Spann, Elizebeth (Betsie) wrote:
Hi,
How is the st value displayed in the 2.6 kernel version of vmstat
computed in general terms?   I've seen it described as time stolen from
a virtual machine.

Steal time is the percentage of time a virtual CPU waits for a real CPU
while the hypervisor (in this case, z/VM) is servicing another virtual
processor.

Are there differences between the value on a RHEL 5.0 system vs a RHEL
5.2 system?  I've seen references to it being "fixed".   ??

The RHEL5.x series (and SLES10 onwards) should all behave in the same
manner.  This wasn't a Red Hat vs Novell fix, it was a feature added to
the kernel which we both incorporated into our distributions.

Previously the kernel would use tick-based CPU accounting (versus
virtual CPU time).  The numbers based on tick-based CPU time accounting
report CPU times spent in a /virtual/ CPU context, versus times spent in
a /real/ CPU context.

Is the vmstat value similar to the mpstat value?
Yes, but the accuracy is different.  mpstat will extend into the 100th
of a percent.

Example:
# mpstat
%sys 4.17
%idle 68.21

#vmstat
sys 4
id 68

Do note that vmstat does *not* count itself as a running process

-Shawn

----------------------------------------------------------------------
For LINUX-390 subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO LINUX-390 or visit
http://www.marist.edu/htbin/wlvindex?LINUX-390

Reply via email to