Problem:
On Windows systems, after you configure the data collector for the WebSphere Applications agent, the application server cannot be started with Java core dump occurred. The following messages are reported in the native_stderr.log file of the application server:

Unhandled exception
Type=Segmentation error vmState=0x00000000

Solution:
This problem occurs because the JVM that is used by the application server is attempting to use the corrupt shared class cache. It might happen to the JVM that is prior to the following versions:

• 6.0.0 SR12
• 6.0.1 SR4
• 7.0.0 SR3

This is an identified problem by WebSphere Application Server. For more information about the APAR, see http://www-01.ibm.com/support/docview.wss?uid=swg1IV29776.

To solve the problem, use either of the following options:

• Upgrade the IBM Java SDK to a level with the APAR fix and then start the application server. For example, upgrade IBM Java 7.1 SR1 to 7.1 SR4.
• Disable the use of shared class cache for the application server. To do this, modify the Generic JVM Arguments of the application server to remove the -Xshareclasses option or to add the -Xshareclasses:none option after all existing -Xshareclasses options.