Dain-
Do you have the data cache enabled? If so, can you disable it and see
if it changes anything? I'm still guessing that this might be due to
dynamic class generation, which the data cache sometimes does.
If you enable verbose logging (e.g., by setting the property
"openjpa.Log: DefaultLevel=TRACE"), what are the last few log
messages you see before the hang?
On Dec 31, 2006, at 12:28 PM, Dain Sundstrom wrote:
On Dec 31, 2006, at 12:04 AM, Marc Prud'hommeaux wrote:
I've never experienced this problem, but as a guess, are you
running with dynamic enhancement (using the "-javaagent" flag)? If
so, what happens if you enhance manually, and then run without the
agent flag?
Currently, I am, but I can recreate the problem without it, it just
takes a lot longer. Actually, it seems be related to the amount of
code executed. If I run the Agent and it enhances, that is a lot
of code executed and the problem happens quickly. If I run derby
instead of HSQLDB, it happens quickly. It is very strange.
Also, you could always try running "javastack" on a process (found
using "jps") from a console and see if you get a stack trace from
that.
I'll try that.
-dain