[UCIMC-Tech] Mailman mass unsubscribe

Mon Apr 23 13:01:57 UTC 2012

Unfortunately it's by IP address, so just renaming the server wouldn't help, it would need to be renumbered by our upstream provider. Fortunately, there is an easy workaround which I've already put in place; mail is currently relaying through pryde.chambana.net, our other offsite server and current primary DNS server. This seems to have all but eliminated the problem as far as I can tell. The biggest issue isn't with institutional blacklists (though those are also a problem), but with a Cisco email appliance called Senderbase which heuristically blocks email for all Senderbase customers. The IP allocation is blacklisted from way back, and nothing fixes it (although I've emailed Cisco) but time. So hopefully we can eventually switch back to sending directly, rather than through a relay (although primary email will hopefully have moved back to new systems at the IMC by then anyway), but it seems to be working fine for now.

Btw, I'm not sure Mailman actually processes the 541 messages we get from these refusals as bounces; the messages just sit in the queue. I think what we are seeing is that bounce processing wasn't working correctly on the old server, and though it was working correctly now the messages weren't all getting through. Now that they are getting through, they are bouncing legitimately for non-existent users. Also, gmail.com and other major providers were never, as Stuart mentioned, blocking us at all. However since I can't be sure about all the bounces, I'm working on a script to parse the mailman bounce log (or what passes for it; Mailman is notoriously bad about logging this sort of thing) and the postfix log and compare to make sure these were bouncing from 550 errors and not the 541's. If there are any outliers, we can add them back in manually. Also if you are an admin for your list, it can't hurt to just add any legitimate looking address back in through mass-subscribe; the worst
that can happen now that mail is relaying properly is that they bounce legitimately and get unsubscribed again.

Jay Schubert <jay.schubert at gmail.com> wrote:

>I also encountered this on and UIUC.edu address.  It wasn't until I
>manually whitelisted chambana in the users CITES control panel for them
>that they received any mails from us.  They had never changed their
>email
>spam filtering level so were rolling completely with default settings. 
>Any
>way to tell how widespread this problem could be?  Any simple way
>around it
>(rename the server?).
>
>Thanks,
>Jay
>
>On Mon, Apr 23, 2012 at 12:38 AM, Stuart Levy <salevy at illinois.edu>
>wrote:
>
>>  This sounds familiar -- I'm seeing such spontaneous un-subscriptions
>on
>> lists that I monitor as well, though not so many.
>>
>> There is some problem (which fortunately Josh understands) where the
>IMC
>> mail server being blacklisted, by *some* mail receivers, due to its
>network
>> being used in the past by spam-senders.
>>
>> This apparently causes some recipients (e.g. att.net/sbcglobal.net)
>to
>> systematically reject mail coming from the IMC.  After some number of
>> rejections (configurable in the Bounce Processing mailman page),
>mailman
>> disables the account, and you as the admin get notified.     But most
>of
>> the bounces are handled in the bowels of mailman where AFAIK we can't
>see
>> them.  It's hard to know just how frequent they are, or which users
>they
>> are happening with, without scanning the mail logs.
>>
>> In other words, it's hard to know just how much mail people should be
>> getting but aren't.   I suspect that people on gmail are getting
>everything
>> (google ignores this blacklist).  It could be that those on
>> att.net/sbcglobal.net and possibly other ISPs are losing many or even
>all
>> messages from IMC lists -- I hope that isn't the case but fear that
>it may
>> be.
>>
>>
>> On 4/22/12 10:48 PM, Josh King wrote:
>>
>> Hey Jay,
>>
>> Unfortunately I'm still working on getting new logging/monitoring
>stuff
>> set up, and didn't finish this weekend. I'll be working on that
>during
>> my free time throughout the week, for as long as that takes. However,
>I
>> did track down most of the addresses you provided to me. They look
>like
>> legitimate bounce processing, with most of them being
>eitherillinois.edu addresses for people who left the University, or
>gmail
>> accounts where gmail claims that they are no longer valid users. So
>my
>> current working theory is that the unsubscriptions are legitimate
>> Mailman bounce processing, which possibly wasn't being processed
>> correctly before the list functionality was temporarily migrated to a
>> new server. It should be possible to turn off bounce processing on
>your
>> list if that is not a desired feature; I can provide instructions.
>> Otherwise, if there are addresses in the list that are being
>> unsubscribed that you know are legitimate, I can investigate further
>in
>> case U of I, Gmail or others are providing illegit responses (which
>> wouldn't be entirely outside the realm of possibility in the arcane
>> world of email processing).
>>
>> On 04/20/2012 09:52 AM, Josh King wrote:
>>
>>  Hi Jay,
>>
>> I've been meaning to clean up the logging. There's now a central log
>server for all system logs, but it currently isn't very well organized.
>If you can forward me the list of addresses off-list, I can track that
>down for you this afternoon, and I'll clean up the logging server and
>send out access information for it this weekend. Sound good?
>>
>> Jay Schubert <jay.schubert at gmail.com> <jay.schubert at gmail.com> wrote:
>>
>>
>>  Tech,
>>
>> Anyone have any advice about what I can check in Mailman to
>understand
>> what's happening when I get notification that 29 or 16 people at a
>time
>> have been unsubscribed from our general mail list (bookstoprisoners)?
>> The
>> only info the email I receive has is "[address] has been
>unsubscribed"
>> (over and over for different addresses x 30 lines) and I've recently
>> gotten
>> a few of them at a surprising rate (I'm used to seeing a few here and
>> there
>> but to see them come in such large batches is alarming).
>>
>> Are there logs that give more info about the reason / action that
>> triggered
>> the unsubscribes?
>>
>> Thanks,
>> Jay
>> _______________________________________________
>> IMC-Tech mailing
>listIMC-Tech at lists.chambana.nethttp://lists.chambana.net/mailman/listinfo/imc-tech
>>
>>
>>
>> _______________________________________________
>> IMC-Tech mailing
>listIMC-Tech at lists.chambana.nethttp://lists.chambana.net/mailman/listinfo/imc-tech
>>
>>
>> _______________________________________________
>> IMC-Tech mailing list
>> IMC-Tech at lists.chambana.net
>> http://lists.chambana.net/mailman/listinfo/imc-tech
>>
>>
>_______________________________________________
>IMC-Tech mailing list
>IMC-Tech at lists.chambana.net
>http://lists.chambana.net/mailman/listinfo/imc-tech

-- 
Josh King

"I am an Anarchist not because I believe Anarchism is the final goal, but because there is no such thing as a final goal." -Rudolf Rocker

Sent from mobile device. Please excuse brevity, typos.