Very High CPU usage of OC4J Processes After Migrating To Sun T5240 Server with UltraSPARC T2 Plus Processors (Doc ID 1099127.1)

Last updated on FEBRUARY 24, 2017

Applies to:

Oracle Containers for J2EE - Version 10.1.2.0.0 to 10.1.3.5.0 [Release AS10gR2 to AS10gR3]
Sun SPARC Sun OS

Symptoms

When running JVM based processes on SUN UltraSPARC T1/T2 processors, the global performance of the JVM is very bad and CPU usage reported by the JVM appears to be abnormally high, while the server is free of resources.

Note that this is a generic issue that affects any kind of JVM based server, and not only OC4J containers.


In our specific case, we saw, on Solaris 10, that running Oracle BPEL Application on OAS 10gR2 with JDK HotSpot 1.4.2, the OC4J_BPEL Instance process monitoring showed following information:

The following commands were used to monitor OC4J

PID USERNAME SIZE RSS STATE PRI NICE TIME CPU PROCESS/NLWP
13421 appbpel 3468M 2417M cpu17 0 10 0:54:35 88% java/201
Here we have lightweight process 201 is using 88% CPU and corresponds to a ‘java’ process thread.
217 processes: 215 sleeping, 2 on cpu
CPU states: 10.9% idle, 28.0% user, 61.1% kernel, 0.0% iowait, 0.0% swap
user/kernel ratio is 1/2.

Changes

The problem appeared during the migration to Sun SPARC T5240 Server.
The target server is multiCPU and has 16 Virtual CPUs assigned to the Solaris Zone where OracleAS 10.1.2 runs.

Cause

Sign In with your My Oracle Support account

Don't have a My Oracle Support account? Click to get started

My Oracle Support provides customers with access to over a
Million Knowledge Articles and hundreds of Community platforms