[Mailman-Users] Mailman is no longer gatewaying news... - SOLVED

Grant Taylor gtaylor at riverviewtech.net
Fri Nov 14 22:29:14 CET 2008

On 11/13/08 11:53, Mark Sapiro wrote:
> You may have never seen this because it may be a one-time transient 
> error. Or, if it did show up from cron, it would probably be emailed 
> as output from the cron, which would be sent to 'mailman' unless you 
> have a MAILTO in the crontab, and depending on how your 'mailman' 
> site list is set up, you may never see it.

I'll look in to setting the MAILTO for the user that is running the cron 
job.  As is I did not see much of any thing at all, other than what I 
sent to the mailing list.

> In any case, the above error says gate_news is closing the connection 
> to the news server, and the underlying Python nntplib received no 
> response to the QUIT command it sent to the server.


> As far as your actual problem is concerned, since gate_news is being 
> run, I am guessing that for some reason, the messages were renumbered 
> on the news server and the current message numbers are all less than 
> the list's usenet_watermark. If this is the case, you have a couple 
> of choices.

This is indeed the case.  As I stated in another message, my news 
provider changed from an old server (with high article numbers) to a new 
server (with low article numbers).

> You can do a mass_catchup from the list's Mail<->News gateways page. 
> If the watermark is the issue, this will allow news->mail to begin 
> again, but previously posted news that hasn't been gated will be 
> lost.

Ugh.  I don't particularly care for this idea.

> To avoid losing messages, you would need to access the news server 
> and determine the current number of the last gated message and use 
> Mailman's bin/withlist or bin/config_list to set the list's 
> usenet_watermark to that value.


This is the route that I went.  I ended up running TCPDump and sniffed 
the traffic from my NNTP client (Thunderbird) and the news server to get 
the article number of the last article that Mailman gatewayed for me.

In the interest of people searching the archives, here is what I did to 
solve my problem.  I then used the following command(s) to set the 
watermark for my mailing list.

    $ bin/withlist -l comp.mail.sendmail
    >>> m.usenet_watermark = 41965
    >>> m.Save()
    >>> ^D

(comp.mail.sendmail is my mailing list and 41965 is what I set the 
watermark to)

When I started writing this reply Mailman / gate_news was busy 
downloading messages from the server (as seen by TCPDump).  And by now 
I'm receiving messages from Mailman in my email client like I'm suppose 
to do.

So I believe that everything is solved.

Thank you all for your help.

In the end I'm glad that it was actually something nefarious (changing / 
renumbering article IDs) on the providers end rather than a problem in 
Mailman.  Things like this help bolster confidence in programs.  :)

Thanks again for all your help!

Grant. . . .

More information about the Mailman-Users mailing list