Full ERROR Message Example:
ERROR [SharedPool-Worker-32] 2019-02-28 19:28:52,904 SEPWorker.java:143 - Unexpected exception killed worker
What does this ERROR message mean?
This error means that the SharedPool-Worker-* thread can't continue processing. It is usually an effect of some other problem. Shared Executor Pool (SEP) threads work on behalf of NTRs (Native Transport Requests) for client connection requests.
Why does this ERROR occur?
If the JVM hits an unrecoverable error (OOM, severe disk problem, etcetera.) DSE or Cassandra will kill the JVM by calling the JVMStabilityInspector class. Leading to the SEPWorker to be killed too and the Unexpected exception killed worker error.
You may see the error if the JVM can't create more threads, for example:
ERROR [SharedPool-Worker-32] 2019-02-28 19:28:52,904 SEPWorker.java:141 - Failed to execute task, unexpected exception killed worker: {} java.lang.OutOfMemoryError: unable to create new native thread
The above error could be due to user-process resource limits being too low, or an insufficient amount of memory available to create new threads. See:
https://docs.datastax.com/en/landing_page/doc/landing_page/recommendedSettings.html
How do you fix this ERROR?
The fix is dependent on why the JVM terminated. A complete stack trace and any earlier messages, particularly those involving JVMStabilityInspector, should be analysed to determine the root cause.
If the reason for the JVMStabilityInspector exception is determined to be an OutOfMemory situation, then a JVM heap dump may be required.
Datastax technical support will be able to advise.