[Mailman-Users] Follow-up - Duplicate Messages Sent - Follow-up

[Dan Collins]

Hi Y'all --

First, thanks so much to Mark Sapiro for answering a number of questions.

Why the duplicate messages?  I'm still not really sure.  But I had debug on 
in a couple of sendmail config files - etc/resolv.conf and etc/qpopper.conf. 
Python didn't like that I guess and as Mark suggested, Mailman's 
qfiles/retry queued the messages for retry. I removed debug references and 
restarted sendmail. Then deinstalled and reinstalled Mailman.  It "appears" 
to work fine now;  I'll sleep a lot better if it still looks fine by happy 
hour tomorrow.

Another unrelated thing that drove me crazy was that every time I added a 
list from the command line, or made any change at all from the command line, 
I had to re-run bin/check_perms and chown mailman:mailman mailman.  That's 
because on my set-up (freeeBSD), the user mailman has no log-in (as per 2.1 
in the Installation manual), and so I set the whole thing up as the root 
user.

When re-installing, remove any lists first, run make deinstall from 
ports/mail/mailman (freeBSD) (and NOT rm mailman dir) THEN run make and make 
install. No shortcuts and follow the installation manual. e-x-a-c-t-l-y.

Dan

"You can lead a horse to water, but if you get it to lie on its back and 
float, THEN you got something."

----- Original Message ----- 
From: "Mark Sapiro" <msapiro at value.net>
To: "Dan Collins" <bigtuner at comcast.net>; <mailman-users at python.org>
Sent: Wednesday, July 27, 2005 7:32 PM
Subject: Re: [Mailman-Users] Duplicate Messages Sent


