[Greylist-users] Re: Why and how often greylisting fails against ordinal servers?

[TSS Support]

 >For Groupwise, see the following:
>  http://support.novell.com/cgi-bin/search/searchtid.cgi?/10084802.htm

Thanks. I hadn't really looked into it because it was easy enough to
whitelist the affected servers and tell them 'your server is broken, fix
it'. It'll be nice to have a reference to toss them at.

>Nope, and we see, all total, ~2million message delivery attempts each day.
>But, we are using 0.05 with at least one stability patch I previously
>posted to this list (it makes sure the database connection doesn't die, or
>reconnects if needed, I'd point you at this URL, but I can't seem to pull
>the patch from it:
>http://lists.puremagic.com/pipermail/greylist-users/2004-May/000586.html).
>
>
>-philip

Thanks. I just noticed this morning that the milter always dies at 6:25AM,
which is coincidentally when our logs our rotated. I suspect that mysql is
being restarted or something similar that is making the milter die. I
figured there's probably a (seemingly) unrelated fix in CVS and have just
been procrastinating on updating until it's called a 0.05. The system is
running threaded Perl 5.8.3, and apart from the very timely 6:25 deaths it's
been basically stable once it was set up. We did have some table corruption
but that cleaned up easily and I completely blame mysql for it. Our site
handles around 500k/attempts per day, split across two machines as much for
redundancy as for load balancing. They are each capable of running the
entire mail load, doing relaydelay->spamassassin->clamav and then passing on
to our main user mail server. They run at a load average of about 0.2-0.3
most all the time. Before relaydelaying, the same systems running the same
spam and virus filtering were running a load average in the 2-5 range
depending on daily mail load. Suffice it to say, we love it.

-Keith