Performance problem under Solaris

Hi everybody,
     My problem is that my disk service time is still
high (avg 100ms). Is that normal?

My HW config:
               Sun Ultra 1, 140 Mhz Cpu with 256 MB of Ram.
               1 x 2 GB external FSCSI disk for access.log and cache.log

               (I did deactivate the store.log)
               1 x 4 GB external FW-SCSI disk for the cache_dir

       Those are my F/S:

       c0t2d0s7 1952573 598882 1295116 32% .../squid/logs
       c1t0d0s7 3939604 3218839 602577 85% .../squid/cache1

My SW config:

         Solaris 2.5.1 as the O/S.
         That server is dedicated to the cache service. I'm now running
          Squid 1.1.17.

           Some of the values in squid.conf:
                     cache_mem 32
                     cache_swap 3800
                     quick_abort 10 80 30

What I'd made so far:
             I made a small script to run "sync" every 5
             seconds. I also got rid of the store.log and
             changed the quick_abort back to its default.
              I'm just asking myself if I should use the
              gnu malloc().

Any help will be appreciate.

Thanks!

Louis Beaulieu
Network O/S & Networking Analyst (Internet)
Analyste S/E reseaux & Reseautique (Internet)
Groupe Quebectel (Solutions Integrees)
louis.beaulieu@globetrotter.net
Received on Thu Jan 29 1998 - 11:09:52 MST