Re: Call for LIST COUNTS testing

> Could people test LIST COUNTS with inn-CURRENT-20091201.tar.gz (or superior)?
> Especially when the news server carries several thousands groups
> and with different overview methods (tradindexed, buffindexed, ovdb).
> We need to be sure that it is not significantly slower than LIST ACTIVE.

Just tested on the news server of The Doctor (whom I thank):

% ls -l act counts
-rw-r--r-- 1 julien user 2206944 Dec  2 13:01 active
-rw-r--r-- 1 julien user 1568808 Dec  2 13:02 counts

% wc -l active counts
  46720 act
  46720 counts

46720 newsgroups with ovdb.

I used:

% getlist -R -h localhost active > active
% getlist -R -h localhost counts > counts

About 1.5 seconds for the first command, 6 seconds for the second.

It seems to be reasonable, especially when we consider that 1.6 MB in 6 seconds
means that we have a 2 Mbit/s transfer (for one client only).

In space (bandwidth), we have a gain of about 30% here.

Incidentally, I wonder whether we could rewrite the contents of the active

[PATCH] newsfeeds.in - NoCeM newsgroup incorrect

With a fresh installation of INN 2.5.1, I noticed that the (helpful)
nocem feed configuration in the provided pathetc/newsfeeds file
specifies an incorrect newsgroup.

Thank you for your work on this project.

--- samples/newsfeeds.in.orig	2009-10-13 05:24:04.000000000 +1100
+++ samples/newsfeeds.in	2009-12-03 11:25:36.000000000 +1100
 <at>  <at>  -118,7 +118,7  <at>  <at> 
 # Feed NoCeM notices to perl-nocem in order to process third-party,
 # PGP-signed article cancellation notices.  See perl-nocem(8) for more
 # details on how to set this up.
-#nocem!:!*,alt.nocem.misc,news.lists.filter\
+#nocem!:!*,alt.nocem.misc,news.lists.filters\
 #	:Tc,Wf,Ap: <at> bindir <at> /perl-nocem

 # News to mail gateway.  Similar to innfeed, this uses a master feed and

--

-- 
John Marshall

_______________________________________________
inn-workers mailing list
inn-workers <at> lists.isc.org
https://lists.isc.org/mailman/listinfo/inn-workers

Re: [PATCH] newsfeeds.in - NoCeM newsgroup incorrect

Hi John,

> With a fresh installation of INN 2.5.1, I noticed that the (helpful)
> nocem feed configuration in the provided pathetc/newsfeeds file
> specifies an incorrect newsgroup.
>
> -#nocem!:!*,alt.nocem.misc,news.lists.filter\
> +#nocem!:!*,alt.nocem.misc,news.lists.filters\

Nice catch, and thanks for having told us.
The group was right in the perl-nocem man page, but not in the newsfeeds
sample file.

Now fixed.

> Thank you for your work on this project.

Do not hesitate to report us any issue you may encounter.

--

-- 
Julien ÉLIE

« Prolonger les adieux ne vaut jamais grand-chose ;
  ce n'est pas la présence que l'on prolonge, mais le départ. » (Bibesco) 

_______________________________________________
inn-workers mailing list
inn-workers <at> lists.isc.org
https://lists.isc.org/mailman/listinfo/inn-workers

Injection-Date and Injection-Info in nnrpd

Hi,

Here is what I suggest for nnrpd (shipped with INN 2.5.2).
Fell free to comment it.

* The Injection-Date: header can be set by a posting agent.  If not
  set, it is always added with POST (not IHAVE).  If set but wrong,
  the article is rejected before its injection.

* The Injection-Info: header cannot be set by a posting agent.
  It is always added with POST (not IHAVE).  Examples:

  Injection-Info: <path-identity>; posting-host="<Client.host>:<Client.ip>";
          logging-data="<pid>"; mail-complaints-to="<complaints>"

  Injection-Info: news.trigofacile.com;
          posting-host="aaubervilliers-151-1-48-186.w83-114.abo.wanadoo.fr:83.114.132.186";
          logging-data="25669"; mail-complaints-to="abuse <at> trigofacile.com"

  Is it the server name in <path-identity>?  (I do not find where it is
  explained in RFC 5536.)  We put here the domain or the virtual domain.

  There is no posting-account information.  Unless you see something interesting
  to give here, known by nnrpd?

  There may be posting-host="83.114.132.186:83.114.132.186" if reversing
  the IP address does not work.

* Go on rejecting articles whose Date: header is too far in
  the past, even though Injection-Date: is fine.

Re: Injection-Date and Injection-Info in nnrpd

Julien wrote:
> * The Injection-Date: header can be set by a posting agent.  If not
>  set, it is always added with POST (not IHAVE).

Why not insert this header using IHAVE of nnrpd? This is injection too,
just using a different command and should be handled like POST. Of course,
it should not be added by IHAVE of innd.

>  If set but wrong, the article is rejected before its injection.

Hmm. Some people use always midnight for Date: header for privacy reasons
(e.g. they are posting during working hours). Injection-Date: will disclose
the exact date and should be made configurable for privacy reasons.

> * The Injection-Info: header cannot be set by a posting agent.
>  It is always added with POST (not IHAVE).  

As mentioned above: IHAVE in nnrpd (not used very often) should be handled
like POST because it is injection too.

>  Is it the server name in <path-identity>?

Should be the same string which is added to the Path: header. Necessary to
identify Path: preloading or article re-injection.

>  There may be posting-host="83.114.132.186:83.114.132.186" if reversing
>  the IP address does not work.

If not resolvable it should be just posting-host="83.114.132.186" as
I understand RFC 5536.

