Additional tests on AIX

Wed Jul 8 08:40:33 EDT 2009

Hi Trustin,
No problem! I'm using the exact same code than before.

Sadly, it seems a bug comes back... but it is not stable...

Using JDK6 on AIX, I've got two kinds of running: (java is run without any
special argument)

A) When it is ok
- Netty server first says : (after 1 seconds or so)
8 juil. 2009 14:30:00 org.jboss.netty.channel.socket.nio.NioProviderMetadata
null
INFO: Using the autodetected NIO constraint level: 1 (Use better NIO
provider for better performance)

- Then when the client starts, the server says before exiting normally:
Channel connected
Msg received and will close channel
Close channel
Free main
Close child channel
End of server
Close main channel

I've got exactly the same when using JDK5 (and stable then).

B) When it is KO (only JDK6 from time to time, until now, each 1 on 3 runs)
- Netty server first says : (after 6 seconds or so)
8 juil. 2009 14:10:09 org.jboss.netty.channel.socket.nio.NioProviderMetadata
null
WARNING: NIO constraint level autodetection timed out.
8 juil. 2009 14:10:09 org.jboss.netty.channel.socket.nio.NioProviderMetadata
null
WARNING: Failed to autodetect the NIO constraint level; using the safest
level (2)

- Then when the client starts, the server says before it hangs:
Channel connected
Msg received and will close channel
Close channel
Free main
Close child channel
End of server
Close main channel

- If I try to launch a new client, the new client says:
Connect to server 3333
Exception in thread "main" java.net.ConnectException: Un hôte distant a
rejeté une tentative de connexion (connect).
        at sun.nio.ch.Net.connect(Native Method)
        at sun.nio.ch.SocketChannelImpl.connect(SocketChannelImpl.java:507)
        at socket.client.Client.main(Client.java:49)
So it cannot connect to the server and the server does not answear neither
it stops.

So a bug comes back, slightly different than previously since now if I
connect a second client later on after the server hangs, the server will not
stop however as before.

For me (if I'm not wrong), there is something that badly happens during the
NIO constraint level autodetection. If the autodetection is ok, then level
is 1 and the test is perfectly ok. If the autodetection hangs for any reason
(around 6 seconds), then level is 2 and the test is KO...

Here is the java dump when the server hangs:

 Full thread dump J9 VM (J2RE 6.0 IBM J9 2.4 AIX ppc64-64 build
jvmap6460sr4ifx-20090409_3325420090409_033254_BHdSMr, native threads):
      "JIT Compilation Thread" TID:0x0000000111CB1700,
j9thread_t:0x00000001101248A0, state:CW, prio=10
            (native thread ID:0x25C04D, native priority:0xB, native
policy:UNKNOWN)
      "Signal Dispatcher" TID:0x0000000111E4ED00,
j9thread_t:0x0000000111D94320, state:R, prio=5
            (native thread ID:0x15A039, native priority:0x5, native
policy:UNKNOWN)
          at com/ibm/misc/SignalDispatcher.waitForSignal(Native Method)
          at com/ibm/misc/SignalDispatcher.run(SignalDispatcher.java:54)
      "Gc Slave Thread" TID:0x000000011206BD00,
j9thread_t:0x0000000111F8FC40, state:CW, prio=5
            (native thread ID:0x26F0F7, native priority:0x5, native
policy:UNKNOWN)
      "NIO constraint level detector" TID:0x0000000112A97000,
j9thread_t:0x0000000112A8A060, state:B, prio=5
            (native thread ID:0x25A033, native priority:0x5, native
policy:UNKNOWN)
          at
sun/nio/ch/SelectionKeyImpl.nioInterestOps(SelectionKeyImpl.java:103)
          at
sun/nio/ch/SelectionKeyImpl.interestOps(SelectionKeyImpl.java:65)
          at
org/jboss/netty/channel/socket/nio/NioProviderMetadata$ConstraintLevelAutodetector.autodetectWithoutTimeout(NioProviderMetadata.java:341)
          at
org/jboss/netty/channel/socket/nio/NioProviderMetadata$ConstraintLevelAutodetector$1.run(NioProviderMetadata.java:251)
          at
org/jboss/netty/util/ThreadRenamingRunnable.run(ThreadRenamingRunnable.java:110)
          at java/lang/Thread.run(Thread.java:735)
      "pool-3-thread-1" TID:0x0000000112959000,
j9thread_t:0x0000000112A8A560, state:R, prio=5
            (native thread ID:0x2580DF, native priority:0x5, native
policy:UNKNOWN)
          at sun/nio/ch/PollArrayWrapper.poll0(Native Method)
          at sun/nio/ch/PollArrayWrapper.poll(PollArrayWrapper.java:116)
          at sun/nio/ch/PollSelectorImpl.doSelect(PollSelectorImpl.java:57)
          at sun/nio/ch/SelectorImpl.lockAndDoSelect(SelectorImpl.java:69)
          at sun/nio/ch/SelectorImpl.select(SelectorImpl.java:80)
          at
org/jboss/netty/channel/socket/nio/NioProviderMetadata$SelectorLoop.run(NioProviderMetadata.java:454)
          at
java/util/concurrent/ThreadPoolExecutor$Worker.runTask(ThreadPoolExecutor.java:896)
          at
java/util/concurrent/ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:918)
          at java/lang/Thread.run(Thread.java:735)
      "DestroyJavaVM helper thread" TID:0x0000000112A95A00,
j9thread_t:0x0000000110123EA0, state:CW, prio=5
            (native thread ID:0x2690EB, native priority:0x5, native
policy:UNKNOWN)

HTH,
Frederic

Trustin Lee wrote:
> 
> Hi Frederic,
> 
> I've just updated the NIO socket accept code which might affect the
> previous fix for AIX freeze.  Would you mind if you test if revision
> 1534 works for your AIX?
> 
> Thanks in advance!
> Trustin
> _______________________________________________
> netty-dev mailing list
> netty-dev at lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/netty-dev
> 
> 

-----
Hardware/Software Architect
-- 
View this message in context: http://n2.nabble.com/Additional-tests-on-AIX-tp3224966p3225198.html
Sent from the Netty Developer Group mailing list archive at Nabble.com.