> Dan Collins wrote:
>>
>>The second message I sent to the mailman list had attachments for testing 
>>scrub; over 50 mb no doubt. That subsequent e-mail has never arrived, but 
>>a Mailman message informing me that there is an administrative task 
>>arrived.  I went to the web admin and discarded the message, but the 
>>administrative warning message continues to arrive every 5 minutes, just 
>>like freakin clockwork. Could there be two identical cron entries causing 
>>this or what?
>
>
> The only cron that should run every 5 minutes is gate_news. Do you have
> anything set up in Mail<->News gateways? The cron that sends the "nn
> LISTNAME moderator request(s) waiting" message is checkdbs which
> normally runs once daily at 8:00 a.m.
>>
>>In usr/local/mailman/locks, mailmanctl starts master-qrunner AND 
>>master-qrunner.bigtxxxx dot com dot 17567 (or some other number) Is that 
>>correct? In usr/local/mailman/data, I have one master-qrunner.pid.
>
> This is correct. One file data/master-qrunner.pid containing the pid
> and two files locks/master-qrunner and
> locks/master-qrunner.your.host.name.pid both containing the same thing
> as the name of the second file with a full path.
>
>
>>Logs --
>>
>>Mailman logs have 137KB of locks entries of the type
>>
>>Jul 27 20:31:00 2005 (9701) mailman.lock unlocked
>
>
> This refers to a lock for the 'mailman' list
>
>>mailman/logs post read
>>
>>Jul 27 18:54:14 2005 (17583) post to mailman from 
>>mailman-owner at uniconexed.org, size=1783, 
>>message-id=<mailman.0.1122485981.24170.mailman at uniconexed.org>, 1 failures
>
>
> This looks like a mailman generated message. Why is it posted to the
> 'mailman' list? Is the'mailman' list an owner or moderator of another
> list or itself? That could probably cause loops.
>
>
>>qrunner logs say
>>
>>Jul 27 20:26:55 2005 (17567) Master watcher caught SIGTERM.  Exiting.
>>Jul 27 20:26:55 2005 (17567) Master qrunner detected subprocess exit (pid: 
>>17582, sig: None, sts: 15, class: NewsRunner, slice: 1/1)
>>Jul 27 20:26:55 2005 (17567) Master qrunner detected subprocess exit (pid: 
>>17581, sig: None, sts: 15, class: IncomingRunner, slice: 1/1)
>>Jul 27 20:26:55 2005 (17567) Master qrunner detected subprocess exit (pid: 
>>17579, sig: None, sts: 15, class: BounceRunner, slice: 1/1)
>>Jul 27 20:26:55 2005 (17567) Master qrunner detected subprocess exit (pid: 
>>17578, sig: None, sts: 15, class: ArchRunner, slice: 1/1)
>>Jul 27 20:26:55 2005 (17567) Master qrunner detected subprocess exit (pid: 
>>17580, sig: None, sts: 15, class: CommandRunner, slice: 1/1)
>>Jul 27 20:26:55 2005 (17567) Master qrunner detected subprocess exit (pid: 
>>17584, sig: None, sts: 15, class: VirginRunner, slice: 1/1)
>>Jul 27 20:26:55 2005 (17567) Master qrunner detected subprocess exit (pid: 
>>17583, sig: None, sts: 15, class: OutgoingRunner, slice: 1/1)
>>Jul 27 20:26:55 2005 (17585) RetryRunner qrunner exiting.
>>Jul 27 20:26:55 2005 (17567) Master qrunner detected subprocess exit (pid: 
>>17585, sig: None, sts: 15, class: RetryRunner, slice: 1/1)
>
>
> Looks like a 'mailmanctl stop' or maybe a kill of the master qrunner.
>
>
>>partial mailman/logs smtp say
>>
>>Jul 27 20:10:10 2005 (17583) 
>><mailman.0.1122488300.32572.mailman at uniconexed.org> smtp to mailman for 1 
>>recips, completed in 3.799 seconds
>>Jul 27 20:25:06 2005 (17583) 
>><mailman.0.1122485981.24170.mailman at uniconexed.org> smtp to mailman for 1 
>>recips, completed in 1.167 seconds
>>Jul 27 20:25:07 2005 (17583) <000a01c592d4$e25b3f40$6401a8c0 at Athlon> smtp 
>>to mailman for 2 recips, completed in 0.408 seconds
>>Jul 27 20:25:24 2005 (17583) 
>><mailman.0.1122488300.32572.mailman at uniconexed.org> smtp to mailman for 1 
>>recips, completed in 6.489 seconds
>>
>>partial smtp-failure logs say
>>
>>Jul 27 20:25:06 2005 (17583) SMTP session failure: -1, es_setoptions(" 
>>debug, msgid: <000a01c592d4$e25b3f40$6401a8c0 at Athlon>
>>Jul 27 20:25:06 2005 (17583) SMTP session failure: -1, es_setoptions(" 
>>debug, msgid: <000a01c592d4$e25b3f40$6401a8c0 at Athlon>
>>
>>vette logs say (regarding the subsequent message, no doubt)
>>
>>Jul 27 18:18:22 2005 (32572) Mailman post from bigtuner at comcast.net held, 
>>message-id=<001601c592d5$b64b3a80$6401a8c0 at Athlon>: Message body is too 
>>big: 12472312 bytes with a limit of 50 KB
>>Jul 27 18:46:20 2005 (7448) mailman: Discarded posting:
>>    From: bigtuner at comcast.net
>>    Subject: test wtih attachments
>>    Reason: Your message was too big; please trim it to less than 50 KB in 
>> size.
>>
>>Finally, in my var/mail logs, I get entries like
>>
>>Jul 27 20:25:22 bigtuner sm-mta[3485]: j6RKPHh3003485: --- 221 2.0.0 
>>bigtuner.com closing connection
>>Jul 27 20:25:22 bigtuner sm-mta[3543]: j6RKPHh1003485: --- 050 
>><webmaster at uniconexed.org>... Connecting to gateway-r.comcast.net. via 
>>esmtp...
>>Jul 27 20:25:22 bigtuner sm-mta[3543]: j6RKPHh1003485: SMTP outgoing 
>>connect on bigtuner.com
>>Jul 27 20:25:54 bigtuner sm-mta[3543]: j6RKPHh1003485: --- 050 
>><webmaster at uniconexed.org>... Sent (ok ; id=20050727202522r2200og6ste)
>>Jul 27 20:25:54 bigtuner sm-mta[3543]: j6RKPHh1003485: 
>>to=<webmaster at uniconexed.org>, delay=00:00:37, xdelay=00:00:32, 
>>mailer=esmtp, pri=12615209, relay=gateway-r.comcast.net.[216.148.227.126], 
>>dsn=2.0.0, stat=Sent (ok ; id=20050727202522r2200og6ste)
>>Jul 27 20:25:54 bigtuner sm-mta[3543]: j6RKPHh1003485: done; 
>>delay=00:00:37, ntries=1
>>Jul 27 20:25:55 bigtuner sm-mta[3543]: NOQUEUE: --- 050 Closing connection 
>>to gateway-r.comcast.net.
>>
>>In usr/local/mailman/data, I have one master-qrunner.pid. In 
>>usr/local/mailman/locks, mailmanctl starts master-qrunner AND 
>>master-qrunner.bigtxxxx dot com dot 17567 (or some other number) Is that 
>>correct?
>
>
> Yes, see above.
>
>
>>Could it be that the locks aren't staying alive long enough to process a 
>>message?
>
> Probably not.
>
>
>>Do I need to restart sendmail whenever I restart Mailman?
>
>
> Not if you didn't make changes to sendmail itself.
>
>
>>What's with the NOQUEUE mail log entry?
>
> Don't know.
>
>
>>Is this a sendmail problem and not Mailman at all?
>
> Maybe. It looks like at least some message deliveries are failing per
> the entries in smtp-failure. The entry
>
> SMTP session failure: -1, es_setoptions(" debug, msgid:
> <000a01c592d4$e25b3f40$6401a8c0 at Athlon>
>
> means the Python smtplib returned an SMTPResponseException where -1 is
> the error code and 'es_setoptions(" debug' is the error message.
>
> Are the messages ending up in Mailman's qfiles/retry queue and being
> resent from there?
>
> --
> Mark Sapiro <msapiro at value.net>       The highway is for gamblers,
> San Francisco Bay Area, California    better use your sense - B. Dylan
>