Re: URL content checking [Vote]

[Michael Shigorin]

On Thu, Dec 02, 2004 at 01:29:30PM +0100, Jan-Henrik Haukeland wrote:
> This is a nice feature and I propose that we try to implement
> this in monit.  We need to vote, if we should add this feature
> in monit. So please cast your votes!

Somehow I've missed the original message (not even in spambox),
but here's my E.05: maybe it's time to consider plugin
infrastructure or callable tests?

We've just been discussing monitoring specific applications
(which aren't of any significance outside quite specific area)
which use some specific protocol (they want us to help with
system tuning and deployment), so writing a module for monit
itself would be a spam although being able to either dlopen()
something local or just run the "pinger" would be enough.

As monit (IMCO) should be quite resistant to system failures
being quite "basemental" software piece (remember SMTP-related
discussion? :), being less fault-prone regarding e.g. broken
libraries for some fancy test (e.g. involving Oracle OCI) would
bring in some solid points over the years and installations.

>       check URL myURLCollection
>               if failed urls [
>                       http://user:address@hidden/foobar?querystring
>                       http://user:address@hidden/foobar
>                       http://blabla3.com/foobar?querystring
>                       http://labla4.com/foobar?querystring
>                       http://user:address@hidden/foobar
>                       ....] # Space or newline separate url-entries
>               then <action>

And what should be the <action>?  As far as I understand,
presumably sending mail/SMS, not restarting apache or restoring
database and/or backup.

I may be wrong but a quick perl, ruby, python, heck -- shell
script to GET or lynx -dump the page(s) and some grepper would do
this particular trick.

At the same time I understand that it's quite a "basic" task, but
it's rather "antivirus"-like than monit-like to me: common
breakins have common signatures, and signature-based approach
would save more time and effort when some new canned
phpnuke-blowing script comes out.  Again, killing/restarting
processes helps nothing here.

So all in all, is monit the targeted system-level bulletproof
tool focusing on local processes or is it aiming to become a
distributed infrastructure wielding dozens of plugins?

I'd say that monit is monit, nagios is nagios, and defacement
detectors are just that -- unless there's a plugin that doesn't
buy everyone such a tool in C running as root.

Sorry for being verbose and kind of "anti"sh again, but I tend to
criticize the things I like most -- the rest isn't worth it :)

(...and I'm just an hour back from the meeting where I've
presented monit to fellow colleagues needing a system monitoring
solution in addition to application-level one they have...)

-- 
 ---- WBR, Michael Shigorin <address@hidden>
  ------ Linux.Kiev http://www.linux.kiev.ua/