Unsubscriptions after DMARC fix
Last month a Mailman (v2.1.18-1) list on my server got hit with a bunch of bounces based on DMARC rejections from Gmail, Yahoo and Hotmail which honor "p=reject" in a DMARC record. The list had not previously been configured to handle DMARC issues so at that time I set dmarc_moderation_action to "Munge From" which I assumed would solve the problem going forward.
Bounce processing for the list is set to send 10 probe messages at 3 day intervals, and last night all of these addresses were unsubscribed. It seems that probe messages to these addresses, which I believe come from the list server's mailman address, were also bouncing. I would assume that messages which come directly from our server and are not reflected through a redirection or mailing list wouldn't be subject to rejection based on DMARC policy.
Why would this be? What can be done to address the problem?
--
Lindsay Haisley | "The first casualty when
FMP Computer Services | war comes is truth."
512-259-1190 |
http://www.fmp.com | -- Hiram W Johnson
On Mon, 2017-04-03 at 12:28 -0500, Lindsay Haisley wrote:
Add comcast.net, msn.com and googlemail.com to the above list of rejecting ESPs/address domains. I assume googlemail.com is an alias for gmail.com.
--
Lindsay Haisley | "The first casualty when
FMP Computer Services | war comes is truth."
512-259-1190 |
http://www.fmp.com | -- Hiram W Johnson
On 04/03/2017 10:28 AM, Lindsay Haisley wrote:
They aren't "probe" messages in the sense of Mailman's
VERP_PROBES = Yes
because VERP_PROBES don't bounce for DMARC (see item 5 at <https://wiki.list.org/x/17891458>, and they are only involved in initially disabling delivery by bounce.
The issue is the user's delivery was disabled by bounce a month ago. At that point further bounces don't matter. The process is controlled by cron/disabled which sends a total of bounce_you_are_disabled_warnings warning messages to the users at intervals of bounce_you_are_disabled_warnings_interval.
If the user doesn't act on one of those warnings and re-enable delivery before the number is exhausted, the user will be unsubscribed.
So basically, the users that were unsubscribed ignored the fact that they weren't receiving list mail and ignored 10 messages informing them their delivery was disabled and telling them how to re-enable it. Maybe the list is better off without them?
-- Mark Sapiro <mark@msapiro.net> The highway is for gamblers, San Francisco Bay Area, California better use your sense - B. Dylan
On Mon, 2017-04-03 at 12:40 -0700, Mark Sapiro wrote:
OK, so I'm in the clear to resubscribe these folks.
Maybe the list is better off without them?
Actually, we try to be gentle with these folks and keep things as simple as possible. Some of them are pretty non-techie and these messages probably elicit a hormonal technophobia reaction. I'll just resubscribe them.
What chaps my dookie is subscribers on AOL and other similar services who don't bother to even _try_ to unsubscribe if they want off a list and just hit "Report Spam" and I get an AOL TOS message from which I have to extract the subscriber address. Not a great problem since I have the subscriber address AES-encrypted into the Resent-message-ID header, but still a PITA.
--
Lindsay Haisley | "The first casualty when
FMP Computer Services | war comes is truth."
512-259-1190 |
http://www.fmp.com | -- Hiram W Johnson
On Mon, 2017-04-03 at 15:03 -0500, Lindsay Haisley wrote:
My list owner asks:
Hormonally-induced techno-alergic reaction!!
Good question! So we want to simply zero out all the bounce counts, and possibly the notification counts for the whole list on a one-time basis. Will turning off bounce_processing for a day or so do this?
I've encouraged Cindy Harris, the list admin, to join this list. She's very tech savvy and although she doesn't admin Mailman on a server, she asks good questions and deals with list management and tech support. I've told her that Mark, Brad, Barry et. al don't bite or act like tech-snobs! Cindy is an accomplished musician and helps manage a small ISP in Pennsylvania.
--
Lindsay Haisley | "The first casualty when
FMP Computer Services | war comes is truth."
512-259-1190 |
http://www.fmp.com | -- Hiram W Johnson
On 04/03/2017 01:21 PM, Lindsay Haisley wrote:
You could turn off bounce processing for longer than bounce_info_stale_after to try to ensure there are no "current" bounces after turning it back on, but there is now a script at <https://www.msapiro.net/scripts/clear_bounce_info.py> (mirrored at <https://fog.ccsf.edu/~msapiro/scripts/clear_bounce_info.py>) which will do it.
We'd be happy to have her join.
