Mqueue.in huge

[Glenn Steen]

On 23/01/07, Marcello Anderlini <m.anderlini at database.it> wrote:
> Thanks for your answer.
> I've checked in in /root/.spamassassin and I found just this files.
>
> -rw-------  1 root root 168087552 Jan 23 10:24 auto-whitelist
> -rw-------  1 root root     97320 Jan 23 10:24 bayes_journal
> -rw-------  1 root root 172904448 Jan 23 10:24 bayes_seen
> -rw-------  1 root root 322408448 Jan 23 10:24 bayes_toks

Ok, no expire files there.... not that problem then:-).

> This it's an extract of my Mailscanner.conf
> ==========================================================
> Spam Score = yes
> Cache SpamAssassin Results = yes
> SpamAssassin Cache Database File =
> /var/spool/MailScanner/incoming/SpamAssassin.cache.db
> Rebuild Bayes Every = 0
> Wait During Bayes Rebuild = no
> ==========================================================
> It's correct ? Should I change something ? Maybe try to use an other
> spamlist ?
You could go for a force expire solution anyway. But you don't really
have to, no.

Next thing to check is if something takes a lot of time completing...
Simplest is to run a test message through (so that you would get the
network tests too) spamassassin. That way you'll see if any of the
digest tests (both Pyzor and DCC have been known to mess things up (in
different ways:)), or BLs, or individual rules seem to take forever.
If you have MailWatch, it has a very nice "timed breakdown" of a lint
run, but unfortunately this will not help you (since you are runnig
3.1.7, which has the network tests turned off for the --lint test).
Did you use Pyzor? Do you use the "alternate" sever at
82.94.255.100:24441
... The default one always seem to time out, this doesn't, so use this one.

Martin Hepworth have posted a few BLs he "habitually" turns off in
SpamAssassin, search the list archives for those... Might help you.
Also look for DCC timeout problems (I don't rightly recall what that
was about... Centered on dccifd IIRC).

Cheers
-- 
-- Glenn
email: glenn < dot > steen < at > gmail < dot > com
work: glenn < dot > steen < at > ap1 < dot > se