when on particular person sends a message to the list dozens of OTHER users have their bounce score incremented
We have been experiencing a strange anomaly. We are running a small-ish list. When a particular subscriber to the list sends an email to the list, a couple of dozen (or more) of the recipient's of the message end up with their bounce scores incrementing. This particular subscriber uses about three different email addresses when he sends his messages and every one of them causes this behavior. There is another subscriber who also causes this behavior when he sends an email to the list, but only one of the two email addresses he sends from causes this behavior. There is a third person who causes this behavior every time. There are dozens of other subscribes who don't generate this behavior when they send to the list.
I can't figure out why this is occurring. I *assume* that Mailman re-writes the headers to add DKIM, etc., before sending them to the list recipients but it sure seems like something in the header from the original sender is ending up in the final header (or something is LACKING) which is not liked by the recipient's MTA's. It doesn't seem like a subscriber sending an email to the list should be able to cause this behavior - but I'm no expert.
The recipients appear to be using many different mail platforms (ie. their addresses are not all, or even mostly, @google.com or some other domain; the recipients appear to be a pretty good cross section of platforms).
This has been going on all year (maybe longer). Any help or troubleshooting suggestions would be greatly appreciated.
On Sat, Aug 24, 2024 at 6:32 AM Ken Alker <mailman3.org@alker.net> wrote:
We have been experiencing a strange anomaly. We are running a small-ish list. When a particular subscriber to the list sends an email to the list, a couple of dozen (or more) of the recipient's of the message end up with their bounce scores incrementing. This particular subscriber uses about three different email addresses when he sends his messages and every one of them causes this behavior. There is another subscriber who also causes this behavior when he sends an email to the list, but only one of the two email addresses he sends from causes this behavior. There is a third person who causes this behavior every time. There are dozens of other subscribes who don't generate this behavior when they send to the list.
I can't figure out why this is occurring. I *assume* that Mailman re-writes the headers to add DKIM, etc., before sending them to the list recipients but it sure seems like something in the header from the original sender is ending up in the final header (or something is LACKING) which is not liked by the recipient's MTA's. It doesn't seem like a subscriber sending an email to the list should be able to cause this behavior - but I'm no expert.
The recipients appear to be using many different mail platforms (ie. their addresses are not all, or even mostly, @google.com or some other domain; the recipients appear to be a pretty good cross section of platforms).
This has been going on all year (maybe longer). Any help or troubleshooting suggestions would be greatly appreciated.
When you say "I *assume* that Mailman re-writes the headers to add DKIM, etc., before sending them to the list recipients", I think there lies the issue. It's either you have configured MM3 to do the DKIM bit or it's done by your MTA. Only one is enough. But even DKIM is not the issue, but DMARC. So my lists are configured this: https://imgur.com/a/Wk82b7Z And I think with proper DKIM+SPF for your MM3s mail host, there should be no such phenomenon you're observing.
-- Best regards, Odhiambo WASHINGTON, Nairobi,KE +254 7 3200 0004/+254 7 2274 3223 In an Internet failure case, the #1 suspect is a constant: DNS. "Oh, the cruft.", egrep -v '^$|^.*#' ¯\_(ツ)_/¯ :-) [How to ask smart questions: http://www.catb.org/~esr/faqs/smart-questions.html]
if this particular subscriber uses apple mail on iphone, and the mails in questions are _answers_ to existing mails, it might be the known issue that the References: header grows beyond a limit given by exim. This is the fault of apple mail and violates standards, historically also microsoft didn't care about that limit. I haven't found a workaround yet; postfix mailer allows to split this line. Best!
Am 24. August 2024 um 10:53 schrieb "Odhiambo Washington via Mailman-users" <mailman-users@mailman3.org>:
On Sat, Aug 24, 2024 at 6:32 AM Ken Alker <mailman3.org@alker.net> wrote:
We have been experiencing a strange anomaly. We are running a small-ish list. When a particular subscriber to the list sends an email to the list, a couple of dozen (or more) of the recipient's of the message end up with their bounce scores incrementing. This particular subscriber uses about three different email addresses when he sends his messages and every one of them causes this behavior. There is another subscriber who also causes this behavior when he sends an email to the list, but only one of the two email addresses he sends from causes this behavior. There is a third person who causes this behavior every time. There are dozens of other subscribes who don't generate this behavior when they send to the list.
I can't figure out why this is occurring. I *assume* that Mailman re-writes the headers to add DKIM, etc., before sending them to the list recipients but it sure seems like something in the header from the original sender is ending up in the final header (or something is LACKING) which is not liked by the recipient's MTA's. It doesn't seem like a subscriber sending an email to the list should be able to cause this behavior - but I'm no expert.
