On Thu, 2004-03-18 at 17:58, suppressed wrote:
> IMHO it is generally a bad idea to make a monitoring system responsible for
> carrying out actions.
>
> If a proc fails then
> 1- Fix the bug that made it fail
> 2 - Use OS facilities to restart it if it fails
> 3 - Write a wrapper script around it that will automatically restart it
> without relying on anything else
I completely agree with you, Pat. That's actually the reason why there
is (still) no support for uxmon taking actions on failures, actively.
On the other hand Big Sister is a generally available monitoring
software, other monitoring software actually often support things like
that, and finally, it's probably better to leave policy issues to the
admin rather than to the developer.
In other words: I hate monitoring software taking countermeasures
against failures, but if you (Kai) file a feature request I could
imagine implementing something like ...
localhost proc=sshd action=red:restart(ssh) procs
localhost partition=/ action=red:rm-rf(/) disk
:-)))
... and leave it to the admin to do something useful with it.
Hmh, I'll have to write a *very*explicit* disclaimer :-)
Best regards,
Tom
----------------------------------------------------------------------------
Thomas Aeby, Kirchweg 40, 1735 Giffers, Switzerland, Voice : (+41)26
4180040
Internet: suppressed PGP public key
available
----------------------------------------------------------------------------
Now available: MS Visual EDLIN for Windows
Attachment:
signature.asc
Description: This is a digitally signed message part
Mail converted by mhonarc 2.6.15
This archive provided courtesy of JSW4.NET, Internet Hosting Services for Small Business.