[Mailman-Users] Help with mailman instance in broken state

[Mark Sapiro]

On 5/15/18 10:27 AM, Kevin Bowen wrote:
> 
>> What's in the various mailman/qfiles/* queues.
> 
> I don't seem to have that directory? I'm running v2.1.9 on RHEL5, do you
> know where it would be located?


In RHEL, it's /var/spool/mailman/ (see <https://wiki.list.org/x/8486953>).



>> What's in Mailman's locks directory? See <https://wiki.list.org/x/17891756
>> .
> 
> Ok, now this is interesting: there are 6 files in /var/lock/mailman named
> <listname>.lock.announce.ucsd.edu.<digits>.0 They have timestamps between 3
> and 4 in the morning, which is probably around the time the issue cleared
> up. Looking in /var/log/mailman/locks I see, starting right around the time
> the problem started yesterday, a bunch of stack traces with
> "<listname>.lock lifetime has expired, breaking" and "<listname>.lock
> unexpected linkcount: -1"


So the issue was a locked list. See <https://wiki.list.org/x/17891756>
for more on locks, but those file names are  if the form
<listname>.lock.<hostname>.<pid>.<seq> so what yo call 'digits' is the
pid of the process waiting on the lock. If there is no file named just
<listname>.lock, the list isn't locked and all the other
<listname>.lock.* files are orphans and their pids are probably gone, so
just remove them.

Actually, when these files are created, they are created with a time
stamp in the future which represents the expiration of the lock lifetime
so that probably is when the list was no longer locked and things
cleared up.

As far as what caused it, that's hard to say, but one strong possibility
is there was an interaction with the list's admin or admindb UI that was
timed out by the web server and left the list locked.


> Also, I notice that in /var/lib/mailman/data, there are currently 4 files
> called heldmsg-<listname>-<digits>.pck, although the admin interface says
> there are currently no messages being held for moderation. Does that mean
> anything?


Those are orphaned held messages, probably orphaned when you deleted and
recreated the list. You can look at them with Mailman's bin/dumpdb and
if you want them, you can reprocess them with the script at
<https://www.msapiro.net/scripts/hold_again>


-- 
Mark Sapiro <mark at msapiro.net>        The highway is for gamblers,
San Francisco Bay Area, California    better use your sense - B. Dylan