The recipients appear to be using many different mail platforms (ie. their addresses are not all, or even mostly, @google.com or some other domain; the recipients appear to be a pretty good cross section of platforms).
This has been going on all year (maybe longer). Any help or troubleshooting suggestions would be greatly appreciated.
(...)
Schulz via Mailman-users wrote on 2024-08-24 12:39:
if this particular subscriber uses apple mail on iphone, and the mails in questions are_answers_ to existing mails, it might be the known issue that the References: header grows beyond a limit given by exim. This is the fault of apple mail and violates standards, historically also microsoft didn't care about that limit. I haven't found a workaround yet; postfix mailer allows to split this line.
This sounds very plausible.
Another thought, less likely but worth looking into, would be in Postfix's main.cf:
mailman_destination_recipient_limit = 1 ## default_destination_recipient_limit = 1
As I understand it, Postfix batches outgoing messages to reduce overhead, so perhaps one member in a batch is failing and Mailman is applying the bounce to everyone in that batch?
Forcing Postfix to send each message individually might help and is simple to implement.
rb
I probably should have said DMARC and not DKIM (I was spam combatant policies all together as if they were all the same/similar, but I should not have done that; too many "new" acronyms to keep straight).
Thank you for providing the screen-shot showing List Settings/Settings/DMARC Mitigations with "DMARC mitigation action" set to "Replace From: with list address" and "DMARC Mitigate unconditionally" set to "Yes".
I tried this and it cleared up the problem (at least with one sender tested so far). I also noticed the my colleague was already using these settings in a list he manages on the same server.
Unfortunately, now when I respond to list emails (with the re-written From address) my (very old) email client (I still use Mulberry!) defaults to responding to the sender instead of the list, as it had before. I'm hoping this is a nuance of my client and isn't something the other members are going to have to grapple with.
On 8/30/24 11:00, Ken Alker wrote:
Unfortunately, now when I respond to list emails (with the re-written From address) my (very old) email client (I still use Mulberry!) defaults to responding to the sender instead of the list, as it had before. I'm hoping this is a nuance of my client and isn't something the other members are going to have to grapple with.
See https://gitlab.com/mailman/mailman/-/blob/master/src/mailman/handlers/dmarc....
-- Mark Sapiro <mark@msapiro.net> The highway is for gamblers, San Francisco Bay Area, California better use your sense - B. Dylan
On 8/23/24 8:31 PM, Ken Alker wrote:
We have been experiencing a strange anomaly. We are running a small-ish list. When a particular subscriber to the list sends an email to the list, a couple of dozen (or more) of the recipient's of the message end up with their bounce scores incrementing.
The notices of bounce score incrementing, assuming Notify owner on bounce increment is Yes, should contain the bounce DSN. What does it say the reason is?
-- Mark Sapiro <mark@msapiro.net> The highway is for gamblers, San Francisco Bay Area, California better use your sense - B. Dylan
See my results posted as a response to Odhiambo Washington above. However for thread completion (and to help others in the future), the answer to your question is:
The DSNs vary based on which provider responded (ie. verizon.net, sbcglocal.net, navy.mil, me.com, hotmail.com, cox.net, etc.), but one example (that very much supports above conclusion) is:
Diagnostic-Code: smtp;550 5.7.509 Access denied, sending domain r1de.net does not pass DMARC verification and has a DMARC policy of reject.
I just found and then changed the "Settings/Alter Messages/Replay goes to list" from "No Munging" to "Reply goes to list" to combat the reply issue I mentioned above. I suppose this means that if someone sent a specific Reply-to: in their original email, it gets overwritten. I suppose there is no perfect solution.
On 8/30/24 14:34, Ken Alker wrote:
I just found and then changed the "Settings/Alter Messages/Replay goes to list" from "No Munging" to "Reply goes to list" to combat the reply issue I mentioned above. I suppose this means that if someone sent a specific Reply-to: in their original email, it gets overwritten. I suppose there is no perfect solution.
If Reply goes to list is set to Reply goes to list, the list posting
address is added to the Reply-To: header. I.e., if there is a Reply-To:
in the incoming message, The resultant Reply-To: contains the original
plus the list posting address unless First strip Reply-To is Yes in
which case it contains only the list posting address.
-- Mark Sapiro <mark@msapiro.net> The highway is for gamblers, San Francisco Bay Area, California better use your sense - B. Dylan
participants (5)
-
Ken Alker -
Mark Sapiro -
Odhiambo Washington -
Ron / BCLUG -
Schulz