[Orca-dev] Re: [Orca-checkins] r395 - trunk/orca/data_gatherers/orcallator

Blair Zajac blair at orcaware.com
Tue Oct 5 10:56:34 PDT 2004


dmberezin at hotmail.com wrote:
> Author: dmberezin at hotmail.com
> Date: Tue Oct  5 08:18:50 2004
> New Revision: 395
> 
> Modified:
>    trunk/orca/data_gatherers/orcallator/orcallator.se
> Log:
> More changes to disk/tape data collection
> 
> * data_gatherers/orcallator/orcallator.se
>   Add check: define USE_KSTAT_IO only if NO_KSTAT_IO is not defined. This
>   allows disabling KSTAT_IO code with command line option.
>   Separate disk and tape statistics gathering. Add
>     define WATCH_TAPE
>     function measure_tape
>   We can only gather tape performance data with KSTAT_IO code at this time,
>   because of the way live_rules.se checks for disk slices to ignore them. It
>   simply checks if there is an 's' in the long_name, which is true for st
>   devices, so they are ignored. We could fix live_rules.se or combine
>   diskinfo.se and tapeinfo.se together to have correct long_name (rmt/xx)
>   appear when live_rules.se accesses it, but perhaps using KSTAT_IO for this
>   is sufficient.

Dmitry,

If you want to have modified versions of the SE .se files, we can do 
that.  Take a look in the lib/SE directory where we keep patched 
versions of some these files.  In fact, there's already a live_rules.se 
in there.  Then when start_orcallator starts orcallator.se, it passes 
-I$libdir/SE/$SE_VERSION on the command line to pick up these versions 
before the standard SE files.

Regards,
Blair

-- 
Blair Zajac <blair at orcaware.com>
Plots of your system's performance - http://www.orcaware.com/orca/



More information about the Orca-dev mailing list