[Mailman-Users] Archiving errors & lock files

[Tim Pepper]

I have a Mailman install (2.0beta2) running on a Debian linux machine with
python 1.5.2 (debian's 1.5.2-10 python-base package) which is giving me untold
grief right now.  Messages coming into lists are getting sent on to the list
properly as best I can tell.  Quite often though, the user sending a message
will also recieve an error message back from the server, similar to the
following (this one pulled from the $prefix/logs/error file):

	Aug 28 09:09:39 2000 post(3233): Traceback (innermost last):
	post(3233):   File "/www/sites/lists/mailman/scripts/post", line 86, in ?
	post(3233):      main()
	post(3233):   File "/www/sites/lists/mailman/scripts/post", line 62, in main
	post(3233):      mlist.Post(msg)
	post(3233):   File "/www/sites/lists/mailman/Mailman/MailList.py", line 1293, in Post
	post(3233):      Mailman.Handlers.HandlerAPI.DeliverToList(self, msg)
	post(3233):   File "/www/sites/lists/mailman/Mailman/Handlers/HandlerAPI.py", line 61, in DeliverToList
	post(3233):      pipeline_delivery(mlist, msg, pipeline)
	post(3233):   File "/www/sites/lists/mailman/Mailman/Handlers/HandlerAPI.py", line 37, in pipeline_delivery
	post(3233):      func(mlist, msg)
	post(3233):   File "/www/sites/lists/mailman/Mailman/Handlers/ToArchive.py", line 32, in process
	post(3233):      mlist.ArchiveMail(msg)
	post(3233):   File "/www/sites/lists/mailman/Mailman/Archiver/Archiver.py", line 200, in ArchiveMail
	post(3233):      lock.lock()
	post(3233):   File "/www/sites/lists/mailman/Mailman/LockFile.py", line 286, in lock
	post(3233):      raise StaleLockFileError(
	post(3233): Mailman.LockFile . StaleLockFileError :  Stale lock file found linked to file: /www/sites/lists/mailman/locks/devel.archiver.lock (requires manual intervention)

The $prefix/locks directory usually then upon inspection has some entries like:

	-rw-rw-r--    1 nobody   mailman        86 Aug 25 14:41 devel.archiver.lock
	-rw-rw-r--   16 mailman  mailman        86 Aug 24 08:07 devel.archiver.lock.lists.10521
	-rw-rw-r--   16 mailman  mailman        86 Aug 24 08:07 devel.archiver.lock.lists.11821
	-rw-rw-r--   16 mailman  mailman        86 Aug 24 08:07 devel.archiver.lock.lists.12474

If I delete all of these and then 'touch devel.archiver.lock' things seem to
work for a while again.  I don't know if the lock file needs to exist, but it
seemed like functioning lists had them even when the ...lock.lists.##### files
weren't there, so I thought I'd recreate the file after killing it.  The
existing ones usually appear to contain a textual reference to the
...lock.lists.##### type files.

I also came across files in the locks directory such as:

	devel<.lock
	devel=20.lock
	docs<br.lock
	docs=20.lock

which almost look like they've got bits of html or something injected in the
name (which is bit disturbing to me).

This is a very low traffic machine with very capable hardware (dual pentiums
with a bunch of memory server upwards of a dosen messages a day).  If it
matters the mailer is exim and the web server is apache.


So...a couple questions:

Anybody know what could be wrong here?

How do these lock files work?  Am I messing stuff up simply deleting them?  I
read through some mailman-devel archives and found some discussion of
lockfiles which don't contain any text (races in writing to them) but use
ctime/mtime offsets.  The sample of python in that discussion was definitely
not what is in my copy of LockFile.py, so either I've got an older version or
else those changes weren't commited/accepted or something?

And...Is there any solid documentation on mailman for the sysadmin?

Tim

--
*********************************************************
*  tim.pepper at opensales dot com     * Venimus, Vidimus, *
*  http://www.opensales.com         * Dolavimus         *
*********************************************************