|
Joseph Mack NA3T [jmack@xxxxxxxx] wrote:
>
> >and I want to distribute the load based on criteria such as:
> >- disk usage
> >- OS load average
> >- CPU usage
> >- custom hooks into my own software
> >- You Name It (TM)
> have you looked at feedbackd (see the HOWTO). It doesn't
> have all the hooks above, but when you get them, it will
> use the info.
Feedbackd has got CPU-monitoring only by default.
It also has a perl plugin that's supposed to let you code
something revelant to you quickly. That's a perfect idea except
original perl plugin is not fully functional because it breaks
some rules regarding linking to C-based modules.
I wrote a patch that's solves the problem (it's against
the latest release - 0.4).
>
> this is a bit of work. If you're doing this for fun that's
> one thing, but it someone has to pay for your time, how
> bad would your setup get if you used one of the standard
> schedulers? Would one realserver be twice as busy as the
> rest? Is this so bad? It's only in the case of persistence
> that realservers get much out of balance.
I guess that's highly application-dependent.
--
Michal Kurowski
perl -e '$_=q#: 13_2: 12/o{>: 8_4) (_4: 6/2^-2; 3;-2^\2: 5/7\_/\7: 12m m::#;
y#:#\n#;s#(\D)(\d+)#$1x$2#ge;print'
|
