[Mailman-Users] Large list problem

[Mark Sapiro]

Chance Eppinette wrote:
>
>I am running Mailman 2.1.5 on a Solaris 2.8 SUN 280R (fully loaded).
>
>I have a number of lists but 1 in particular is giving me problems.  The
>list currently has about 14000 members of which about 1100 are disabled due
>to bounces.
>The problem seems to arise when a message is sent to the list (around every
>2-3 weeks).
>
>The CPU time for a python process jumps up to around 50% for an indetermined
>amount of time (> 24 hours).


Which python process (i.e. presumably one of the qrunners, but which
one)


>While the system is "working" on this list message, GUI access to the list
>is nearly impossible (can't login); while other lists seem to function OK.


The list is probably locked.


>Up to this point I have had to stop/kill all Mailman processes and then
>restart Mailman several times to bring things back to normal with the List
>service.
>The intended message being sent to the list does arrive in users mailboxes
>(which are worldwide).
>
>I believe the culprit to be a combination of managing the number of bounces
>due to the size of the list membership.


This doesn't seem likely.


>Also, we prefer that once a member has been disabled for 3 bounces, that it
>also be unsubscribed from the list.  This does seem to occur, but it seems
>to take forever to occur.  Like I said there are about 1100 addresses
>disabled right now.  However, they have yet to be removed.  It is like the
>system is waiting on something else to occur, but I don't know what.


It's waiting n weeks for the n weekly warning messages to be sent where
n is the bounce_you_are_disabled_warnings setting. It seems you want
this to be 0.

You might be interested in seeing what's in Mailman's bounce log.

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