Mailbox stuck in ratelimit "to"

sean

I had a mailbox that was hitting ratelimit “to” continuously, although it was not actually getting any new mail in.

It got a few spams in a bunch, nothing like the 100/1s the limit is supposed to be, but then got rate limited “to” and stayed in this state for ~ 12 hours, rejecting all lincoming emails.

Restarting the rspamd container had no effect.

I enevtually changed the rate limit in the ratelimit.conf file to reduce the to rate to 200/1m , and this fixed it.

Perhaps it was just the container rebuild that fixed it, don’t know.

Just a heads-up in case this is a new rspamd bug turning up.

BiboX

Thank you for the hint.

I do not understand why the ratelimit is being triggered at all. It is not activated for the domain or mailbox.
The problem only occurs for one mailbox.
Any ideas? Thx!

pkernstock

It looks like this could be related to the existing GitHub issue?
See: mailcow/mailcow-dockerized5168

sean

That’s sounds like it exactly, don’t know how I missed it!

The linked rspamd bug doesn’t sound like a good root cause though. Just double-counting the recipients woulnd’t get anywhere near the default limit, and doesn’t explain why it doesn’t reset.

Given that lowering the limit to 200 / 1m fixed it for me perhaps there’s a math issue somewhere that isn’t working with the short time “bucket leak” rate.