The 15 minute delay is the default delay from sendmail in processing
the mailq.
I've seen this happen on our all to old mail server that has about
3 times as many people as it was designed to handle on it.
Procmail actually would hang trying to do local delivery and
leave the message in the mailq where delivery would be attempted by
sendmail in the 15 minute mailQ run.
In our case.....
We receive a lot of mail (for the server we are running). procmail
is our LDA and lock the mail spool file for the user.
That user receives another mail (or two or ten), befor the first
procmail is done writing it's message. This is mostly due
to the processing speed of the machine (Sparc 20 with 60MHz
processor), and that /var/mail used to be on an overly congested
disk. We've added services over time, but inadverently had
/var/mail, /var/spool/mqueue, our procmail.log and our mailman
mail archive logs all on the same disk (which is slow because
of the 10MB/s interface on the Sparc 20).
Any given message required writing to the same disk 3 times, or
4 if it was a mailman mailing list message.
If we received 4 emails to mailman mailing list, our I/O on the
disk would go to 100%. We moved the directories around and
isolated them on different disk and life is better....
Now to the really bad part. We are trying to replace our mail
server and happen to have a spare E4500 that's not doing a whole
lot, so it was going to become our mail server (to replace 4
Sparc20's running mail).
After some initial test, we can still overload our mail server
and cause procmail processes to hang. We did this by sending
4000 emails to the same account. It wasn't a disk I/O issue
in this case. We had 3000 procmail processes trying to access
the same inbox and waiting on locks and whatnot.
What we ended up doing was switching to Maildir format and now
we are going to have to really rethink our test because our
same test before only has 4 outstanding procmail processes worse
case. We don't know if this is going to be our final implementation
of our new mail server, but it looks promising. With this setup,
we can handle our current volume of mail in 1 day in about 15
minutes.
Chris G.
Carles Pina i Estany wrote:
Hi,
On my university we have a server running procmail.
If we see mailq, we can see that some e-mails stay on queue 15 minutes and
then e-mail is sent (without problem).
It is sent to local machine!
Normally, the mails are sent on the moment.
Anybody has any idea about it?
Thank you very much
----
Carles Pina i Estany
E-Mail: cpina(_at_)linuxfan(_dot_)com || #ICQ: 14446118 || Nick: Teufeus /
Pine
URL: http://www.salleurl.edu/~is08139
-Que significa Windows NT? -Windows que No Tira :-DD
_______________________________________________
procmail mailing list
procmail(_at_)lists(_dot_)RWTH-Aachen(_dot_)DE
http://MailMan.RWTH-Aachen.DE/mailman/listinfo/procmail
_______________________________________________
procmail mailing list
procmail(_at_)lists(_dot_)RWTH-Aachen(_dot_)DE
http://MailMan.RWTH-Aachen.DE/mailman/listinfo/procmail