On 6/17/07, Ritesh <[EMAIL PROTECTED]> wrote:
I am facing a strange problem of CPU spikes every 12-13 mins. This is
a sun fire V210 machine (with 2 processors) and solaris 8. (5.8
Generic_117350-26 sun4u sparc SUNW,Sun-Fire-V210)
I have oracle, tomcat and a client-server (c++) application with few
threads running on the system(this is a simple messaging application).
I tried monitoring with prstat -Lm 1 but that shows me every thread of
the application pretty much consuming consistent cpu (user and
system).
mpstat however do shows the spikes happening 12-13 mins. i am
unable to catch the culprit. i do see the prstat -m shows ICWX
(involuntary context switches) for application threads incrasing
during the spike and so does the latency (it goes from 0.0 to 4.5).
Sounds to me like you may have something that is kicking off a bunch
of short-lived processes. Perhaps "sar -c 1" during this period will
show an increase in forks and execs.
How do i catch the culprit. my oracle is pretty much supposed to be
idle, so does the tomcat/ java and any other process.
If my hunch is right, you could enable process accounting to see what
commands (or 8 characters of them) that are being executed. This may
be enough of a hint to help you figure out what is going on.
Being on Solaris 10 or later would make this a much easier problem to solve.
Mike
--
Mike Gerdts
http://mgerdts.blogspot.com/
_______________________________________________
perf-discuss mailing list
perf-discuss@opensolaris.org
