Hi,
A few updates on the general issues related to bouncing:
1. We worked with Mailman developers to improve the bounce parsing so it's less likely miss-process a bounce message. This might help in a few cases where list owners were getting unsubscribed. This was deployed a few hours ago. https://phabricator.wikimedia.org/T285120
2. We're looking into enabling "verp probes". If an address hits the bounce score, Mailman will send them a dedicated email, and if *that* mail bounces, then they get unsubscribed. This will prevent getting unsubscribed when spam/viruses go to the -owner address and are bounced. https://phabricator.wikimedia.org/T285361
3. We're discussing with upstream about just disabling bounce processing of the -owner address altogether. https://gitlab.com/mailman/mailman/-/issues/912
If you're still running into bouncing issues or other issues please let us know and we'll look into them.
Thanks, -- Kunal
Hi,
On 6/23/21 2:26 AM, Kunal Mehta wrote:
- We're looking into enabling "verp probes". If an address hits the
bounce score, Mailman will send them a dedicated email, and if *that* mail bounces, then they get unsubscribed. This will prevent getting unsubscribed when spam/viruses go to the -owner address and are bounced. https://phabricator.wikimedia.org/T285361
The "verp probes" feature is now enabled.
Additionally, all messages with a spam score of >=6.0 are now globally discarded. This was previously enabled in Mailman2 but got missed in the migration, thanks to MarcoAurelio for noticing.
At this point if you're still having issues with bounce processing and people getting unsubscribed, please let us know. I'd like to think we've closed most of the holes but could've easily missed something.
-- Kunal
On 7/6/21 3:04 PM, Platonides wrote:
I guess those are spamassassin points? or is it using a different system?
Yep, SpamAssassin. There's some slightly outdated technical documentation at https://wikitech.wikimedia.org/wiki/Mailman/Spam_fighting.
-- Kunal
listadmins@lists.wikimedia.org