[CXF-2606] ThreadGroup in the default workqueue gets prematurely destroyed

• Type: Improvement
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 2.2.2
• Fix Version/s: 2.1.9, 2.2.6
• Component/s: Core

Environment

IBM JDK6

• Assignee: Daniel Kulp
• Reporter: Hadrian Zbarcea
• Created: 08/Jan/10 03:41
• Updated: 01/Feb/10 19:02
• Resolved: 08/Jan/10 14:25

Description

An IllegalThreadStateException gets thrown as shown in the stack below when a new Thread is added to the default workqueue. This does not happen with the Sun JDK and it was not encountered with the IBM JDK5 either.

The Exception gets thrown because the ThreadGroup is destroyed by the time we add the new thread because the core pool size (low watermark) is 0, the number of threads is 0 and is a daemon thread.

Throwable occurred: java.lang.IllegalThreadStateException
        at java.lang.ThreadGroup.checkNewThread(ThreadGroup.java:147)
        at java.lang.Thread.initialize(Thread.java:332)
        at java.lang.Thread.<init>(Thread.java:267)
        at java.lang.Thread.<init>(Thread.java:225)
        at org.apache.cxf.workqueue.AutomaticWorkQueueImpl$AWQThreadFactory.newThread(AutomaticWorkQueueImpl.java:162)
        at java.util.concurrent.ThreadPoolExecutor.addThread(ThreadPoolExecutor.java:672)
        at java.util.concurrent.ThreadPoolExecutor.addIfUnderCorePoolSize(ThreadPoolExecutor.java:697)
        at java.util.concurrent.ThreadPoolExecutor.execute(ThreadPoolExecutor.java:652)
        at org.apache.cxf.workqueue.AutomaticWorkQueueImpl.execute(AutomaticWorkQueueImpl.java:249)
        at org.apache.cxf.interceptor.OneWayProcessorInterceptor.handleMessage(OneWayProcessorInterceptor.java:83)
        at org.apache.cxf.phase.PhaseInterceptorChain.doIntercept(PhaseInterceptorChain.java:236)
        at org.apache.cxf.transport.ChainInitiationObserver.onMessage(ChainInitiationObserver.java:104)

A workaround is to set the lowWaterMark to something >=1 so that the ThreadGroup never gets destroyed.