[Orca-users] Orca Data Entry Question

[Blair Zajac]

Adam Levin wrote:
> 
> On Wed, 13 Nov 2002, Scott A. Manicke wrote:
> > I have a parameter collected file that is in following format:
> ...
> > Note the multiple "INSTANCES" on separate lines.  Is there a way to load
> > this data and plot it on the same plot in Orca, or do I need to build a
> > script to manipulate the date into separate columns.
> 
> You need to build a script to make it column-based instead of row-based.
> 
> > The only issue I
> > have will doing that is I don't always get a data point from both CPUs
> > for example on the same exact epoch time stamp.  That would make
> > building a script very difficult, and I don't want to substitute a zero
> > for a missing data point as well.
> 
> How about splitting into two files when you process it?  You can have one
> file for CPU_0 and one for CPU_1, and that way the differences in time
> stamp won't throw Orca off.  You won't be able to graph both CPUs on the
> same graph that way, though (I don't think...).

Agreed.  Right now Orca can't handle data from different sources in
lines.  You could have something like this:

1037017986 host1 CPU_0 ProcessorTimePercent 27.1692 host1 CPU_1 ProcessorTimePercent 10.359

and because it's consistent format for each line and from the same sources,
then that would work.  However, this isn't scalable if you had different
numbers of CPUs on different boxes.

The best way is to generate separate files for each CPU.

Best,
Blair

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