[Orca-users] Orca and Large Systems

[Liston Bias]

On Sun, 29 Dec 2002, Mark Goldenberg wrote:

> The heavy load appears to be Orca processing the disk information from
> two SF15k domains and a SF4800.  When you have this many disk devices
> I'm not sure how useful the graphs are.  Can I simply comment out the
> disk sections of the orcallator.cfg file to shutdown this processing?
> Or can I turn it off on a system by system basis by editing the
> orcallator.se script?  Does anyone on this list have any suggestions?
> I would like to keep Orca as real time as I possible.  I'm currently
> trying to use the Orca output to assist with troubleshooting
> performance problems with a new project.

Commenting out the disk section should help a lot.  I have thousands of
luns connected to EMC drives throughout the environment and creating
graphs for those drives was little/no use for us.

I try not to reduce was orcallator.se collects as individual system
processes has little effect on the overall performance of orcallator.
Plus, I can also go back and graph history data on something if I have
collected the data.

If commenting out orcallator.cfg lines doesn't get you the performance you
are seeking then you may also want to look at running multiple orca
processes to produce your data.  You would first need to run orcallator
with all systems to get the correct html graphs... and then you could run
multiple orca processes (specifying subset of systems in find_files) with
the no-html option.

- Liston