On 10/20/06, Brian Kolaci <[EMAIL PROTECTED]> wrote:
I think it is a bug that the load of zone A affects the reported load
in zone B thereby causing a process in zone B (sendmail) to make bad
decisions that are based on perceived load average.
I have always been a little unclear on _how_ the load
_statistic_ is managed by the system. There doesn't seen to be a tool
to track it over time (other than things like SunMC). It doesn't
(directly) appear in sar data, not vmstat output.
I assume the reason there is not a 'per non-global zone' load
value, is that there is really only one kernal with one scheduler and
one run queue. I think load is defined as the number of runnable and
running processes at any given point in time. I assume each process is
tagged with the zone it belongs to (zoneid 0 being the global zone),
in which, would it be possible to create a load value for each zoneid
?
Please fogive my ignorance here, I am an admin, not a developer :-)
--
Paul Kraus
_______________________________________________
zones-discuss mailing list
zones-discuss@opensolaris.org
