[Mailman-Users] Qrunner lock

[Jon Carnes]

----- Original Message -----
From: "Lalo Castro" <laloc at cats.ucsc.edu>
To: <mailman-users at python.org>
Sent: Wednesday, September 26, 2001 7:02 PM
Subject: [Mailman-Users] Qrunner lock


> Hello,
>     I recently installed mailman  2.0.6 on our OpenBsd 2.9 mail machine,
> running python 1.5.2.  Our qrunner is taking an extraordinarily long
> time to get going.  450 second smtp cycles.
>     The qrunner log shows 'Could not acquire qrunner lock'.
>     The smtp error log shows:   smtp-failure:Sep 26 15:16:31 2001 (1628)
> 450 who at saaczpl.org (4.7.1 <who at saaczpl.org>... Can not check MX records
> for recipient host saaczpl.org)

Well "saaczpl.org" is non-existent.  So that is a valid error.
Plus that error happens almost 2 hours after the qrunner error.

>     And the general error log is:
> error:Sep 26 13:40:32 2001 qrunner(2995): Traceback (innermost last):
> error:Sep 26 13:40:32 2001 qrunner(2995):   File
> "/home/mailman/cron/qrunner", line 282, in ?

I believe that line is:
    mlist = open_list(listname)

So the error occurs when it tries to access the file with the email
addresses in it.  Some other process must have this file locked open.

> error:Sep 26 13:40:32 2001 qrunner(2995):      kids = main(lock)
> error:Sep 26 13:40:32 2001 qrunner(2995):   File
> "/home/mailman/cron/qrunner", line 195, in main
> error:Sep 26 13:40:32 2001 qrunner(2995):      lock.refresh()
> error:Sep 26 13:40:32 2001 qrunner(2995):   File
> "/home/mailman/Mailman/LockFile.py", line 204, in refresh
> error:Sep 26 13:40:32 2001 qrunner(2995):      raise NotLockedError
> error:Sep 26 13:40:32 2001 qrunner(2995): Mailman.LockFile .
> NotLockedError
>
>     Is there a conflict that mailmans qrunner could be having with
> another process?  It seems like from what we've read that qrunner is
> working (there is a lock file), but that there's a conflict somewhere
> with another process or message.  Has anyone else had this same problem?
>
I've never had this problem... well I had a similar problem with one of my
admins using an experimental browser that left processes hanging and a ton
of lock files in the lock directory.  I had to kill the left over processes
and then delete the lock files.  Fortunately, the lock file names identify
the processes.

No matter how much I whacked this guy, he wouldn't stop screwing up his
Mailman lists with his POS MS-beta browser so I wrote this script which
kills the dead processes and then removes the lock files.
===
#!/bin/bash
# Kill current Mailman processes and then remove lock files
# Jeff B is running a misconfigured browser and locks up the admin interface
# on a daily basis...  Jon C
for i in `ls /home/mailman/locks |cut -f7 -d. `; do kill $i; done
rm -f /home/mailman/locks/*
===

It shouldn't hurt to stop the processes and kill the lock files.

Jon Carnes