We have a situation where creating an EJB container by doing
EJBContainer.createEJBContainer() causes our test to run out of heap space.
This is on a box with 3.48 G of memory with the -Xmx1024m option. The
same option on a Mac with 8 G of memory--same test, rather, with the same
-Xmx option--runs fine.
The out of memory error is coming from within the DynamicReloader thread of
Glassfish.
Is there a way to influence the EJBContainer API to tell the underlying
provider what its reload interval should be? I'm worried that perhaps
embedded glassfish is trying to load my EJB test over and over again in the
background, like perhaps the computer that is running the test is slow enough
that Glassfish is trying to reload the app, while on the Mac it's speedy
enough to never hit this dynamic reloading behavior.
Best,
Laird
--
[Message sent by forum member 'ljnelson']
View Post: http://forums.java.net/node/713182