[Mailman-Developers] Re: [Mailman-Users] forever pending subscriptions

barry@wooz.org barry@wooz.org
Thu, 9 Nov 2000 14:13:30 -0500 (EST)


Trying to come back to this problem.  It looks like Eric Wang is also
seeing problems, so there must be /something/ going on.  I can't
figure out what though, so perhaps one of you guys can give me ssh
access to the host in question?

>>>>> "AG" == Andreas Gietl <gietl@gietl.com> writes:

    AG> i'm sorry, but there is no pipeline entry:

    | that's the db-file:
    | {   'filebase': 'ccfe93988bf681ad271ad049fa54dcd7260a60ac',
    |     'listname': 'bittecomgietltest2',
    |     'torequest': 1,
    |     'version': 2}

Okay, this looks pretty good, assuming the file is
qfiles/ccfe93988bf681ad271ad049fa54dcd7260a60ac.db

We don't get a `pipeline' key in the dictionary because, by virtue of
the `torequest' key, qrunner sends this message through the email
command parser instead of the delivery pipeline.  So far so good.

    AG> and this is the .msg file:

[...most headers deleted...]

    | To: bittecomgietltest2-request@bitte.com
    | Subject: Re: Bittecomgietltest2 -- confirmation of subscription --
    | request 569170
    | References: <E13s2o6-0003dd-00@d18.x-mailer.de>
    | Content-Type: text/plain; charset=iso-8859-1
    | Content-Transfer-Encoding: 8bit

I notice that the Subject: line is wrapped, but not in an rfc 822
valid way.  I'm assuming that something got munged in the forwarding
of the message to this list, and that the text `request 569170' is
actually at the end of the Subject: line.

In my own tests, if the message looked exactly as you sent it, with
the line being wrapped and `request' showing in column zero, the
mailcmd parser will generate an error message back to the sender.  But
you're not seeing that.  If the `request' line were wrapped a la rfc
822, Mailman should be able to grok out the command just fine.  But
you're not seeing that either.

Another log file to look at: doe you see anything in logs/bounce?  I
don't expect you to -- just wondering.

We're at the point of using debug statements now, and I don't want to
bore mailman-users with that.  If I can't get ssh access to your
installation, let's carry on the debugging conversation on
mailman-developers only.

-Barry