>       Is it required to use the update method in libkstat to gather
> correct data?  Is it possible to just call a specific kstat_read every
> time I need to access statistics?

I haven't used the C interface in a while, but my assumption is no.  You
only need update to correct your chain, not to read data.

>       Now I understand what the update method does (refreshes private
> kstat data with updated kstat data), but do I really need it?  I ask
> this 2nd question because when running through the perl script
> sometimes reports 100%+ idle time up to 105% idle time, which is quite
> unlikely. 

Without seeing what it's doing, I would assume that the perl script is
doing sloppy math with the kstat data that is presented to it.

> The machine is a 6900 with a cpu count of 24.  If I run the
> same perl script against a couple of the cpus then it reports back
> seemingly correct data.  The script runs through 2 main loops.  The
> first loop (loop_1) being a simple counter.  Inside loop_1 is a loop
> that prints out statistics for each processor.  At the end of loop_1 I
> call the update function, and then does a "sleep 1;".  This reports an
> idle time in excess of 100 sometimes.  Is this a possible bug?

Possible, but without more information, I'd assume it to be in the
script.

> anyone duplicate this?  I can duplicate this on any system with a cpu
> count of 2 or higher and with an uptime of 124 days or greater running
> Solaris 10 (sparc).

I've got one that's within about 15 days of getting there.  Can you give
any more details about what the perl script is doing (algorithm or
code)?

-- 
Darren Dunham                                           [EMAIL PROTECTED]
Senior Technical Consultant         TAOS            http://www.taos.com/
Got some Dr Pepper?                           San Francisco, CA bay area
         < This line left intentionally blank to confuse you. >
_______________________________________________
Solaris-Users mailing list
[email protected]
http://node1.filibeto.org/mailman/listinfo/solaris-users

Reply via email to