Eugene: I think I understand why your microstate values aren't adding up to 100%. The fact that you're running Solaris 9 has a lot to do with the problem.
Microstate accounting only updates its timestamps when a lwp transitions from one state to another. So, in your case, your lwp has been idle for a long time and none of its new idle time has been added into the value that is being sampled by prstat. If the lwp were to change state, the values would be updated and would probably add up to 100%. This has been fixed in Solaris 10, where microstate is enabled all of the time. The timestamps are still only updated when lwps transition from one state to another, but proc usage and lusage files have been updated so that they include the time spent in the current microstate when they are read. Now, every time prstat reads these files it gets the amount of time spent in the current microstate, so the values will always add up to 100%. I hope this helps, -j On Sat, Feb 10, 2007 at 01:01:07PM -0500, Eugene Margulis wrote: _______________________________________________ perf-discuss mailing list perf-discuss@opensolaris.org
