Re: [Wrapper-user] Wrapper kills the application unexpectedly

[Leif M. <le...@ta...>]

Erik,
The Wrapper is receiving a SIGCHLD signal from the OS and interpreting
it as the Java process having stopped.  From the log however, it
appears that JVM still exists as it continues on to shut it down.

Can you tell me a little about your application?   Does your Java
application run any child processes of its own?  I am wondering if
such a grandchild process exiting might be causing this.  I am going
to need to go back and run some more tests if that is the case.

As a note, when you query the status, it is reporting as running
because the Wrapper process itself is still running.  At the end of
the log, it appears the Wrapper is just about to launch a second JVM
invocation to recover from this problem.

Cheers,
Leif

On Fri, Mar 13, 2009 at 12:54 AM, Erik Drolshammer <eri...@fj...> wrote:
> Hi!
> Some time after starting the application (20min perhaps) the wrapper
> seems to invent a problem and kills the jvm. See debug log:
>
> INFO   | jvm 1    | 2009/03/12 16:28:24 | Send a packet PING : ok
> DEBUG  | wrapperp | 2009/03/12 16:28:24 | read a packet PING : ok
> DEBUG  | wrapper  | 2009/03/12 16:28:24 | Got ping response from JVM
> ERROR  | wrapper  | 2009/03/12 16:39:26 | JVM exited unexpectedly.
> DEBUG  | wrapper  | 2009/03/12 16:39:26 | Signal trapped.  Details:
> DEBUG  | wrapper  | 2009/03/12 16:39:26 |   signal number=17 (SIGCHLD),
> source="unknown"
> DEBUG  | wrapper  | 2009/03/12 16:39:26 | Received SIGCHLD, checking JVM
> process status.
> DEBUG  | wrapper  | 2009/03/12 16:39:26 | JVM process exited with a code
> of 1, setting the wrapper exit code to 1.
> INFO   | jvm 1    | 2009/03/12 16:39:26 | Read Timed out. (Last Ping was
> 292200 milliseconds ago)
> INFO   | jvm 1    | 2009/03/12 16:39:26 | Read Timed out. (Last Ping was
> 585600 milliseconds ago)
> INFO   | jvm 1    | 2009/03/12 16:39:26 | Wrapper Manager: JVM did not
> exit.  Give up.
> INFO   | jvm 1    | 2009/03/12 16:39:26 | Wrapper Manager: ShutdownHook
> started
> INFO   | jvm 1    | 2009/03/12 16:39:26 | WrapperManager.stop(0) called
> by thread: Wrapper-Shutdown-Hook
> INFO   | jvm 1    | 2009/03/12 16:39:26 | Send a packet STOP : 0
> INFO   | jvm 1    | 2009/03/12 16:39:26 | Thread, Wrapper-Shutdown-Hook,
> handling the shutdown process.
> INFO   | jvm 1    | 2009/03/12 16:39:26 | calling listener.stop()
> INFO   | jvm 1    | 2009/03/12 16:39:26 | WrapperSimpleApp: stop(0)
> INFO   | jvm 1    | 2009/03/12 16:39:26 | returned from listener.stop() -> 0
> INFO   | jvm 1    | 2009/03/12 16:39:26 | shutdownJVM(0)
> Thread:Wrapper-Shutdown-Hook
> INFO   | jvm 1    | 2009/03/12 16:39:26 | Send a packet STOPPED : 0
> INFO   | jvm 1    | 2009/03/12 16:39:26 | Closing socket.
> INFO   | jvm 1    | 2009/03/12 16:39:26 | Wrapper Manager: ShutdownHook
> complete
> DEBUG  | wrapperp | 2009/03/12 16:39:26 | server listening on port 32000.
> INFO   | wrapper  | 2009/03/12 16:39:26 | Wrapper Process has not
> received any CPU time for 144 seconds.  Extending timeouts.
> DEBUG  | wrapper  | 2009/03/12 16:39:26 | Waiting 5 seconds before
> launching another JVM.
>
> Any idea what causes this?
>
> If I call status I get the response that the application is running, but
> my application is surely stopped. However, a wrapper process seems to be
> alive.
>
>
> I have tried extending the timeouts, but this does not seem to effect
> the problem.
> wrapper.ping.timeout=300
> wrapper.startup.timeout=300
>
>
>
> System:
> Linux 2.6.18-92.el5 #1 SMP i686 i686 i386 GNU/Linux
> Wrapper: 3.2.3 with WrapperSimpleApp Integration
> java version "1.6.0_10"
> Java(TM) SE Runtime Environment (build 1.6.0_10-b33)
> Java HotSpot(TM) Server VM (build 11.0-b15, mixed mode)
>
> --
> Best regards,
>
> Erik Drolshammer