Summary: Performance tuning on a SPARC V running Solaris 2.5

Mohammed Ali (ali@platinum.com)
Mon, 03 Mar 1997 10:08:40 -0600 (CST)

Hi,

Here's the summary of all the responses I got. My Original Question was:

I have a SPARC V with 160 MB RAM running solaris 2.5 with all the
recommended patches loaded. The system is very,very slow.
The vmstat shows:

procs memory page disk faults cpu
r b w swap free re mf pi po fr de sr f0 s0 s1 s2 in sy cs us sy id
5 1 0 154312 70352 0 269 20 0 0 0 0 0 43 44 9 564 1464 326 26 64 10

Do I need to tune some kernel parameters, increase the buffer cache size
etc or anything else ?

Any suggestions/recommendations ?

Following are the responses I got, I hope I have not missed anyone:
My thanks to all the friends who took time to answer my question.

brion@dia.state.ma.us wrote:
---------------------------
Find out why find is running. Then kill it. If it is run automaticaly,
find out how
to control when it is run and fix it.

If system is still slow, you probably have too many users.

Please mail back if I can help further.

<Kevin.Sheehan@uniq.com.au> wrote:
---------------------------------
system time that high is generally a worry, but the bottom line is
that this is not enough information to tell. WAG would be that
you have a paging problem with an app, or need to tune your disks.

Get the Solaris Performance and Tuning book by Adrian Cockcroft.

l & h,
kev

peterb@uniq.com.au wrote:
-------------------------
The only useful data in this output is the run queue. You've got
5 processes which are runnable on a single CPU machine. However,
the cpu appears to be about 10% idle. So, the amount of information
you've provided isn't anywhere near enough to pinpoint a slowness
problem.

We would need to see a section of vmstat output to see what's
happening. Also, output from 'iostat -xc'. In order to look at
buffer cache hits, you'll have to run 'sar -b'. However, this
shouldn't critically affect the performance of the machine.

Read the manual pages on vmstat, iostat and sar.