SOLVED: RE: Some mail (up to 7 days old) is stuck in /var/spool/mqueue

[Chris W. Parker]

Thanks to everyone who participated in this discussion and tried to
help. At this point I've managed to get the mails out of the queue and
to their recipients.

Some of you may remember about a thread I started on July 19th (hint
hint) called "'I/O error on connection' problem. MailScanner related?".
In that thread Julian suggested that I adjust the sendmail QueueLA to
something greater than the norm of 8. I thought, "Why increase the
number? Instead I'll just temporarily set this to unlimited." So I put a
0 for QueueLA and RefuseLA. I saw another option a few lines down that
says:

# load average at which we delay connections; 0 means no limit
#O DelayLA=0

So I took this to mean that 0 could be put in for all the ***LA values.
I'm not sure if I was wrong or not with this assumption but this change
is apparently what caused the problem.

While looking at the heart-pounding, gut-wrenching, ever growing output
of 'mailq' I noticed that each of the message IDs had an X at the end.
So I went to the man page to hopefully find out what that meant. While
reading I noticed:

"... The status characters are either * to indicate the job is being
processed; X to indicate that the load is too high to process the job;
and -  to  indicate that the job is too young to process."

Aha! Load average. So I stopped MailScanner, went back into sendmail.cf
and changed the 0 values back to their original values, moved all the
queued messages back into mqueue.in and then restarted MailScanner.

While watching maillog with tail -f I noticed that one of the message
IDs I was familiar with finally had a line that said "Queued mail for
delivery". Great!

After a few minutes of repeatedly executing mailq I noticed that as the
queue filled up, it also started to be cleaned out. Phew!

So for all of us in the future (although this problem is probably pretty
rare) if you see someone's output from mailq and all their messages have
an X at the end of the message ID, it's probably a good idea to look at
the machines load average and possibly adjust the settings in
sendmail.cf.


Unfortunately though having had this experience I'm afraid to say that I
will probably be moving away from MailScanner and going to an all
Exchange solution. I attribute this to a high learning curve and a very
beginner Linux admin. It's also a second Linux box that I have to keep
track of and maintain and one is enough already (web server) for someone
like me! :)

Anyway, this community is easily one of the best I've ever participated
in and I'm glad to have gotten used to some of your names. I haven't
been able to contribute much but I've definitely had a lot of questions
answered. :)



Chris.