Re: [Exim] handling HUGE backlogs

On Wednesday, December 5, 2001, at 01:07 AM, Chris wrote:
> We have a rather substantial backlog to a company who has been
> experiencing
> severe mail problems as of late. By substantial, I mean, well, we have
> *several hundred thousands* of messages queued up on a handful of
> machines.
> The queues are massive.

We have this problem all the time. There are two easy solutions that we
have come up with.. One is much faster than the other.

(1) On the same machine or another machine, set up an exim install that
uses a different directory base
(spool_directory=/path/to/somewhere/else). If it is on the same
machine, you obviously need to set it on a different IP and handle the
delivery to a different port -- I won't go into those details. In the
director configuration add a domain_director before lookuphost: like:

yahoo_director:
driver = domainlist
transport = remote_smtp
route_list = yahoo.com 10.0.0.138 byname

This one obvious handles the yahoo.com domain (replace it with your
problem domain) and 10.0.0.138 is my other machine that is dedicated to
yahoo.com problems.

This second exim instance should be in queue_only mode. As only emails
to this problem domain will be in the queue for that exim instance, it
will *always* find a reusable connection.

(2) Same thing, but do some magic with perl or exim -bpr and sh to find
all the clogging messages and simply move them to the new queue. We do
it this way because it takes far less resources.

It is also possible to take the queued messages and turn them into a
BSTMP stream. At that point, contact the other admin and FTP the file.
Let him handle the grunt work. I guess that all depends on who you are
and the nature of the messages -- do they want their users to get your
messages or do you want their users to get your messages? :-)

--
Theo Schlossnagle
1024D/82844984/95FD 30F1 489E 4613 F22E 491A 7E88 364C 8284 4984
2047R/33131B65/71 F7 95 64 49 76 5D BA 3D 90 B9 9F BE 27 24 E7