I run a small mail server in a VPS host. I often get errors like
this in my panic log:
2009-02-09 12:23:47 1LWX15-00020Y-0R spam acl condition: error reading
from spamd socket: Connection timed out
I use ACL spam checking, using something like this:
deny condition = ${if < {$message_size}{SPAM_MAXSIZE}}
spam = Debian-exim:true
condition = ${if >= {$spam_score_int}{SPAM_SCORE_REJECT_INT}}
message = We do not accept spam here. (Score: $spam_score)
warn condition = ${if < {$message_size}{SPAM_MAXSIZE}}
spam = Debian-exim:true
set acl_m_spam_status = ${if >=
{$spam_score_int}{SPAM_SCORE_REQUIRED_INT}
{Yes}{No}}
set acl_m_spam_score = $spam_score
set acl_m_spam_required = SPAM_SCORE_REQUIRED_DISPLAY
set acl_m_spam_bar = $spam_bar
I don't really know what makes spamassassin sometimes take very long
to check messages, perhaps the system gets too busy, after all there are
several other slices in the same physical server. Anyway, this causes
messages to be temporarily rejected and stay queued on the sending
server. They do get delivered sometime, but with delays and possibly
out-of-order. Nothing critical, but annoying.
Is there a way to tell exim that if spamassassin is taking too long,
just let go and let the message be delivered anyway?
--
There are no great men, buster. There are only men.
-- Elaine Stewart, "The Bad and the Beautiful"
Eduardo M KALINOWSKI
eduardo@???
http://move.to/hpkb