log noise level

[John Gerth]

John H Terpstra wrote:

> On Tuesday 04 January 2005 11:28, John Gerth wrote:
> 
>>In the wake of moving over the last year from NT to Samba PDCs
>>(both for my own group and helping other people), I've had to chase
>>a number of annoying residual problems for which the samba debug
>>log messages proved invaluable.  However, they would've been much
>>more valuable several times if there had been an option to get
>>the IP address included in the bracketed message header, e.g.
>>    [2006/01/04 09:26:39, 1, 10.0.0.1] ....
>>in order to be able to see where the request was being generated.
> 
> 
> I would strongly object to such a modification as it will add needlessly to 
> the already burgeoning volume of log information. As you show in the remained 
> of your message that I cut for brevity, it is possible to collect log info 
> per IP address using the %m and %M macros to the "log file" parameter. I 
> believe this is the preferred method.
> 
> No offense is meant, but this modification just adds to the log noise level. 
> That's a bad thing in my NSHO.
> 
  No offense taken...that's why I wrote later in the post that it be
  optional. Subsequent responses suggested that something like "debug pid"
  would be appropriate.

  You also raise a good general question about log noise level.  I'm relatively
  new at using the Samba logs, but from a past life as a software developer I know that
  it's hard to get the right granularities assigned. It's not clear how they
  work in Samba.  For me, I find that 99% of our problems can be solved at
  level 4 or less, but even there I see lots of messages about alloc/dealloc
  which I would've thought to have required a much higher level to trigger.

Is there any developer document which indicates how to to assign levels
to messages?

  I also remember that sometimes messages are added to tackle particular,
  usually internmittent, problems but then have a way of living on even
  though the problem was found and fixed.

Do Smaba log messages get reviewed on some periodic basis?