[Orca-users] Re: Question on changing probe interval in orcallator.se butleaving RRD alone.

[Blair Zajac]

Sean O'Neill wrote:
> 
> I know this question has been asked already but "wading" through all the
> email back history ... well ... I'm lazy.
> 
> The orcallator.se script "sleep" interval can easily be modified to
> anything.  I have a need for a time interval of 1 minute - damn developers ;>
> 
> Anyway, my question is on the RRD/ORCA Graphing side of the house.  Let's
> just say somebody wants to see the graphs.  I've seen various posts that
> there will be blood, wailing, and gnashing of teeth if someone tries to
> mess around with what RRD and the "graphing half" of ORCA understand the
> time interval to be.
> 
> What will happen if I feed 1 minute time interval data into "ORCA's
> other"/RRD which is expecting 5 minute data?  Will it just average the data
> up or will I get crap ?

Yes, it'll handle it just fine and average it.

> 
> I posted this to the cricket-users group by accident but I did get an answer:
> 
> -----
> No clue about Orca, but RRDtool will just average the samples (weighted
> according to their dispersion over time).
> 
> For fast ethernet devices, I routinely feed 1 minute interval data into
> Cricket with default rrd settings, to avoid counter wrap.
> -----
> 
> I assume the same is true for ORCA considering it uses RRD just like Cricket ?

Yes.  The only difference may be how RRD's RRAs (Round Robin Archives) consolidate
the data.  The RRDs created by Orca by default average the data, although using
the maximum or minimum values are also available.

Best,
Blair

-- 
Blair Zajac <blair at orcaware.com>
Web and OS performance plots - http://www.orcaware.com/orca/