[HLUG] Exim 4.50 on Debian 3.1 external greet_pause?

[John Hedges]

> >The backup  should deliver
> >immediately when the primary is up not to mention that other delivery
> >agents should direct deliver to the primary when it is reachable (note
> >that I leave out spammers tactics to deliberately deliver to
> >secondaries as the anti-spam/AV measures are usually less there).

Could someone explain why it is useful to have a secondary MX routing
through the primary. If the primary is down then mail is queued on the
secondary - without the secondary, surely the mail would be queued on
the sending host (or returned, depending on their own policy)? It seems
like a lot of effort for very little advantage.

> I am also doing CLAM on the secondary, and at some point I will be
> integrating SpamAssassin (though I am not sure whether that is pushing the
> memory on the virtual machine).  
 
Clam AV and SpamAssassin work on my basic VMs without too much problem
(about 1000 mails per host per day). The only problem I had was with
large (>20Mb) emails where either clam or sa were loading the whole mail
and so exhausting memory. I now limit mails to 5 megs (Debian defaults
to 50).

> On another note, have you ever had to set up POP3 using virtual users (not
> in /etc/passwd)?  I am trying to do something using Solid-POP3D and Pam
> using htpasswd files, but get the general feeling I am re-inventing the
> wheel.

Courier handles virtual users well with its userdb support. I have
exim and courier configs for virtual domains and virtual users if you
are interested.

Cheers

John