SUMMARY: Unusual slowness on a SB100 w/Solaris 8.

From: Andres Rojas <arojasm_at_adexus.cl>
Date: Wed Nov 13 2002 - 13:24:51 EST
Dear Managers,

Thanks for your promptest replies.

>>> It happened to be a problem with NIS+ which is still under
investigation. <<<

------------------------------------------------

Special thanks must go to these people. I included their recommendations
and my comments below with "***" if anyone cares to take a look at it.

Hindley Nick - suspect that you may have a loop in your filesystem? run
fsck on it.

*** The filesystems were checked and determined as sane.

Soeren Schaper - I had the same problem on an V880. The only solution I
found was a reboot.With patches on a Blade 100 I had some weird
experiences. My Blade does not run with solaris8 02/02 or solaris8 07/01
and recommended patches while all others in our company have no problems.

*** Yes, rebooting was a termporary solution until the SB began with
slowness again. Would it be a NIS+ problem?

 Julie Peers - I know of problems with ShowMeTV on Sunblades. Do you have
this running? Check in /var/adm/messages as you may have errors in there.
Check for 111179-04 and ensure flash-prom up to date!!

*** No, the only application is Apache Web Server. The eeprom is up todate
and /var/adm/messages contains warnings for apache and ntp only.

Mike - df -k to see if any diskspace is low or out, ie. /var or /tmp, etc.
use iostat, vmstat, prstat or top to see if any other processes are running
that's taking too much memory/cpu/disk, etc.

*** Absolutely agree. iostat shown disk usage equals to zero during the
slowness, vmstat shown more than 2GB ram available and mpstat shown between
97% and 100% idle CPU. Did I mention it was a weird behavior?

Galen Johnson - I've found that whenever I have a broken mount, ps, df, and
ls (and any other command that views the filesystems) will hang on me.
Check your mounts.  Of course, there is always the possibilty this system
has been
compromised and non-standard ps binaries have been put in place but unless
you have good reason to believe this may be the case it's an outside
possibility.

*** Good to know that df would hang if there's anything broken at
filesystem level.

John Douglass - Sounds like an NFS problem.  Check to make sure all your
mounts are responding properly.  For instance does a "df -kl" work??  My
guess is that it will.  Your cron jobs are probably running from the NFS
mounted area which is why crons were hanging around.

*** NFS filesystem problems this time. Good to know also.

- Andris
____________________________________________________
Andris Rojas M.                       Adexus S.A.
Ing. Soporte Sun                      Miraflores 383, piso 3
Tel: +56 2 355-1348                   http://www.adexus.cl
Fax: +56 2 355-1005                   http://www.sun.com
Central de Servicios: +56 2 355-1850  mailto:arojasm@adexus.cl
Mobile: +56 9 244-0524
_______________________________________________
sunmanagers mailing list
sunmanagers@sunmanagers.org
http://www.sunmanagers.org/mailman/listinfo/sunmanagers
Received on Wed Nov 13 13:30:06 2002

This archive was generated by hypermail 2.1.8 : Thu Mar 03 2016 - 06:42:57 EST