Re: [Wrapper-user] wrapperKillProcess()

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

Richard,
    When are you seeing this message?  Does it happen shortly after your 
application
was launched (within 30 seconds) or after your app has been running for 
a while?

    The first message "JVM appears hung: Timed out waiting for signal 
from JVM."
is displayed when the Wrapper process first decides that there is a 
problem with the
JVM.   It then attempts to ask the JVM to shut itself down cleanly by 
sending it a
stop command.  This gives the JVM a chance to do any cleanup and shut itself
down cleanly.

    If the JVM fails to shutdown on its own then the Wrapper will give 
up and
forcibly kill it with a "kill -9".  That is the message that you are seeing:
"JVM did not exit on request, terminated "  It is printed right after 
the kill
command because at that point, the JVM has failed to exit.   The kill 
command
will always work because it is killing a child process.

    I assume that you digging into the code because you are having a 
problem.
could you describe it?

Cheers,
Leif

Richard Emberson wrote:

> I get the following in wrapper logs on linux (not windows or solaris):
>
> JVM appears hung: Timed out waiting for signal from JVM.
> JVM did not exit on request, terminated
>
> Looking through the code in wrapper_unix.c I see the following in the
> function wrapperKillProcess() :
>
>         kill(jvmPid, SIGKILL);
>         log_printf(WRAPPER_SOURCE_WRAPPER, LEVEL_ERROR, "JVM did not 
> exit on request, terminated");
>
> What does the message "JVM did not exit on request, terminated" mean?
> There is no test after the kill function is called to tell whether or
> not the JVM exited or not so how can one say that the JVM did not exit?
>
> Thanks.
> Richard