[Mailman-Users] listmembers being "(ignored)", lockfile problems, 2.0b5

[Brett Dikeman]

Greetings.

This is a edited version of a post I sent to the developers this 
morning.  No response, so I can only assume no one cares to respond.

I searched through my archives.  I couldn't search through the list 
archives because the search engine on python.org appears to be 
broken(and yes, I followed the instructions, choosing only SIG 
archives.)

I found a bunch of posts related or similar to my problems, but not 
much in the way of resolutions or anthing that seemed to help me.

The setup we're running is Solaris, python 1.5.2, which I compiled 
myself.  The list was working great, until people started loosing 
mail delivery. I thought the auto-bounce handler was removing people, 
but they're still listed as subscribed, and I never received a 
message from the bounce handler saying they would be removed.

Now, what I'm seeing in the logs:

in logs/smtp-failure:

smtp-failure:Aug 30 11:16:15 2000 (122) -1 <email addr> (ignore)
(many, many times)

This user is in fact not getting email from either of the two lists 
he belongs to.  The number in ()'s changes, though not with every 
entry in the logs.

and, in logs/post, I see constant entries about:
Aug 30 04:53:08 2000 (23571) post to <listname> from 
<listname>-request@<hostname>, size=28501, 177 failures

(always the same number of failures and size; seems to be some sort 
of unrelated mail loop of some kind.)

and there are posts that go through:

Aug 30 04:44:42 2000 (23411) post to <listname> from <some 
subscriber's email address>, size=528, success

Qrunner seems to be working:

Aug 30 02:55:04 2000 (22304) qrunner begining
Aug 30 02:55:09 2000 (22304) qrunner ended

(this is repeated every time qrunner is launched, which is wasteful 
in terms of space since it runs once a minute; I've got 56k of text 
for less than 12 hours of running)

Now, this mystery "177 failures" message shows up again:
Aug 30 06:50:10 2000 (25068) All recipients refused: please run connect() first
Aug 30 06:50:10 2000 (25068) smtp for 177 recips, completed in 2.620 seconds

In logs/bounce, I'm not sure I understand the meanings:

Aug 29 23:13:31 2000 (18077) <listname>: <someuser> - 27 more allowed 
over 426977 secs

then there will be:

Aug 29 23:13:33 2000 (18077) <listname>: <differentuser> - 0 more 
allowed over 395006 secs


The non-delivery problem affects digest and non-digest users. 
Archives are working fine.  One list is 1,500 people; 500 normal, 
1000 digest users; the other is about 100 people, mostly non-digest. 
We're using sendmail, and I have the threaded delivery turned on due 
to the large number of slow/bad mail servers(and experience has shown 
that splitting them up into chunks results in a dramatic improvement 
in delivery times.)

Some mail is getting by, and some users report that after 
re-subscribing, their problems have gone away, although I made some 
changes(like clearing out qfiles/ and also removing stale lockfiles) 
around the same time, so it's hard to tell.)  However, several users 
are still not getting any mail no matter what I do(they do get the 
list welcome message when I re-subscribe them, however.)


Thanks all for any help/tips.  I wouldn't mind applying some patches 
that have been introduced since 2.0b5 that will fix this stuff, if 
they're available.

Brett