|
From: Leif M. <le...@ta...> - 2006-04-10 02:27:01
|
Jim,
What version of the Wrapper are you using? If 3.1.2, try setting the
wrapper.use_system_time=FALSE property. It is default in 3.2.0. This
new timing
mechanism handles heavily loaded systems much better.
If that fails, it may be necessary to set the following two timeout
properties.
Be sure to familiarize yourself with the docs of any of these properties
before
setting them however.
wrapper.cpu.timeout=180
wrapper.ping.timeout=300
http://wrapper.tanukisoftware.org/doc/english/prop-cpu-timeout.html
http://wrapper.tanukisoftware.org/doc/english/prop-ping-timeout.html
These will probably get rid of the problems you are seeing, but they
will
also slow down the Wrapper's ability to recover from a crashed jvm.
Cheers,
Leif
Jim Redman wrote:
> We have a graphical application that we run as an .exe using the
> wrapper. It restarts on a timeout:
>
> ERROR | wrapper | 2006/04/07 00:08:05 | JVM appears hung: Timed out
> waiting for signal from JVM.
> ERROR | wrapper | 2006/04/07 00:08:05 | JVM did not exit on request,
> terminated
>
> I don't really case about the wrapper timeout, the application is
> interactive and the user can take care of real problems, but I'm
> wondering what we're doing that's causing a problem.
>
> Are the "pings" sent from the Java Event thread? The application
> could use a little optimization and is a memory hog. We tend to go
> into phases in the program where we'll take 100% CPU time, such as
> compiling applications, so we get, eg this:
>
> INFO | wrapper | 2006/04/06 23:32:15 | Wrapper Process has not
> received any CPU time for 141 seconds. Extending timeouts.
>
> but even during this the screen is repainting (usually and slowly).
>
> Not too important, but any thoughts appreciated.
>
> Jim
>
|