Opened 16 years ago

Closed 16 years ago

#175 closed defect (fixed)

Mailman mass unsubscriptions triggered?

Reported by: neteler Owned by: warmerdam
Priority: critical Milestone:
Component: SysAdmin Keywords: mailman
Cc:

Description

[here now as ticket]

I have observed that > 100 people have been unsubscribed from several GRASS lists running at lists.osgeo.org.

All precisely at Nov 29, 2007 3:00 PM (not sure which time zone, I guess UTC-1:

from	mailman-bounces lists.osgeo.org
to	grass-dev-owner lists.osgeo.org,
date	Nov 29, 2007 3:00 PM
subject	grass-dev unsubscribe notification
mailing list	grass-dev.lists.osgeo.org
mailed-by	lists.osgeo.org
	
name@somewhere.it has been removed from grass-dev.

NOTE: I have anonymized the email address. In this way estimated 100 addresses at the same time.

What's happening? Is this a list server cleanup? Do I now have to manually re-enable them (grrr)?

Original header of one message (I receive stuff at my gmail account):

Delivered-To: neteler.osgeo@gmail.com
Received: by 10.100.5.10 with SMTP id 10cs73186ane;
        Thu, 29 Nov 2007 06:14:11 -0800 (PST)
Received: by 10.142.107.1 with SMTP id f1mr1860223wfc.1196344807853;
        Thu, 29 Nov 2007 06:00:07 -0800 (PST)
Return-Path: <mailman-bounces@lists.osgeo.org>
Received: from lists.osgeo.org (lists.osgeo.org [66.223.95.242])
        by mx.google.com with ESMTP id m75si6940317wrm.2007.11.29.06.00.07;
        Thu, 29 Nov 2007 06:00:07 -0800 (PST)
Received-SPF: pass (google.com: best guess record for domain of mailman-bounces@lists.osgeo.org designates 66.223.95.242 as permitted sender) client-ip=66.223.95.242;
Authentication-Results: mx.google.com; spf=pass (google.com: best guess record for domain of mailman-bounces@lists.osgeo.org designates 66.223.95.242 as permitted sender) smtp.mail=mailman-bounces@lists.osgeo.org
Received: by lists.osgeo.org (Postfix)
	id B157FE025A3; Thu, 29 Nov 2007 09:00:05 -0500 (EST)
Delivered-To: neteler@osgeo.org
Received: from osgeo1.osgeo.org (localhost.localdomain [127.0.0.1])
	by lists.osgeo.org (Postfix) with ESMTP id A754EE025A2
	for <neteler@osgeo.org>; Thu, 29 Nov 2007 09:00:05 -0500 (EST)
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
Subject: grass-dev unsubscribe notification
From: mailman-bounces@lists.osgeo.org
To: grass-dev-owner@lists.osgeo.org
Message-ID: <mailman.41.1196344801.4567.grass-dev@lists.osgeo.org>
Date: Thu, 29 Nov 2007 09:00:01 -0500
Precedence: bulk
X-BeenThere: grass-dev@lists.osgeo.org
X-Mailman-Version: 2.1.5
List-Id: GRASS GIS Development mailing list <grass-dev.lists.osgeo.org>
X-List-Administrivia: yes
Sender: mailman-bounces@lists.osgeo.org
Errors-To: mailman-bounces@lists.osgeo.org

Markus

Change History (7)

comment:1 by neteler, 16 years ago

The first people start to complain about this mass unsubscription :( It seems I have to go through all of them manually and subscribe back. Or could the addresses be retrieved from the Mailman logs (for mass subscription including apologies)?

Markus

comment:2 by warmerdam, 16 years ago

Owner: changed from sac@… to warmerdam

Markus,

It seems they (211 items) were unsubscribed by the bounce processing system. I'm not clear on why, but presumably it relates to my having done something wrong during the transition.

I have harvested and segregated the addresses by list so you can bulk add them, and emailed those to you.

I can't guarantee that this won't happen again since I don't understand what is going on. But if you see the problem again, I'll reharvest the addresses.

It may also turn out to be necessary to disable bounce processing for these lists.

comment:3 by neteler, 16 years ago

Priority: blockercritical

Frank,

I have re-subscribed the members which I have received from you.

Suggestion:

http://veenet.value.net/~msapiro/scripts/ get_bounce_info.py

Prints the bounce information for all list members that have any.

could give some insights if many members have a (too) high bounce score [1]. Just to predict the next event (which hopefully doesn't happen). As you said in IRC, maybe the scores were badly transferred to the new list server (although it's both Mailman 2.1.5).

[1] http://www.gnu.org/software/mailman/mailman-admin/node25.html

The Bounce Processing Category

comment:4 by warmerdam, 16 years ago

I have installed the script. Running it I get reports like:

<bounce info for member wolfplusgrass@bergenheim.net
        current score: 1.0
        last bounce date: (2007, 10, 6)
        email notices left: 3
        last notice date: (1970, 1, 2)
        confirmation cookie: None
        >

Some have bounce scores that are 2.0. I'm a bit surprised that the scores (wolf's for instance) wouldn't have already been cleared to zero. Thanks for the additional tool.

comment:5 by neteler, 16 years ago

Here another hint:

http://mail.python.org/pipermail/mailman-users/2007-March/055903.html

>i would like to ask whether there is a
>possibility to reset the bounce status for a member (or all) to "nothing".
>
>i guess this can be done with "withlist", but before i start learning
>withlist again, i'd like to ask if somebody has a ready to use solution
>at hand and is willing to share.


There are several withlist script examples at
<http://veenet.value.net/~msapiro/scripts/> (mirror
<http://fog.ccsf.edu/~msapiro/scripts/>).

To remove bounce info from all members, you could modify the
fix_bad_bounce_info.py script by removing the line

        if not mlist.isMember(member):

To re-enable members you have been disabled by bounce but not yet
removed, use the reset_bounce.py script.

comment:6 by neteler, 16 years ago

The question is if we can just bulk-reset all bounce statuses of the grass-* lists or if that harms. I think that is doesn't harm (Mailman can just collect bounce scores from scratch then) and it should not touch disabled subscribtions since that's another flag.

comment:7 by warmerdam, 16 years ago

Resolution: fixed
Status: newclosed

Using the indicated technique I think I have stripped the bounce stats off all the existing grass lists now (and gdal-dev).

Note: See TracTickets for help on using tickets.