Andrew Edunov wrote:
>> I've seen this behaviour before on UNIX boxes.
>> It can happen when the CPU('s) are so heavily saturated that the OS is
>> spending
>> a lot of time doing scheduling and process management instead of letting
>> the user
>> processes do the real work.
>>
>> There is however also the possibility that your user processes initiates
>> some kernel
>> level operations.
>>
>> Have you taken a look at top while the load is as described above?
>>  
>
>
> When I start top I saw only oracle process.
>
> Best regards,
>  Andrew.
>
>  
Ok, lets see if it is the oracle processes are consuming all the CPU.

Two ways of doing this (without requiring the diagnostics pack) is
either statspack or
a select from v$sysstat.
The latter is probably the quickest:

From SQL*Plus (logged in as sysdba) do:

set time on
select name, value from v$sysstat where statistic#=12;

then wait a minute or two

then reexecute last sql:

/

By subtracting the CPU stats from the two runs you'll get the number
of centiseconds CPU usage between the two measurements.

Divide this difference with 100 * "No of CPU's" and compare this with
the elapsed
time.

If this comparision shows that all of the consumed CPU (user+os) is used
within Oracle
then there is reason to believe that the os-part of the CPU-consumption
is caused
by kernel-level calls from Oracle.

If the comparision only shows 2/3 of the total CPU consumption there is
reason to
believe that the sar output is showing you a server that is highly
CPU-saturated.

--
Best regards

Martin Berg

email: martin@(protected)
web: www.berg-consult.com



--
To unsubscribe, email: suse-oracle-unsubscribe@(protected)
For additional commands, email: suse-oracle-help@(protected)
Please see http://www.suse.com/oracle/ before posting