When catalog as many vector layers (around 600), WFS transactions cause memory leaks and are very slow
------------------------------------------------------------------------------------------------------
Key: GEOS-3907
URL: http://jira.codehaus.org/browse/GEOS-3907
Project: GeoServer
Issue Type: Bug
Components: WFS
Affects Versions: 2.0.x
Environment: OpenSUSE 10, JDK 1.6, Jetty
Reporter: Luca Morandini
Assignee: Andrea Aime
Priority: Minor
I noticed, using jmap, that Xerces-related memory of live objects always grows (to the tune of 2MB at every WFS-T insert, same for XSD-related memory) but never goes down (not even when forcing Garbage Collecton from the web interface). Actually, I observed Xerces-related memory growing up to to 123MB.
The issue is definitively linked to the number of vector layers (see the attached insert script):
A) 600 layers
Used by Xerces: 73.699MB
Used by XSD: 54.323MB
Transaction 1 of 1
....
real 0m14.339s
user 0m0.008s
sys 0m0.000s
Used by Xerces: 76.732MB
Used by XSD: 56.550MB
....End
A) 10 layers
Start...
Used by Xerces: 76.732MB
Used by XSD: 56.550MB
Transaction 1 of 1
....
real 0m1.862s
user 0m0.004s
sys 0m0.000s
Used by Xerces: 77.186MB
Used by XSD: 56.901MB
....End
Therefore, the difference is not only in memory consumption, but in speed too.
To add 600 copies of sf:archsites layer (the one used by the afore-mentioned insert script), please look at the attached script.
This issue may be a duplicate of either: http://jira.codehaus.org/browse/GEOS-3534
or: http://jira.codehaus.org/browse/GEOS-3653
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators: http://jira.codehaus.org/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira