-
Bug
-
Resolution: Cannot Reproduce
-
P3
-
6
-
None
-
sparc
-
solaris_9
Calling method start on a JMX connector server is said to have an effect the first time and no effect from the second call.
On Sparc Solaris 9 (Ultra 10), I experiment a memory leak when looping over that call to start.
BUT on Windows XP Pro, there is no leak at all so it could be OS and/or hardware specific.
Tests done:
1 hour long & expect no memory consumption the last 30 minutes
3 hours long & expect no memory consumption the last 90 minutes (output attached)
rmi pass the two tests.
iiop fails the small one and pass the long one.
Long test does 6812207542 calls and stop consuming memory after around 80 minutes, so roughly we need 6812207542 / 180 * 80 = 3027644769 calls to reach a stable state.
There is no reason iiop does worse than rmi for an API call said impact less.
###@###.### 2005-1-28 17:25:12 GMT
On Sparc Solaris 9 (Ultra 10), I experiment a memory leak when looping over that call to start.
BUT on Windows XP Pro, there is no leak at all so it could be OS and/or hardware specific.
Tests done:
1 hour long & expect no memory consumption the last 30 minutes
3 hours long & expect no memory consumption the last 90 minutes (output attached)
rmi pass the two tests.
iiop fails the small one and pass the long one.
Long test does 6812207542 calls and stop consuming memory after around 80 minutes, so roughly we need 6812207542 / 180 * 80 = 3027644769 calls to reach a stable state.
There is no reason iiop does worse than rmi for an API call said impact less.
###@###.### 2005-1-28 17:25:12 GMT