Hi,
I'm in a situation where geoserver builds fine from command line,
yet trying to run all of the wfs tests from Eclipse OOM's after
some tens of tests are run (to try out, right click on the test
source folder and "run as unit test", this will run all of the
unit tests contained in the folder).
Now, in Eclipse this happens because the more limited amount of
memory given to the VM (64mb) compared to the command line
(where I believe surefire is instructed to run with 256/512m
or memory).
The fact that in eclipse all tests after a certain point
fail with an OOM error makes me believe there is a memory
leak that is able to suck 64MB or ram after a while of time.
This did not happen on 1.6.x
Now, I don't know what it is, but does it ring any bell
to you?
Cheers
Andrea
I've got a report of a memory leak for a WFS-only geoserevr installation - it uses complex schemas, but my initial tests with these schemas didnt look like it was leaking, so I suspect it might be a leak in the actual request or protocol handling, not the response.
Rob
Andrea Aime wrote:
Hi,
I'm in a situation where geoserver builds fine from command line,
yet trying to run all of the wfs tests from Eclipse OOM's after
some tens of tests are run (to try out, right click on the test
source folder and "run as unit test", this will run all of the
unit tests contained in the folder).
Now, in Eclipse this happens because the more limited amount of
memory given to the VM (64mb) compared to the command line
(where I believe surefire is instructed to run with 256/512m
or memory).
The fact that in eclipse all tests after a certain point
fail with an OOM error makes me believe there is a memory
leak that is able to suck 64MB or ram after a while of time.
This did not happen on 1.6.x
Now, I don't know what it is, but does it ring any bell
to you?
Cheers
Andrea
-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
Geoserver-devel mailing list
Geoserver-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/geoserver-devel
Yes, I posted this problem a week or go or so. I noticed it slightly
after the move to trunk.
One thing I found was that the mock data directory was being created and
destroyed twice for every test. Fixing that helped but did not solve the
problem. I wonder if the spring context being destroyed and recreated
every for every test. This should not be necessary, the same context
should be able to be used for every single test.
After profiling for a while it was still not evident. But I think better
managing of the spring context will help.
This is still on my todo list. There is a jira for it here.
http://jira.codehaus.org/browse/GEOS-1524
-Justin
Andrea Aime wrote:
Hi,
I'm in a situation where geoserver builds fine from command line,
yet trying to run all of the wfs tests from Eclipse OOM's after
some tens of tests are run (to try out, right click on the test
source folder and "run as unit test", this will run all of the
unit tests contained in the folder).
Now, in Eclipse this happens because the more limited amount of
memory given to the VM (64mb) compared to the command line
(where I believe surefire is instructed to run with 256/512m
or memory).
The fact that in eclipse all tests after a certain point
fail with an OOM error makes me believe there is a memory
leak that is able to suck 64MB or ram after a while of time.
This did not happen on 1.6.x
Now, I don't know what it is, but does it ring any bell
to you?
Cheers
Andrea
-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
Geoserver-devel mailing list
Geoserver-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/geoserver-devel
!DSPAM:4007,47472623115221137850744!
--
Justin Deoliveira
The Open Planning Project
http://topp.openplans.org