gate_news should catch lost connection

[Jarrell-users]

Sep 20 10:51:35 2002 gate_news(9301): Traceback
(most recent call last):
Sep 20 10:51:35 2002 gate_news(9301):
File "/home/mailman/cron/gate_news", line 270,
in ?
Sep 20 10:51:35 2002 gate_news(9301): main()
Sep 20 10:51:35 2002 gate_news(9301):
File "/home/mailman/cron/gate_news", line 250, in
main
Sep 20 10:51:35 2002 gate_news(9301):
process_lists(lock)
Sep 20 10:51:35 2002 gate_news(9301):
File "/home/mailman/cron/gate_news", line 194, in
process_lists
Sep 20 10:51:35 2002 gate_news(9301): conn,
first, last = open_newsgroup(mlist)
Sep 20 10:51:35 2002 gate_news(9301):
File "/home/mailman/cron/gate_news", line 89, in
open_newsgroup
Sep 20 10:51:35 2002 gate_news(9301):
password=mm_cfg.NNTP_PASSWORD)
Sep 20 10:51:35 2002 gate_news(9301):
File "/usr/local/lib/python2.2/nntplib.py", line
117, in __init__
Sep 20 10:51:35 2002 gate_news(9301):
self.welcome = self.getresp()
Sep 20 10:51:35 2002 gate_news(9301):
File "/usr/local/lib/python2.2/nntplib.py", line
203, in getresp
Sep 20 10:51:35 2002 gate_news(9301): resp =
self.getline()
Sep 20 10:51:35 2002 gate_news(9301):
File "/usr/local/lib/python2.2/nntplib.py", line
192, in getline
Sep 20 10:51:35 2002 gate_news(9301): line =
self.file.readline()
Sep 20 10:51:35 2002 gate_news(9301): IOError :
[Errno 131] Connection reset by peer

[http://sourceforge.net/tracker/index.php?func=detail&aid=620033&group_id=103&atid=100103]

[Barry Warsaw (barry)]

I didn't realize that nntplib.NNTP() can raise an IOError --
I thought it would have been a socket.error. Oh well,
here's an untested patch. I think I'll check this in just
because I don't see how it could break anything.

[Barry Warsaw (barry)]

I'm applying this patch

[Jarrell-users (jarrell-users)]

Didn't work... This comes up, btw, not during initial
connection, but if the connection goes away during a
gate_news run, like the daemon on the other end dies,
or the news server reboots, etc. You get an IOerror on
the mailman side.

[Barry Warsaw (barry)]

Hmm, the specific traceback you pasted indicates an error
during the initial connection. I think that's about the
best we're going to do for MM2.1 final.