Auteur: Brian Datum: Aan: exim-users Onderwerp: Re: [exim] exim 'problem'?
Jethro R Binks wrote:
> On Fri, 5 Nov 2004, Brian wrote:
>
>
>>About two months ago we put exim into 'daemontools', running great not a
>>problem at all. A few days ago (monday) we decided to try running with the
>>exim spool file in a freebsd mfs (memory file system a.k.a. ram-disk).
>
> ...
>
>>So my question is, is exim mounted on a mfs partition not supported? Are
>>there some other options that we would need to set either in exim or
>>FreeBSD to make this working?
>
>
> I think before you ask that question, there is something more fundamental
> to be considered.
>
> Do you really want to have exim's spool, which contains the queue of
> mail-yet-to-be-delivered, on an in-memory filesystem whose contents will
> be deleted at system boot (or crash) time?
>
> Or does this mfs have some magic beyond what you've described?
>
> Jethro.
>
> . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
> Jethro R Binks
> Computing Officer, IT Services
> University Of Strathclyde, Glasgow, UK
>
Exim and freebsd (for us) has never been anything but stable, we've
never had a 'panic' or crash.
the problem turned out to be something with spamassassin as there were a
few mb's of mail in the scan dir. and a few hundred messages in the
que. The machine wasn't finishing up smtp properly and people we're
timing out.. so they would reconnect and try and send again..
Since exim had taken the message but not issued a 250 OK to them, they
kept trying to redeliver it. After stopping inbound delivery and
restarting spamassassin and doing a -qqff; we've flushed the messages
and are delivering messages as usual.
sa is 3.0.1_1 and clamd is 0.80 again all running in daemontools and
just must have been a fluke SA thing.
this seemed to have been the first entry in the panic log:
2004-11-04 18:19:39 1CPqod-000KTT-Ot spam acl condition: cannot parse
spamd output
then tons of these:
2004-11-05 09:15:57 1CQ4sV-0007Rl-0c spam acl condition: error/short
send to spamd
2004-11-05 09:15:57 1CQ4q8-0007Lk-1U spam acl condition: error/short
send to spamd
2004-11-05 09:15:57 1CQ4q2-0007Ld-20 spam acl condition: error/short
send to spamd
2004-11-05 09:15:57 1CQ4qx-0007O1-GV spam acl condition: cannot parse
spamd output
..