[exim] queue problem needs troubleshooting

I have an interesting problem: I have customer whose outgoing e-mail is
queuing up and not going out. Currently he has about 1000 outgoing messages
on his server.

Here are some of the relevant details:

The server is a web host using Red Hat Linux and cPanel.

Mail delivers to recipients with accounts on the server.

The domain resolves correctly, the server resolves to the domain.

Debugging from the command line seems to indicate that addresses are being
resolved and passed successfully.

Forcing the queue to send shows multiple attempts to send to mail servers at
a large ISP (for instance earthlink.net). The message is "operation timed
out" and 6 more attempts are made to try to deliver the mail before
deferring it. (This seems to take up a lot of time.) Removing messages to
earthlink.com recipients moves the problem to other ISP's such as yahoo.com.

Multiple messages are bounced and re-tried to recipients who have full
mailboxes (over quota).

There is a message "Unfrozen by forced delivery" which pops up occasionally,
but doesn't seem to have any other parameters.

Some mail seems to go through after 5-7 days. The problem has been going on
for two weeks and I've only been working on it for two days, so I can't make
distinctions between the mail that's being delivered and the mail that's
not.

All server users experience the same problems.

telnet troubleshooting to recipient servers (for instance, earthlink.net)
seems to work except for those servers that specifically exclude telnet
connections on port 25.

I'm new to exim, so I'm open to any suggestions. I'd specifically like to
move mail from the queue to another directory and take messages one at a
time to test their delivery. I'd also like to determine if the message order
has anything to do with the problem. (For instance, are messages queuing up
behind an undeliverable message?) Can I move failed messages to another
directory for examination? Can I limit the number of re-tries on timed-out
smtp connections and then move the mail? What specific logging levels do you
suggest I set, and where's the best place to put the logs, and what would
you look for in the logs if you had the problem? I'd kind of like to be able
to identify the delivery order of the queued up mail and know what the rules
are for deferring failed messages.

That's a lot, but I thank you in advance for any help or advice.

Mike Burke

"There are 10 kinds of people; those who understand binary and those who
don't."