[collectd] collectd-nagios problems with ds queries

[Mariusz Gronczewski]

2011/8/8 Ulrich Habel <rhaen at cpan.org>:
>> It's still one *type* of value per file ;)
>> http://collectd.org/faq.shtml section "Why is the CPU usage split up
>> in so many files? " :)
>
> Thanks, I've already noticed :-). I didn't know that the df stats
> changed recently :-).
>
> It's perfectly reasonable to change the way collectd works and stores
> its data. However, from what I understand for now is that
> collectd-nagios is just not working anymore for some plugins e.g. the
> df plugin. This leads to the next consequence - use collectd only for
> data collection (which it was meant for) - for monitoring purposes use
> nagios (which it was meant for). Don't try to combine them with
> collectd-nagios, write a different tool which queries the data and do
> the math.
>
> A different solution might me: use nagios to monitor and collect data
> about e.g. df stats for long time monitoring (e.g. when is my disk
> full) and don't use the df plugin in collectd  (and maybe others) at
> all.
>
> I am still a big fan of the collectd data collector system with
> proxy/multicast, it's awesome. It would just be nice to use a tool to
> query all the data in one place and to use ranges for monitoring.
>
Well tbh what would be needed is just better version of
collectd-nagios that can do simple math like "if ( (df-free/df-total)
< 0.2 ){alert();}"
It wouldn't be very hard to write just i guess noone needed it that
badly ;]. I've wrote something similiar (only much simpler, i needed
only average/min/max of few values) and interfacing with collectd is
pretty easy.

PS. add list address in Cc: so other ppl can see/discuss about ideas ;)