Re: [leafnode-list] SIGPIPE Issues

[Matthias Andree <ma@xxxxxxxxxxxxxxxxxxxxxxxxxxxx>]

"Michael O'Quinn" <michael@xxxxxxxxxxx> writes:

> Mar 19 02:35:33 shire fetchnews[26161]: >XOVER 23631-30565^M
> Mar 19 02:37:50 shire fetchnews[26161]: {GROUP 1} 23631: reply 500 No server having article is reachable
> Mar 19 02:38:35 shire fetchnews[26161]: {GROUP 1} 23632: reply 500 No server having article is reachable
>
> This goes on for a while, since the server we are connecting to is a sort
> of a caching server (NewsPlex) and it's overview is a little behind. 

Well, fix NewsPlex. It should not offer articles it cannot fetch, and
leafnode is logging the 500 replies when it tries to fetch things.

> Mar 19 03:58:08 shire fetchnews[26161]: >GROUP {GROUP 2}^M
> Mar 19 03:58:08 shire fetchnews[26161]: >GROUP {GROUP 3}^M
> Mar 19 03:58:08 shire fetchnews[26161]: >GROUP {GROUP 4}^M
> Mar 19 03:58:08 shire fetchnews[26161]: >GROUP {GROUP 5}^M
> Mar 19 03:58:08 shire fetchnews[26161]: >GROUP {GROUP 6}^M
> Mar 19 03:58:08 shire fetchnews[26161]: >GROUP {GROUP 7}^M
> Mar 19 03:58:08 shire fetchnews[26161]: >GROUP {GROUP 8}^M
> Mar 19 03:58:09 shire fetchnews[26161]: >GROUP {GROUP 9}^M

No new articles offered.

> Mar 19 03:58:12 shire fetchnews[27012]: reading XOVER info from /var/spool/news/{GROUP 1}/91930
> Mar 19 03:58:13 shire fetchnews[27012]: reading XOVER info from /var/spool/news/{GROUP 1}/91931

Looks correct.

> First, fetchnews should have fetched a lot more articles in {GROUP 1}.
>
> Second, it's likely Groups 2 thru 9 really had no articles.  I had just
> subscribed to them, and NewsPlex needs a few minutes to explore the
> downstream servers before any actual articles are present.  I am running
> fetchnews as we speak, and this time it's picking up articles in all the 
> groups, and it appears to have to completeed {GROUP 1} successfully.
>
> I'd wait for it to finish, but I have to get to work, and I want to post 
> this SOMETIME before the final release.
>
> Third is just a minor logging issue.  I am connecting to {SERVERNAME} port 
> 2000, but the log reported it as port 200.

That's not the port number, but the response of the server. 200: you may
post. 201: you cannot post.

ChangeLog extract:

2002-03-20  Matthias Andree  <matthias.andree@xxxxxx>

	* nntputil.c: Log port number in "connected to" syslog message,
	and mark the "200" with "reply:" so users know it's not the port
	number. Suggested by Michael O'Quinn.

Log messages now look like this:

Mar 20 12:59:39 merlin fetchnews[25834]: connected to 127.0.0.1:119, reply: 200

> Finally, is there something I should be doing which will give you more
> information about this?

Nope, that's sufficient to know the bug is outside leafnode, and
leafnode has functioned seemingly correctly.

Nonetheless, 1.9.20.rc8 will log many more replies to NNTP commands than
rc7, so check back after 1.9.20.rc8 is released (really soon now).

-- 
Matthias Andree

-- 
leafnode-list@xxxxxxxxxxxxxxxxxxxxxxxxxxxx -- mailing list for leafnode
To unsubscribe, send mail with "unsubscribe" in the subject to the list