[Mailman-Developers] URGENT Please Help: Duplicates with 2.0 final

[Todd Joseph]

BW>
BW>Naw, you just have to offer to finish my taxes for me.  Fortunately
BW>for you, I'm on hold with the IRS so I get to help you <wink>.

I know the feeling.  Ever filed an extension? :)

BW>
BW>    TJ> Oh, and I'm running Python 1.5.2
BW>
BW>And Mailman 2.0.3?  I'd recommend upgrading even though I doubt any of
BW>the fixes between 2.0 and 2.0.3 will address your problems.

I'm all for going to 2.0.3 but I want to find whatever went wrong
first.  I'm wary of upgrading any masking/removing the problem before
it's found.

BW>The error your getting in the log file indicates that one of the
BW>queued files is getting deleted out from under qrunner.  That could be
BW>because you're getting multiple qrunner processes because of some
BW>locking failure, or because your qrunner lock is getting stomped on.

I kinda thought so.  Any ideas on what could cause that?  

One more tidbit, I have set the "Action when critical or excessive
bounces are detected" to "Do nothing" So, I do have a fair number of
bogus addresses in my list.  My sendmail queue does get kinda large.
I've set it up this way out of paranoia in maintaining the integrity
of the list (IE not removing valid email addresses).

BW>Have you changed any of the default values in your mm_cfg.py (or
BW><gasp> Defaults.py) file?  

OK, I'm bad.  I changed them in the Defaults.py file.  I tried to use
the mm_cfg.py file, but ran into a problem (was a few months ago,
don't remember what exactly happened). I've included a copy of my
"Defaults.py" and "mm_cfg.py" at the bottom of this message for
reference.

BW>You ought to see a bunch of "Could not
BW>acquire qrunner lock" messages in the logs/qrunner file.  Perversely
BW>enough, that indicates that qrunner's lock is working.

:) yes.

BW>Are you still seeing duplicates, or was this a one-time occurance?

I'm currently using Mailman to send announcements about once a month
which are critical to our business.  The duplicates were sent with the
last announcement and we need to send another next week (wouldn't you
know it, Marketing picks this week to send 2 in a month!).  I can't
send another announcement until I can be sure we won't have any more
duplicates.  We'll probably outsource next weeks mailing to someone if
I can't get this resolved this weekend.

BW>Which DELIVERY_MODULE are you using?

SMTPDirect (would I be better off with Sendmail?)

BW>  If SMTPDirect, are you using any of the experimental thread stuff? 

Not that I know of.  See attached Defaults.py and mm_ctg.py for more
info.

BW> Are you running a news gateway?

No.

BW>That's the one handler that still does an os.fork() so if for some
BW>reason your child processes are returning to the main HandlerAPI loop,
BW>then that might explain both symptoms.
BW>
BW>Check your MTA log files.  Can you match the duplicate messages with
BW>deliveries by your MTA?  If Mailman were sending them out twice, you
BW>should see each entry in your MTA logs.

Yes, I have 2 different Message-Id's from sendmail 1 for each message
delivered.  I actually picked that up from the message headers (thanks
to "Roger B.A. Klorese" <rogerk@QueerNet.ORG> for that pointer).  I
think Mailman is sending the twice -- not Sendmail.

BW>
BW>I know I'm giving you more questions than answers, but I'll definitely
BW>need more information before we can figure out your problem.
BW>
BW>-Barry

Thank you!