Re: [Wrapper-user] JVM ping timeout

[Mark L. <ma...@bi...>]

Leif,

Thanks for your reply.
(and sorry for the duplicate post - my original post I used the wrong 
email address)

 >    First of all, I would look into what is causing your JVM to not
 > respond for 30 seconds.
Yes, this turned out to be an anti-virus app going nuts as we wrote to 
the log.

 >    As for your question about the wrapper.ping.timeout.  There was a bug
 > fix in
 > version 3.0.2 which fixed this problem. ...
Good, so my understanding was OK, just my version wasn't as up-to-date 
as it could have been.

 >    There have been lots of other fixes as well, so you might want to
 > upgrade to
 > 3.0.5, at some time. ...
I will do that, especially because the wrapper is so well written that 
it is almost fully self-configuring - thank you to all contributors!

Mark.


Leif Mortenson wrote:

> Mark,
>    The Wrapper monitors the health of the JVM in a number of ways.  It 
> is able to tell if
> the JVM process crashes without any hooks in the JVM.  But a common 
> failure mode for
> Java is that the JVM hangs but does not crash.   In order to catch and 
> recover from this,
> the Wrapper pings the JVM once every 5 seconds and will kill the JVM if 
> it does not
> receive a response for more than 30 seconds (the ping timeout)
> 
>    First of all, I would look into what is causing your JVM to not 
> respond for 30 seconds.
> The most common cause of this is garbage collection on very large 
> applications which do
> not have enough system memory.  When the OS has to do swapping to be 
> able complete
> a GC sweep, the sweep can take a couple orders of magnitude longer than 
> normal to
> complete.
> 
>    Anyway, as you said, increasing the ping timeout will work around 
> this.  But the
> drawback to this is that it also increasing the amount of time before 
> the Wrapper is
> able to detect that the JVM has hung.
> 
>    As for your question about the wrapper.ping.timeout.  There was a bug 
> fix in
> version 3.0.2 which fixed this problem.   The pings were detected 
> correctly from
> the Wrapper side things, but there is analogous feature where the JVM 
> will quit
> to let itself be restarted if it has not been pinged by the JVM for more 
> than 30
> seconds.  Up until 3.0.2, this could not be set.  They are now both set 
> using the
> same wrapper.ping.timeout property.
> 
>    There have been lots of other fixes as well, so you might want to 
> upgrade to
> 3.0.5, at some time.  At least go take a look at the release notes and 
> see if any
> of the changes may affect, or benefit you.
> http://wrapper.tanukisoftware.org/doc/english/release-notes.html
> 
> Cheers,
> Leif
> 
> Mark Lassau wrote:
> 
>> I'm using wrapper v3.0.0 to run jBoss as a winNT service.
>>
>> I was getting the following when trying to run our program on a W2K 
>> Server:
>>
>> jvm 1    | 03:12:43,203 INFO  [STDOUT] The Wrapper code did not ping 
>> the JVM
>> for 31 seconds.  Quit and let the wrapper resynch.
>>
>> We have actually solved the problem now, but during analysis I tried 
>> setting the ping timeout to 60 seconds with the following conf option:
>>
>> wrapper.ping.timeout=60
>>
>> The strange thing was, I kept getting the same error message 
>> (including the 31 seconds value)
>>
>> Could someone tell me if I'm missing something here, or maybe this is 
>> a known problem with this version?
> 
> 
> 
> 
> 
> 
> -------------------------------------------------------
> This SF.net email is sponsored by: SF.net Giveback Program.
> Does SourceForge.net help you be more productive?  Does it
> help you create better code?  SHARE THE LOVE, and help us help
> YOU!  Click Here: http://sourceforge.net/donate/
> _______________________________________________
> Wrapper-user mailing list
> Wra...@li...
> https://lists.sourceforge.net/lists/listinfo/wrapper-user
> 
>