[Mailman-Users] postfix+mailman w/o chunking?

Chad Walstrom walst005 at tc.umn.edu
Fri Jan 3 22:01:20 CET 2003

(Please include me in the "Cc:" as I'm not a member of the list.)

OS Sun Solaris 2.6 + security patches
Postfix Version: 1.1.12
Mailman Version: 2.0.13

I am having problems getting postfix to deliver emails to large
recipient lists without chunking (SMTP_MAX_RCPTS=0).  I had found the
problem to be related to postfix's default settings for
smtpd_recipient_limit.  After raising that to a setting more akin to the
size of the lists (10,000), an almost successful delivery of a message
to this list happened around 7:00 this morning.  The postmaster email
address received this, however:

	Out: 250 Ok
	In:  rcpt TO:<hoyt at uwec.edu>
	Out: 250 Ok
	In:  rcpt TO:<bernl001 at umn.edu>
	Out: 250 Ok
	In:  rcpt TO:<holmx034 at umn.edu>
	Out: 250 Ok
	In:  data
	Out: 354 End data with <CR><LF>.<CR><LF>
	Out: 451 Error: queue file write error
The list has 2290 recipients.  The qrunner picked it up at:

	Jan 03 03:53:25 2003 (16522) Reading qfile:

It quit processing this message at:

	Jan 03 07:06:46 2003 (8432) Keeping message.

(These messages I've added in the qrunner script).  You've seen the
error it produced.  There are two problems I see with this.  First is
that this is awfully slow (probably because my overly strict
smtp_recipient_restrictions).  Second, it choked and held up the rest of
the hosted lists by staying resident in memory.  The error in the error
log is as follows:

Jan 03 07:06:46 2003 qrunner(8432): Traceback (innermost last):
Jan 03 07:06:46 2003 qrunner(8432):   File "/opt/mailman/cron/qrunner",
line 292, in ?
Jan 03 07:06:46 2003 qrunner(8432):      kids = main(lock)
Jan 03 07:06:46 2003 qrunner(8432):   File "/opt/mailman/cron/qrunner",
line 200, in main
Jan 03 07:06:46 2003 qrunner(8432):      lock.refresh()
Jan 03 07:06:46 2003 qrunner(8432):   File
"/opt/mailman/Mailman/LockFile.py", line 204, in refresh
Jan 03 07:06:46 2003 qrunner(8432):      raise NotLockedError
Jan 03 07:06:46 2003 qrunner(8432): Mailman.LockFile . NotLockedError  

So something VERY strange is going on here.  

Since then, I've removed the file from the queue and tried to resend it.
I've also set the SMTP_MAX_RCPTS=500.  Things seem to be much more
successful, as the queue is filled with bounces for that list regarding
the resent message, now.  However, the last time the message was loaded
by qrunner was:

Jan 03 14:12:24 2003 (9396) Reading qfile:

It completed the job at:

Jan 03 14:52:05 2003 (9396) Keeping message.
Jan 03 14:52:06 2003 (9396) qrunner ended

The post log reported 40 failures instead of 2286!  Yay! ;-)  Looks like
most of those are legitimate "Domain not found" errors.

I've rearranged the smtp_recipient_restrictions settings a bit so that
hostname lookups and rbl stuff happen after the check_relay_domains.
I'm hoping that'll speed things up a bit.

Do you have any suggestions on how to fix the aforementioned problem.
Any idea why the lock attempt would fail?  Any ideas on how to speed
things up?  

Chad C. Walstrom <walst005 at tc.umn.edu>                  247 Gortner Hall
Asst. Coordinator of IT                                 Phone: 612-625-6208
College of Biological Sciences, University of MN

More information about the Mailman-Users mailing list