Crashes

The worst case problem from an SAP JVM point of view is an actual VM crash. Although the VM crash should be detected and the server node will be restarted, it indicates a serious problem.

Putting it simple, there are two possible reasons for a VM crash: an out-of-memory situation that cannot be handled gracefully, or a VM bug (although some other causes are imaginable, for example hardware issues etc.).

To render further analysis possible, the SAP JVM will write an extensive protocol about its current state and the last actions performed before the crash. This protocol can be found in a file called hs_err_pid<process id>.log in the working directory. Although the content of the protocol can be very technical and hard to understand without expert knowledge, it can help to recognize out-of-memory situations or give hints about the crash cause to non-expert users as well. In any case, if you want to report the problem to SAP, it will certainly help to provide the crash protocol along with your message.