INN-current (20090414) and some log messages

[Petr Novopashenniy]

Good day, Julien!

On Mon, 27 Apr 2009, Julien ?LIE wrote:

JL> Hi Petr,
JL> 
JL> > I have ~2000-3000 entries in my daily news log file:
JL> 
JL> Not mentioned as unknown lines by innreport in its daily mail I assume.
Yes, no "unknown" lines in innreport.

JL> 
JL> 
JL> > Apr 27 15:10:01.245 - ? (null) 439 Bad "Message-ID" header
JL> > 
JL> > It is possible to log peer name?
JL> 
JL> Well, as innd did not log it, I believe it is because there was no Path:
JL> header in the article.  It takes the peer name from the Path: header
JL> except when logipaddr is set to true in inn.conf.

Oh! Are you confident in that?

In all of my "news" logs (INN-current and 2.4.*), I see only names (or 
IPs, if no FQDN for this peer in incoming.conf) from icoming.conf, for 
example:

Apr 28 03:00:47.975 + news-out.glorb.com 
<49f6391d$0$5486$bbae4d71 at news.suddenlink.net> 1884 

No "news-out.glorb.com" in "Path" header, but in incoming.conf:
hostname: "news-out.glorb.com"

And so it was very long, as I remember.


JL> Maybe we should try to behave as though logipaddr were true when
JL> there is no Path: header.

My "logipaddr: true" in all our INN installations, and this is default 
(according to the documentation) for INN.

JL> 
JL> 
JL> > I have no example of these Message-ID's (or whole articles), but maybe
JL> > possible to remember them from "takethis" output? Or is it the wrong way?
JL> 
JL> I see that the message-ID which was found is "reset" when the error
JL> is found.  I can try to log it at the end of the error.
JL> 
JL> As for remembering it from the argument to TAKETHIS or IHAVE, I do not
JL> know whether it is a good idea.  The standard allows to say "IHAVE <a>"
JL> and send <b>...
JL> 
JL> 
JL> Could you please try that patch and tell me how your logs now look like?
JL> 
JL> 
JL> Index: innd/art.c
JL> ===================================================================
JL> --- innd/art.c  (r?vision 8431)
JL> +++ innd/art.c  (copie de travail)
JL> @@ -413,6 +413,8 @@
JL>                 data->Feedsite = hops[0];
JL>             else
JL>                 data->Feedsite = "localhost";
JL> +        } else if (cp->Address.ss_family != 0) {
JL> +            data->Feedsite = RChostname(cp);
JL>         }
JL>     }
JL>     ARTlog(data, ART_REJECT, text != NULL ? text : cp->Error);
JL> @@ -1051,9 +1053,10 @@
JL> 
JL>   /* assumes Message-ID header is required header */
JL>   if (!ARTidok(HDR(HDR__MESSAGE_ID))) {
JL> +    sprintf(buff, "%d Bad \"Message-ID\" header (%s)",
JL> +            ihave ? NNTP_FAIL_IHAVE_REJECT : NNTP_FAIL_TAKETHIS_REJECT,
JL> +            HDR_FOUND(HDR__MESSAGE_ID) ? HDR(HDR__MESSAGE_ID) : "null");
JL>     HDR_LEN(HDR__MESSAGE_ID) = 0;
JL> -    sprintf(buff, "%d Bad \"Message-ID\" header",
JL> -            ihave ? NNTP_FAIL_IHAVE_REJECT : NNTP_FAIL_TAKETHIS_REJECT);
JL>     TMRstop(TMR_ARTCLEAN);
JL>     return false;
JL>   }
JL> 

Now I will try it.

JL> -- 
JL> Julien ?LIE
JL> 

--pety