Re: [Wrapper-user] JBoss ping timeout, shuts down but doesn't restart

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

Eoin,
    I was able to reproduce your problem.   This is a bug in the Wrapper 
that I just got
fixed in CVS.  It will be in the next (3.1.1) release of the Wrapper.

    If the Wrapper process does not get any CPU for some reason for an 
extended
period of time then it will fail to ping the JVM.   Normally this will 
only happen if the
system is under extremely heavy load.  However in such cases, the JVM 
will also
experience the same problem and compensate.  It appears that in your 
case, the
JVM had been getting at least a little CPU.

    The JVM is designed to shut itself down in the event that the 
Wrapper process
is ever killed.   Thus the message that it is attempting to shut itself 
down and then
let the JVM restart it.  Once the JVM decides to shut itself down this 
will happen
very quickly.

    The problem is that when the JVM decided to shut itself down, the 
Wrapper
process was thinking that the JVM wanted to exit cleanly and was 
shutting itself
down as well.   The Wrapper should have been launching a new JVM.

    I had reworked the state engine in 3.1.0 and it looks like I broke 
this.  3.0.5
had been working correctly.

    For now, the best option for you is to set the ping timeout to a 
high enough
value that you do not encounter this.  Read the docs for the property so you
understand the issues involved.
http://wrapper.tanukisoftware.org/doc/english/prop-ping-timeout.html

    I'll try and get a release out soon.  There are
a few other 3.1.0 problems that I want to get fixed first however.

Cheers,
Leif


Eoin Curran wrote:

> I'm using wrapper 3.10 on linux, using:
>
> wrapper.java.mainclass=org.tanukisoftware.wrapper.WrapperSimpleApp
> wrapper.app.parameter.1=org.jboss.Main
>
> I got the following in my logs:
> INFO   | jvm 1    | 2004/05/13 21:14:59 | Wrapper Manager: The Wrapper 
> code did not ping the JVM for 60 seconds.  Quit and let the Wrapper 
> resynch.
> INFO   | jvm 1    | 2004/05/13 21:14:59 | 21:14:59,954 INFO  [Server] 
> JBoss SHUTDOWN: Undeploying all packages
>
> the server shut down and was not started up. I don't see anything in 
> the logs that might explain why the ping failed, and there was nothing 
> special happening at that time that I know of. I don't have debug 
> output from wrapper, but am enabling it in case this happens again.
>
> I'm wondering how the ping could fail, but the wrapper was able to 
> tell jboss to shut down so quickly (<1 second later). Also, should the 
> wrapper not restart the server if it kills it as a result of timed out 
> ping?
>
> regards,
> eoin.