Maybe this is overkill (or maybe I'm missing the point completely), but
wouldn't Splunk ( http://www.splunk.com ) be a good solution, or tool
for creating a solution, for this problem?
For those that haven't heard of it, it collects data from many different
data sources, syslog being one of them, and provides you with a web
based interface to search through them. It allows for complex searches
and has the ability to alert on any of the searches. They also have
something called SplunkBase, which is a community driven database of
what many of these messages mean. It is a commercial product, but they
have a free version that will work with up to 500Mb per day of data.
I haven't implemented this myself yet, but I have played around with it
and look forward to finding the time to try to really implement it for
myself.
Mike Robbert
Hari Sekhon wrote:
I'm also extremely interested in expanding my log watching to include
a massive amount of comprehensive pattern matching alerting.
I currently have some but need to expand it. The problem is that this
is really a difficult thing to approach because it can only catch
known patterns in this fashion. And whitelisting is really not
practical in this context as the logs generated are practically
infinite and not really able to whitelist them.
I think that there should really be a well maintained project of
regexs for this purpose, one official champion for us to build our
baselines on... with frequent updates...
Anyone got any ideas or regexs they want to share?
Isaac, you would do well to have things like "I/O Error" for disk
problems... "hardware hung"... etc etc, but this list is practically
endless, you should look at your logs and decide which ones you'd like
to be alerted on.
-h
Hari Sekhon
Isaac Perez Moncho wrote:
Hello,
I just installed swatch, and used this configuration file for the
checks:
http://www.loganalysis.org/sections/signatures/log-swatch-skendrick.txt
Anyone knows any other common phrase or word that I should find the logs
for hardware and system errors?
Or what you consider important to monitor in the logs?
Thanks
