Print

[row_perfect]

Hi,

I'm getting an odd EOFException in the driver on task completion; the result of which is that the single task submitted gets passed to a new node, the original still shows as executing and this repeats until all nodes appear busy and the grid is stuck. Killing the client frees everything up again. As far as I can tell, there are no Exceptions in the client.


I've trapped it in the debugger grabbed some of the variables:

Thread [NodeClassServer-thread-30] (Suspended (exception EOFException))   
   owns: SelectionKeyWrapper  (id=241)   
   ChannelInputSource.read(ByteBuffer) line: 91   
   ChannelInputSource.read(byte[], int, int) line: 61   
   MultipleBuffersLocation.nonBlockingTransferFrom(InputSource) line: 161   
   MultipleBuffersLocation.transferFrom(InputSource, boolean) line: 117   
   PlainNioObject.read() line: 94   
   BaseNioMessage(AbstractNioMessage).readNextObject(ChannelWrapper<?>) line: 145   
   BaseNioMessage(AbstractNioMessage).read(ChannelWrapper<?>) line: 104   
   ClassContext(SimpleNioContext<S>).readMessage(ChannelWrapper<?>) line: 49   
   WaitingNodeRequestState.performTransition(ChannelWrapper<?>) line: 70   
   WaitingNodeRequestState.performTransition(ChannelWrapper) line: 35   
   StateTransitionTask<S,T>.run() line: 85   
   Executors$RunnableAdapter<T>.call() line: 471   
   FutureTask$Sync.innerRun() line: 334   
   FutureTask<V>.run() line: 166   
   ThreadPoolExecutor.runWorker(ThreadPoolExecutor$Worker) line: 1110   
   ThreadPoolExecutor$Worker.run() line: 603   
   Thread.run() line: 722   


ChannelInputSource.read(ByteBuffer) line: 91   

this   ChannelInputSource  (id=372)   
   channel   SocketChannelImpl  (id=398)   
      blocking   false   
      closeLock   Object  (id=400)   
      fd   FileDescriptor  (id=401)   
      fdVal   162   
      interrupted   null   
      interruptor   AbstractInterruptibleChannel$1  (id=402)   
      isInputOpen   true   
      isOutputOpen   true   
      keyCount   1   
      keyLock   Object  (id=403)   
      keys   SelectionKey[3]  (id=405)   
      localAddress   InetSocketAddress  (id=406)   
      open   true   
      provider   EPollSelectorProvider  (id=312)   
      readerThread   0   
      readLock   Object  (id=407)   
      readyToConnect   false   
      regLock   Object  (id=408)   
      remoteAddress   InetSocketAddress  (id=409)   
      socket   SocketAdaptor  (id=410)   
      state   2   
      stateLock   Object  (id=412)   
      writeLock   Object  (id=413)   
      writerThread   0   
n   -1   
remaining   4   
count   0   
data   HeapByteBuffer  (id=384)   
      java.nio.HeapByteBuffer[pos=0 lim=4 cap=4]
tmpBuffer   DirectByteBuffer  (id=386)   


Paul

[lolo]

Hi Paul,

Sorry for the late answer.
From the stack, the EOFException most likely means that the network connection with the node was interrupted somehow, on the log node side.
Is there any chance you could look at the node's and see if there is an exception in it?
It seems to be happening while the node is sending a class or resource loading request.

Thanks,
-Laurent