[Orca-users] Re: issues with large floats and the 'U' value

[Blair Zajac]

"Jason D. Kelleher" wrote:
> 
> Anyone having problems with large floating point numbers and Orca?  I
> have a counter that's now a 13 digit float and my graphs all became 0
> for the value.  I don't think it's a Perl issue:
> 
> [mktmon at barbaros]:/opt/perf/data/dw-ibm/sp/stats/worf $ zcat
> stats.2001-08-29-9999.Z | cut -d ' ' -f1,13| head -5
> timestamp       TOTAL_disk_rK/s
> 999061200       1015910202326.000
> 999061500       1015916647851.000
> 999061800       1015921386649.000
> 999062100       1015940506358.000
> [mktmon at barbaros]:/opt/perf/data/dw-ibm/sp/stats/worf $ perl -e
> 'print 1015940506358.000 - 1015921386649.000, "\n";'
> 19119709
> [mktmon at barbaros]:/opt/perf/data/dw-ibm/sp/stats/worf $
> 
> Also, I switched to 0.27b2 to be able to feed 'U' into Orca for
> unknown values, but I still get the following warning:
> 
> Argument "U" isn't numeric in subtraction (-) at (eval 22) line 3,
> <FD> line 210.
> 
> I'm going through SourceFile.pm now...
> 
> Regards,
> 
> jason
> 

Jason,

It could be an issue with RRDtool, since it does the differencing between
the old and new values.

You may want to try running rrddump on your generated RRD files to see what
values are stored in there.

Regarding the 'U', this will work if the only expression in the 'data'
statement is value being measured.  Orca doesn't check if the value is 'U'
and then decide if its ok to do math.

So something like

	data		usr%

will work but something like

	data		100 - usr%

won't.  For those data's that you want to use U on, you may want to do
something like

	data		usr% ne 'U' ? (100 - usr%) : 'U'

Obviously, these expressions get complicated if there are multiple
measurements being used in a single data.

Best,
Blair

-- 
Blair Zajac <blair at orcaware.com> - Perl & sysadmin services for hire
OS & web analysis - http://www.orcaware.com/orca/