log level inconsistancies (was: bind 9.1.0 issues :-\)

sidster patrick at boxsoft.com
Tue Feb 6 07:31:58 UTC 2001



Hi Bind-ites, Brett,

I'm not sure if I have found the problem or not.

I've been browsing through the source code from the bind 9.1.0 I
downloaded and installed.

One of the issues I have is the fact that I was unable to get either of
my name server to log any of the xfer-out messages.  I read the docs
(namely bind-9.1.0/doc/arm/Bv9ARM.ch06.html ) but there seemed to be
something missing.

I looked at (relative to the bind-9.1.0/ directory):

   bin/tests/named.conf : line 455

There you'll find this comment:

   /*
    * Channels have a severity level.  Messages at severity levels
    * greater than or equal to the channel's level will be logged on
    * the channel.  In order of decreasing severity, the levels are:
    *
    ...
    */

Which is consistent with section 6.2.10.1 in Bv9ARM.ch06.html:


   Messages which are not at least of the severity level given
   will not be selected for the channel; messages of higher
   severity levels will be accepted


However, looking at:

   lib/isc/log.c : line 1277

The comment says otherwise:


   /*
    ...
    * If the level is (mathematically) less than or equal to the
    * highest_level, or if there is a dynamic channel and the level is
    * less than or equal to the debug level, the main loop must be
    * entered to see if the message should really be output.
    ...
    */


And in fact that is what the code seems to be doing on line 1292:


In function isc_log_wouldlog( isc_log_t*, int ) called from function
isc_log_doit( ... ) defined after it (line 1298).

Please also note that it will not log at all if the log level is
ISC_LOG_DYNAMIC (see line 1320)!


   /* ... */
    return (ISC_TF(level <= lctx->logconfig->highest_level ||
            (lctx->logconfig->dynamic &&
             level <= lctx->debug_level)));



Is this the intended behavior?


To test my guess I changed the severity for my xfer_channel to debug 9
and all of a sudden i see zone transfers being logged.




My second issue with the classless reverse lookup is still outstanding
with no suggestions offered to me from either the list or the news
group.

Hopefully the transfer logging will shed some light on this problem.
But I will appreciate any input whatsoever from anyone regarding this
matter.  :-)


Regards,


* Brett Gianpetro (brettg at whatupg.com) [20010204 21:54]:
> > verify that my post was successful until you emailed me.  My post never
> > showed up after I posted to the group.  :-\
> 
> Since the bind-users mailing list and the comp.protocols.dns.bind groups
> are one and the same, you can always check the archives at isc.org.
> 
> > Please do the same,
> 
> Will do.  
> 
> -Brett
> 
> 
> > 
> > 
> > * Brett Gianpetro (brettg at whatupg.com) [20010203 13:01]:
> > > I am having the same problem with logging zone transfers (see my message
> > > from a couple of days ago, "logging zone transfers in bind
> > > 9.1.0").  I've tried using the category "xfer_out", and I've also tried
> > > letting the default logs catch it.  Either way I can't get it to work.  So
> > > I think the question becomes this: Are outgoing zone transfers being
> > > logged at all in Bind 9.1.0?  If so, is it OS specific?  I am also running
> > > Linux (slackware) on x86.
> > > 
> > > -Brett



patrick
--
It's a damn poor mind that can only think of one way to spell a word.
      -- Andrew Jackson


More information about the bind-users mailing list