[Mailman-Users] Current stump (newbie)

[Richard Barrett]

On 8 May 2006, at 22:04, William D. Tallman wrote:

> On Mon, May 08, 2006 at 10:51:41AM -0500, Patrick Bogen wrote:
>> On 5/8/06, William D. Tallman <wtallman at olypen.com> wrote:
> <snip>
>
> This is the point at which Mailman initiates a response.
>
>>> May  7 20:25:31 mailhost sm-mta[4082]: k483PUSs004082:
>>> from=<mailman-bounces at mailhost.locallan>, size=2772, class=-60,
>>> nrcpts=1, msgid=<mailman. 
>>> 1.1147058729.3838.mailman at mailhost.locallan>,
>>> proto=ESMTP, daemon=MTA, relay=localhost [127.0.0.1]
>>> May  7 20:25:31 mailhost sm-mta[4084]: k483PUSs004082:
>>> to=<wtallman at olypen.com>, delay=00:00:00, xdelay=00:00:00,  
>>> mailer=relay,
>>> pri=140772, relay=smtp.olypen.com. [208.200.248.8], dsn=5.1.1,  
>>> stat=User
>>> unknown
>>> May  7 20:25:31 mailhost sm-mta[4084]: k483PUSs004082:  
>>> k483PVSs004084:
>>> DSN: User unknown
>> Mailman has generated a response, and attempted to send it to
>> wtallman at olypen.com, but it was rejected by the remote SMTP server.
>> The SMTP server reported 'User unknown'. Check the logs on
>> smtp.olypen.com, if you can, for more information about this.
>
> Called Olypen, and it would seem that the smtp session must have gone
> like such:
>
> telnet smtp.olypen.com 25  //  Get okay.
> HELO mailhost.locallan  // Get okay (250).
> MAIL From: mailman-bounces at mailhost.locallan // User unknown?
>
> The tech support guy said that 'mailman-bounces' is an   
> unauthenticated
> User, or so he thought.  I'll know more later when the guy at the NOC
> gets back to me.  Maybe it will be a case of getting Mailman
> authenticated on their smtp server.  That sounds like the server will
> only accept mail from subscribers, which makes sense.
>
> -----------------------------
> Addendum, just before sending this off:  The NOC guy just called back,
> and said that indeed 'mailman-bounces' is unauthorized at their smtp
> server.  After some discussion (very few of their subscribers run  
> Linux,
> although they themselves run RedHat on their servers), I discovered  
> that
> they are contemplating email list service as an addition to what they
> already offer.
>
> Such service would be industrial/commercial and cost several hundred
> bucks a month, I suggested, and we agreed I wouldn't be paying that  
> sort
> of freight.  OTOH, as they really like OS/FSF software, it would be  
> very
> likely they would choose Mailman.  With that in mind, I said I'd be  
> very
> happy to share my experiences and what I learned if I could  
> successfully
> set up my server.
>
> On that note, the guy said he would advocate opening their port 25 for
> me, the decision to be made by the email sysadmin.  So we'll see what
> happens.
>
> OYAH (On Yet Another Hand...), if I could configure Mailman to use
> 'my_list' instead of 'mailman-bounces', all this would be resolved, as
> 'my_list' is an authenticated User; it's one of my mailbox addresses.
>
> If that is possible, that's what I would do.  Hafta look at the source
> code...  and I've never even looked at Python...  this oughta be fun!

The problem you get with this solution and the reason why outbound  
mail comes from the <listname>-bounces alias rather than the  
<listname> alias is that if outbound traffic to your subscribers  
bounces then the bounce messages end up being re-distributed to the  
subscribers, just like regular posts to the list! The bounces alias  
captures bounce messages and the Mailman bounce handler deals with them.

> -----------------------------
>> -- SNIP --
>>
>>> That's the end of the grunner log file.  Note that no activity is
>>> recorded for the time period of the maillog.
>> The qrunners each, more or less, have their own logs. the main  
>> qrunner
>> log only records what you see; when they started, when they failed,
>> when they restarted.
>
> Okay, that makes sense.  Didn't know what I should find there.
>
>>> And no mail is sent back out, apparently, because none is  
>>> received by
>>> the list subscriber.
>> See my comments above.
>>
>>> Now, I've already demonstrated my abysmal ignorance of this stuff in
>>> previous comments, but I'd sure like to get this running if I  
>>> can.  Is
>>> there enough information here to diagnose this situation?  If  
>>> not, what
>>> can I provide?
>> Mailman *seems* to be working, but I might be missing something. If
>> nothing else, I don't know that 'my_list at ...' should be sending to  
>> the
>> mailman list.
>
> Well, 'my_list@' is one of the email addresses I get with my
> subscription to Olypen's services (5 with aDSL).  So Fetchmail gets  
> mail
> from that mailbox and sends it to Mailman.
>
> fetchmailrc snippet:
>          user 'my_list' there with password '******' is 'mailman' here
>
> 'mailman' is, of course, a user on this machine, so messages sent from
> Fetchmail wind up in /var/spool/mail/mailman.  Apparently, Mailman  
> likes
> that just fine.
>
> The deal breaker here would be if Olypen insists on establishing
> mailman-bounces as a separate subscriber.  At that point, this entire
> setup -> /dev/null.
>
> Thanks for this response; it verified my suspicions and pointed me to
> the presumed source of the problem: the ISP.
>

Time to change?

> Thanks for reading,
>
> Bill Tallman
>