Re: [logs] most popular reports...?

From: Stephen P. Berry (spb@private)
Date: Thu Aug 19 2004 - 00:35:31 PDT


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1


Devdas Bhagat writes:

	[...]
>Top N blocked spam sending systems (by IP, by domain and by ip whois).
>	-- Useful for judging how good/bad a DNSBL is with respect to
>		your requirements
>Top N servers handling the mail load.
>	-- Ideally, this should be equally well distributed.
>Top N ports being probed.
>	-- New attacks?
	[deletia]

The interesting thing I get out of reading the report ideas people have
posted to this list is this:  syslog really has no mechanism by which
such information may be unambiguously conveyed.  Certainly we can
use swatch(1) or its moral equivalent to fling logs through a regex(3)
sieve, but these general classes of events cannot be identified with
the concision (or portability) that, say, the service or severity can.

I've actually made this observation (and the related suggestion:  syslog(3)
should really have better expressive power), but haven't really seen
much discussion about this sort of thing (including in the discussion
of the new IETF draft).

Am I just nuts (about this, I mean)?  One of the things that sucks
about vanilla syslog is the transport.  There is, however, vo-friggin-luminous
discussion of this end of the problem.  The larger problem, to my mind
anyway, is the ability to unimbiguously evaluate what a line from syslog
means (or, more properly, what it is intended to represent).

My (perhaps naive) thought is that a better-enunciated diction would be
solve a big whollop of this problem.  I.e., having an application report,
via syslog, the equivalent of `application foo (with PID bar) on machine
baz has reported a failed authentication attempt' (think something
like `syslog(LOG_NOTICE, LOG_AUTH, AUTH_FAIL, "[This space intentionally \
left blank");' ), rather than attempting to convey this information entirely
in the human-readable portion of the log[0].

Granted, an individual application could create a set of tokens for this
purpose, assuming a parsing widget at the far end with the same lexicon.
But what I'm talking about is a general and (insofar as a new rev of
syslog would be) universal diction.




- -spb

- -----
0	Only a strawman example, obviously.  The actual form of such
	a diction is beyond the scope of this message, although I'd
	be delighted to discuss the idea in greater (and even tiresome)
	depth.

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.2 (OpenBSD)

iD8DBQFBJFg5G3kIaxeRZl8RAiqJAJ4gQOuQre8XLeW1ARo78Zn2QrULRgCg9c+B
x0YCfcgA3yhe4mx5tgQ7bd4=
=q0c0
-----END PGP SIGNATURE-----
_______________________________________________
LogAnalysis mailing list
LogAnalysis@private
http://lists.shmoo.com/mailman/listinfo/loganalysis



This archive was generated by hypermail 2.1.3 : Thu Aug 19 2004 - 09:52:54 PDT