Re: SpamAssassin timed out [Scanned]

[Justin Mason <jm...@jmason.org>]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1


David Thurman writes:
> On 10/20/04 8:40 AM, "Matt Kettler" wrote:
> 
> > Mailscanner is inappropriately impatient with SpamAssassin. It's timeouts
> > were designed in the pre-bayes era, and are not designed to accommodate
> > bayes housekeeping chores like expiry and journal syncs.
> > 
> > In the short term, you can help by running a sa-learn --force-expire on
> > your bayes DB.
> > 
> > In the longer term, here's some suggestions I use on my own MailScanner
> > server: (I use all of these together)
> 
> Matt
> 
> I wanted to personally say thanks! We too had time out issues with the new
> 3.0 and MS. Certain spams happened to sneak in randomly and looking at the
> headers showed SA timing out. Applying your suggestions works! This morning
> when I came in to check our servers and mail, no SA time outs on any mail.
> 
> Kudos' again for a great tip. Needs to be in the FAQ of SA or at the least
> MS.

That's a good idea, actually...  worth adding to our FAQ.

- --j.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)
Comment: Exmh CVS

iD8DBQFBerBFMJF5cimLx9ARAhzoAJ49McXHDSvqFEpMJsKgKAJxIHKLqwCfXXRq
cyUjxSpvKHPlv7Q68Q1VdOE=
=tvmo
-----END PGP SIGNATURE-----

Re: SpamAssassin timed out [Scanned]

[Matt Kettler <mk...@evi-inc.com>]

At 03:25 PM 10/23/2004, Justin Mason wrote:
> > > In the longer term, here's some suggestions I use on my own MailScanner
> > > server: (I use all of these together)
> >
> > Matt
> >
> > I wanted to personally say thanks! We too had time out issues with the new
> > 3.0 and MS. Certain spams happened to sneak in randomly and looking at the
> > headers showed SA timing out. Applying your suggestions works! This morning
> > when I came in to check our servers and mail, no SA time outs on any mail.
> >
> > Kudos' again for a great tip. Needs to be in the FAQ of SA or at the least
> > MS.
>
>That's a good idea, actually...  worth adding to our FAQ.

As a side-note, I've been talking with Julian. He's decided to up the 
default timeout from 40 to 75 seconds, but reduce the excessive timeouts 
threshold from 20 to 10 consecutive timeouts. This should help 
considerably, although I think that disabling bayes_auto_expire will be 
required to make it work well.

(side note: I've seen SA 2.64 run off and run an expiry process during 
message delivery anyway, even with bayes_auto_expire set to 0 in local.cf. 
It happens rarely, but it does happen here about once a month. Not sure if 
this happens in 3.x or not but I'll check for it when I get round to the 
upgrade. If I detect it there, I'll investigate and open a bug as needed.)