-- Mark Sapiro <mark@msapiro.net> The highway is for gamblers, San Francisco Bay Area, California better use your sense - B. Dylan
On Apr 05, 2017, at 11:36 AM, Stephen J. Turnbull wrote:
Barry encourages participation by musicians!
Not official Barry-spokesman, but reasonably good at channeling :-)
Don't forget the secret bassplayer.py handler! You email it a drum pattern and it replybots you a funky groove in E.
-B
On Mon, 2017-04-03 at 12:28 -0500, Lindsay Haisley wrote:
Add comcast.net, msn.com and googlemail.com to the above list of rejecting ESPs/address domains. I assume googlemail.com is an alias for gmail.com.
--
Lindsay Haisley | "The first casualty when
FMP Computer Services | war comes is truth."
512-259-1190 |
http://www.fmp.com | -- Hiram W Johnson
On 04/03/2017 10:28 AM, Lindsay Haisley wrote:
They aren't "probe" messages in the sense of Mailman's
VERP_PROBES = Yes
because VERP_PROBES don't bounce for DMARC (see item 5 at <https://wiki.list.org/x/17891458>, and they are only involved in initially disabling delivery by bounce.
The issue is the user's delivery was disabled by bounce a month ago. At that point further bounces don't matter. The process is controlled by cron/disabled which sends a total of bounce_you_are_disabled_warnings warning messages to the users at intervals of bounce_you_are_disabled_warnings_interval.
If the user doesn't act on one of those warnings and re-enable delivery before the number is exhausted, the user will be unsubscribed.
So basically, the users that were unsubscribed ignored the fact that they weren't receiving list mail and ignored 10 messages informing them their delivery was disabled and telling them how to re-enable it. Maybe the list is better off without them?
-- Mark Sapiro <mark@msapiro.net> The highway is for gamblers, San Francisco Bay Area, California better use your sense - B. Dylan
On Mon, 2017-04-03 at 12:40 -0700, Mark Sapiro wrote:
OK, so I'm in the clear to resubscribe these folks.
Maybe the list is better off without them?
Actually, we try to be gentle with these folks and keep things as simple as possible. Some of them are pretty non-techie and these messages probably elicit a hormonal technophobia reaction. I'll just resubscribe them.
What chaps my dookie is subscribers on AOL and other similar services who don't bother to even _try_ to unsubscribe if they want off a list and just hit "Report Spam" and I get an AOL TOS message from which I have to extract the subscriber address. Not a great problem since I have the subscriber address AES-encrypted into the Resent-message-ID header, but still a PITA.
--
Lindsay Haisley | "The first casualty when
FMP Computer Services | war comes is truth."
512-259-1190 |
http://www.fmp.com | -- Hiram W Johnson
On Mon, 2017-04-03 at 15:03 -0500, Lindsay Haisley wrote:
My list owner asks:
Hormonally-induced techno-alergic reaction!!
Good question! So we want to simply zero out all the bounce counts, and possibly the notification counts for the whole list on a one-time basis. Will turning off bounce_processing for a day or so do this?
I've encouraged Cindy Harris, the list admin, to join this list. She's very tech savvy and although she doesn't admin Mailman on a server, she asks good questions and deals with list management and tech support. I've told her that Mark, Brad, Barry et. al don't bite or act like tech-snobs! Cindy is an accomplished musician and helps manage a small ISP in Pennsylvania.
--
Lindsay Haisley | "The first casualty when
FMP Computer Services | war comes is truth."
512-259-1190 |
http://www.fmp.com | -- Hiram W Johnson
On 04/03/2017 01:21 PM, Lindsay Haisley wrote:
You could turn off bounce processing for longer than bounce_info_stale_after to try to ensure there are no "current" bounces after turning it back on, but there is now a script at <https://www.msapiro.net/scripts/clear_bounce_info.py> (mirrored at <https://fog.ccsf.edu/~msapiro/scripts/clear_bounce_info.py>) which will do it.
We'd be happy to have her join.
-- Mark Sapiro <mark@msapiro.net> The highway is for gamblers, San Francisco Bay Area, California better use your sense - B. Dylan
On Apr 05, 2017, at 11:36 AM, Stephen J. Turnbull wrote:
Barry encourages participation by musicians!
Not official Barry-spokesman, but reasonably good at channeling :-)
Don't forget the secret bassplayer.py handler! You email it a drum pattern and it replybots you a funky groove in E.
-B
participants (4)
-
Barry Warsaw -
Lindsay Haisley -
Mark Sapiro -
Stephen J. Turnbull