On 13/04/2016 12:11, Josep M Beleta wrote: > I was wrong in my last message. One of the threads is different, is is not > related to AJP. This thread is: > > "NioBlockingSelector.BlockPoller-2" #15 daemon prio=5 os_prio=0 > tid=0x000000001ce0c800 nid=0x19b8 runnable [0x000000001dfff000] > java.lang.Thread.State: RUNNABLE > at sun.nio.ch.WindowsSelectorImpl$SubSelector.poll0(Native Method) > at sun.nio.ch.WindowsSelectorImpl$SubSelector.poll(Unknown Source) > at sun.nio.ch.WindowsSelectorImpl$SubSelector.access$400(Unknown Source) > at sun.nio.ch.WindowsSelectorImpl.doSelect(Unknown Source) > at sun.nio.ch.SelectorImpl.lockAndDoSelect(Unknown Source) > - locked <0x00000006c016abd0> (a sun.nio.ch.Util$2) > - locked <0x00000006c016abc0> (a java.util.Collections$UnmodifiableSet) > - locked <0x00000006c016aa50> (a sun.nio.ch.WindowsSelectorImpl) > at sun.nio.ch.SelectorImpl.select(Unknown Source) > at > org.apache.tomcat.util.net.NioBlockingSelector$BlockPoller.run(NioBlockingSelector.java:342) > > Locked ownable synchronizers: > - None > > All they have in common is that all them are running inside > sun.nio.ch.WindowsSelectorImpl$SubSelector.poll0.
Isolating it to the poller threads is very helpful. > Could be a Java problem? Maybe. A Tomcat bug is more likely at this point. > Could I find a workaround? We need to figure out the root cause first... > What could have been the cause that triggered that behavior? Hard to say. Maybe some edge case error condition that puts things into an unexpected state. It sounds like the Poller is in a tight loop. The next thing to figure out is what code path it is following and why. Are you able to configure a test instance to allow remote debugging? If you can, wait for the problem to occur, set a break point on NioBlockingSelector.java:342, and attach the debugger. Check that the poller thread keeps coming back to this point. If it does, step through the code and record the path followed. Alternatively, if you find a way to recreate the issue, other people on this list will be able to help investigate. Mark --------------------------------------------------------------------- To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org For additional commands, e-mail: users-h...@tomcat.apache.org