Hi All, 

Has there been any activity to resolve this problem. This is a big
show-stopper for us. It only takes a few minutes with a handful of users
before we run out of memory. If we can't get a resolution fairly soon,
we might be forced to abandon OJB (even though I think it's a great
product).

Thanks Much, 
Greg Plummer

-----Original Message-----
From: Armin Waibel [mailto:[EMAIL PROTECTED] 
Sent: Monday, November 17, 2003 8:52 AM
To: OJB Users List
Subject: Re: [rc4 / BUG] major memory leak in MetadataManager


Hi Roger,

Janssen, Roger wrote:
>>are you using ojb from CVS or download?
> 
> i am not using CVS head but release rc4
> 
> searching the internet on 'java threadlocal memory leak', there are a 
> number of hits. some refer to threadlocal usage with the tomcat 
> application container. although i use tomcat as the container for my 
> webapp, i am able to reproduce the problem with my own unit tests, not

> using tomcat.
>
So, can you confirm it's a OJB bug, or is it a
bug in J2SDK?

In OJB test suite shipped with a test called MetadataMultithreadedTest.
This test use the 'pre thread metadata changes' feature. I modified the
test and repeat the test 60 times and check the free memory of the JVM.
All seems ok - using JVM 1.4.1_05-b01

Test results:

RuntimeMetadataChanges test take: 31135 ms
Free Memory before test:          125MB
Free Memory after test:           92MB (JVM gc seems work correct
and free memory)
Free Memory after test after gc:   123MB


regards,
Armin


> Roger Janssen
> 
> 
> -----Original Message-----
> From: Jason Pyeron [mailto:[EMAIL PROTECTED]
> Sent: maandag 17 november 2003 15:06
> To: OJB Users List
> Subject: Re: [rc4 / BUG] major memory leak in MetadataManager
> 
> 
> On Mon, 17 Nov 2003, Janssen, Roger wrote:
> 
> 
>>can anybody confirm this, is anybody working on this?
> 
> 
> are you using ojb from CVS or download?
> 
> I will test it from CVS later on, hopefully today.
> 
> 
>>for me this is a showstopper, which is kind of a problem because it 
>>occurs in a live hosted webapp.
> 
> 
> if true, then DITO.
> 
> -Jason Pyeron
> 
> 



---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]


---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to