RE: Syslog - monitoring the bigger picture
"Tina Bird" <[email protected]> Thu, 13 Sep 2007 16:11:37 -0700
| Newsgroups | gmane.comp.sysutils.loganalysis |
|---|---|
| Message-ID | <018101c7f65b$6fc692d0$1701a8c0@lindesfarne> |
> Just curious, what is wrong with all other past logging projects, that > seek to accomplish just about the same? now *there's* a question for the ages. what i usually tell my students: "some junior sys admin who just took a perl/C/lisp/whatever class is tasked with centralizing and monitoring system logs at her job. after getting all the data centralized, she begins to look at the data and to try to understand what's important and what isn't. once she starts figuring out what she needs in terms of reports and alerts, she hits the net, because *surely* this is a solved problem. she gets a lot of google hits for swatch, so she starts there -- but it doesn't handle thresholding traffic, which is critical for the firewall, and the job of coming up with all the right keywords is less than exciting. so she asks around a bit, and someone mentions logsurfer, cos it lets you deal with multi-line messages and context, so she can get the contexting she needs. but logsurfer is pretty complicated for what she needs, and she can't tell whether it's still being maintained, or, for that matter, if it *matters* whether or not it's still being maintained. and her manager keeps asking for progress reports. so finally she throws together a little script/program/spell that does *exactly* what her organization needs. it may be tweaked for a particular vendor's products, or the specific reporting requirements of her industry, or whatever. but it does the trick for her and her manager is happy and she can finally get onto something more interesting. then she posts it on her website, and (with any luck) pretty much forgets about it for the rest of her life." ...lather, rinse, repeat. let me pre-emptively apologize to all list members who have contributed to the log analysis and management tools that are out there. it's clear that many of the folks involved in these projects take a much more big-picture view than i've painted here, and i don't mean to smear them with the same brush. but wow, after 5 years of trying to maintain an up-to-date list of log parsing and analysis tools -- which included attempting to figure out how they differed from each other -- this is the only explanation i've been able to come up with. cheers -- tbird