Re: Injection-Date and Injection-Info in nnrpd

Julien ÉLIE <julien <at> trigofacile.com> writes:

>  Is it the server name in <path-identity>?  (I do not find where it is
>  explained in RFC 5536.)  We put here the domain or the virtual domain.

Yes, the intention was to use <path-identity> here.

>  There is no posting-account information.  Unless you see something
>  interesting to give here, known by nnrpd?

We could give the authenticated user, but we probably shouldn't by
default.

>  There may be posting-host="83.114.132.186:83.114.132.186" if reversing
>  the IP address does not work.

It looks like we're supposed to omit the part before the colon if we just
have an IPv4 or IPv6 address.

> * Also-Control:, Article-Names:, Article-Updates: and See-Also: are
>  obsolete and refused in INN 2.5.2.  [Or INN 2.6.0?]

> * No longer add the Lines: header in INN 2.5.2.  [Or INN 2.6.0?]

I tend to make changes like that in a major release, but it probably
doesn't matter too much either way.

--

-- 
Russ Allbery (rra <at> stanford.edu)             <http://www.eyrie.org/~eagle/>

Re: Injection-Date and Injection-Info in nnrpd

Hi Heiko,

>> * The Injection-Date: header can be set by a posting agent.  If not
>>  set, it is always added with POST (not IHAVE).
>
> Why not insert this header using IHAVE of nnrpd? This is injection too,
> just using a different command and should be handled like POST. Of course,
> it should not be added by IHAVE of innd.

RFC 3977:
   [IHAVE] differs from the POST command in that it is intended
   for use in transferring already-posted articles between hosts.  It
   SHOULD NOT be used when the client is a personal news-reading
   program, since use of this command indicates that the article has
   already been posted at another site and is simply being forwarded
   from another host.

Besides, IHAVE in nnrpd already behaves like this.  For instance,
a post without a message-ID will be rejected by nnrpd when using IHAVE.
It is the same for a post without a Path: header.

>>  If set but wrong, the article is rejected before its injection.
>
> Hmm. Some people use always midnight for Date: header for privacy reasons
> (e.g. they are posting during working hours). Injection-Date: will disclose
> the exact date and should be made configurable for privacy reasons.

When I say "wrong", I mean that it is an invalid date (something that
cannot be parsed as a date).  So I believe the issue you report will
not occur.

Re: Injection-Date and Injection-Info in nnrpd

Julien wrote:
>   [IHAVE] differs from the POST command in that it is intended
>   for use in transferring already-posted articles between hosts.

This is usually true. But for "transferring already-posted articles between
hosts" people use innd and not nnrpd.

> It SHOULD NOT be used when the client is a personal news-reading program,
> [...]

Then IHAVE is not necessary in nnrpd at all. In fact, it was not
implemented for long time, if I remember correctly. Unfortunately people
use versions of Microsoft Exchange as NNTP client which always uses IHAVE
(instead of POST).

I don't see any reason why endusers should be handled different if they use
POST and IHAVE. With a trivial script it is possible for endusers posting
articles with IHAVE command in nnrpd and avoid insertion of Injection-...
headers.

Distinction between endusers (--> nnrpd) and feeding servers (--> innd)
makes more sense to me.

Heiko

Heiko Schlichting          Freie Universität Berlin
heiko <at> CIS.FU-Berlin.DE     Zentraleinrichtung für Datenverarbeitung (ZEDAT)
Telefon +49 30 838-54327   Fabeckstraße 32
Telefax +49 30 838454327   14195 Berlin
_______________________________________________

Re: Injection-Date and Injection-Info in nnrpd

Hi Heiko,

>>   [IHAVE] differs from the POST command in that it is intended
>>   for use in transferring already-posted articles between hosts.
>
> This is usually true. But for "transferring already-posted articles between
> hosts" people use innd and not nnrpd.
[...]
> Then IHAVE is not necessary in nnrpd at all.

Maybe to reinject already posted articles (via suck, newsx, pullnews)
in case the user who does that only has an access to nnrpd?
But these programs can also use POST...

> In fact, it was not
> implemented for long time, if I remember correctly. Unfortunately people
> use versions of Microsoft Exchange as NNTP client which always uses IHAVE
> (instead of POST).

New in INN 2.4.0:
    nnrpd now optionally supports article injection via IHAVE, see readers.conf(5).
    Any articles injected this way must have Date, From, Message-ID, Newsgroups,
    Path, and Subject headers.  X-Trace and X-Complaints-To headers will be added
    if the appropriate options are set in readers.conf, but other headers will
    not be modified/inserted (e.g. NNTP-Posting-Host, NNTP-Posting-Date, Organization,
    Lines, Cc, Bcc, and To headers).

> Distinction between endusers (--> nnrpd) and feeding servers (--> innd)
> makes more sense to me.

mailpost whoes about undefined $_

Hi folks,

I've got an error message from mailpost telling me

------ pipe to |/usr/lib/news/bin/mailpost -b /var/tmp my.newsgroup
       generated by my.newsgroup <at> mailhost ------

Use of uninitialized value $_ in pattern match (m//) at /usr/lib/news/bin/mailpost line 204, <STDIN>
line 2.

Most EMails are processed without this problem. Here is a
suggested patch:

--- a/frontends/mailpost.in	2009-05-21 22:08:33.000000000 +0200
+++ b/frontends/mailpost.in	2009-12-07 09:47:12.000000000 +0100
 <at>  <at>  -201,7 +201,7  <at>  <at> 
     ($_, $line) = ($line, $_) ; # Swap $line and $_.

     last if defined($_) && /^$/ ;
-    next if /^$/ ;		# Only on first header will this happen.
+    next unless defined($_) ;		# Only on first header will this happen.

     push  <at> errorText, "($_)\n";

Would you mind to check?

Many thanx

Harri