Re: [Wrapper-user] Wrapper not really terminating the JVM

[Christian M. <chr...@ta...>]

Hi Victor,

sorry I forgot to mention that if you can't use the port range approach,
please use property wrapper.restart.delay and set it to a value of 4
minutes or larger, for instance:
wrapper.restart.delay=240

This way the Wrapper will always wait with the restart until the system
freed the port.

Best Regards,

Christian



On Tue, Mar 5, 2013 at 3:50 PM, Christian Mueller <
chr...@ta...> wrote:

> Hi Victor,
>
> none. The Wrapper will always terminate the JVM eventually. So your
> problem is rather a problem with your system, or how TCP/IP ports are
> working.
>
> I think what are you seeing is that your application is binding a socket
> at a certain port (port C) and to establish to create a connection to
> another port (port S).
> So when your application is now being stopped forcibly by the Wrapper (or
> a crash or anything causing the application to stop immediately), then the
> system is leaving port C for a defined period of time in state TIME_WAIT,
> which is around 1 - 4 minutes and then after this period the system is
> going to free the port.
> As long as the port is in TIME_WAIT, it can't be reused. The TIME_WAIT
> period is usually twice the max packet lifetime, this protects actually
> your client socket from receiving data from the previous connection.
>
> I would recommend to rather use a port range for your client port, so even
> if your first attempt to bind fails, you can try the next one etc, one
> eventually works.
>
> Please let me know if you have any further questions.
>
> Thank you,
>
> Christian Mueller
> Tanuki Software, Ltd.
>
>
>
>
>
> On Mon, Mar 4, 2013 at 10:27 PM, ויקטור ר <har...@wa...> wrote:
> >
> >
> > Hi all
> >
> > It seems to me that sometimes the JVM doesnt get terminated during the
> restart for us.
> >
> >
> > We use Wrapper Professional Edition 64-bit 3.3.9 for our project
> (Windows OS) and sometimes we see the following in our logs:
> >
> >
> > wrapper  | JVM requested a restart.....
> > wrapper  | Shutdown failed: Timed out waiting for the JVM to terminate.
> > wrapper  | Dumping JVM state.
> > wrapper  | JVM did not exit on request, terminated
> > wrapper  | Reloading Wrapper configuration...
> > wrapperp | port 1777 already in use, using port 32001 instead.
> > wrapper  | Launching a JVM...
> > jvm 2    | -XX:+CMSClassUnloadingEnabled -XX:+CMSIncrementalMode
> -XX:CompileCommand=exclude,jregex/Pretokenizer,next
> -XX:+HeapDumpOnOutOfMemoryError -XX:InitialHeapSize=536870912
> -XX:+ManagementServer -XX:MaxGCPauseMillis=250 -XX:MaxHeapSize=1073741824
> -XX:MaxPermSize=201326592 -XX:MaxTenuringThreshold=6 -XX:NewRatio=2
> -XX:OldPLABSize=16 -XX:PermSize=134217728 -XX:+PrintCommandLineFlags
> -XX:+UseAdaptiveGCBoundary -XX:+UseCompressedOops -XX:+UseConcMarkSweepGC
> -XX:-UseLargePagesIndividualAllocation -XX:+UseParNewGC
> > jvm 2    | CompilerOracle: exclude jregex/Pretokenizer.next
> > jvm 2    | WrapperManager: Initializing...
> > jvm 2    | Starting the Probe ...
> >
> > After that we have some application specific exception because some RMI
> that we use is not free because it seems like the previous instance of our
> application is still alive.
> > Are there cases when the wrapper reports that it terminates the JVM but
> it isnt?
> >
> > Thanks in advance
> > Victor
> >
> > ________________________________
> > Walla! Mail - Get your free unlimited mail today
> >
> > �
> >
> > ________________________________
> > Walla! Mail - Get your free unlimited mail today
> >
> ------------------------------------------------------------------------------
> > Everyone hates slow websites. So do we.
> > Make your web apps faster with AppDynamics
> > Download AppDynamics Lite for free today:
> > http://p.sf.net/sfu/appdyn_d2d_feb
> > _______________________________________________
> > Wrapper-user mailing list
> > Wra...@li...
> > https://lists.sourceforge.net/lists/listinfo/wrapper-user
> >
>