank you for the quick reply. I found the issue and fixed it. The flood of .mbox.gz archive export requests from web crawlers overloaded uWSGI workers, leading to instability.
Kind regards, Prasanth
On Sat, 5 Apr 2025 at 19:52, Stephen J. Turnbull <steve@turnbull.jp> wrote:
Prasanth Nair writes:
Our Mailman3 server crashes with OOM and high CPU usage. As soon as we start the mailmanweb service (uwsgi process), the system crashes. It was working fine, but this issue started recently. We haven't changed any recently.
Any idea how to fix this?
You're not giving us anything to work with, except "get more memory for the server".
or any suggestions?
By "system crashes" do you the Mailman system, or does it take the OS with it?
By "no changes" do you mean no change to Mailman, or do you mean *nobody* has touched *anything* on the system recently (no new applications, no upgrades of existing applications or system daemons, etc)?
Tell us system resource limits. Limited CPU should not cause a crash, and I'm not aware of reports that CPU was a limiting factor in a Mailman problem. But a Mailman 3 system needs at least 2GB of memory to function properly unless you're really crafty about configuring it, and 4GB to be really comfortable. Do you have swap enabled? If not, 2GB-4GB of swap should get you past the crash, although if there really is that much memory pressure performance will probably be terrible.
Check mailman.log, mailmanweb.log, error.log (usually in .../mailman/var/logs). I don't know what you're looking for, so start by looking at many pages of logs when mailman was working normally. Then look at the recent ones when it started crashing, and just report "anything that looks different".
Check memory usage and CPU load just before starting mailmanweb.
Steve