Thanks! My "feature type cache size" in the "Global Settings" was set to zero. I increased this setting to 25 (my server has 20 layers). I was then able to run my load test for 10 minutes.
I then got ambitious and tried an hour long load test. The hour long test case ran fine so now I'm going to try longer. Thanks again! On Fri, Dec 31, 2010 at 2:45 PM, Andrea Aime <[email protected]> wrote: > On Thu, Dec 30, 2010 at 10:45 PM, Leon Doud <[email protected]> wrote: >> While doing some load testing with WFS-T (inserting information) I ran >> into java.lang.OutOfMemoryError (Java heap space). I reran the test >> several times, each time watching the amount of available memory in >> Tomcat. GeoServer would work fine but the amount of available memory >> would steadily decrease until it about 3 mb (from 500 mb) and then run >> out of heap space. >> >> The load testing is 90 to 100 insertions a second for about 5 minutes >> solid. The stack trace I saw while trying to shutdown Tomcat (after >> the heap space error) mentioned a thread named >> "WeakCollectionCleaner". Is there a hashtable that is "caching" the >> feature type data? If there is such a cache, is there a way to >> configure a cap on the size of the hashtable or a frequency for >> cleaning the hashtable? > > There is one, but it's not working the way you think: the leak will present > itself any time the feature type cache in ResourcePool is not large > enough to accomodate all feature types. This will result in some reloads > of the XML cache schemas, which in turn triggers a memory leak. > > The problem appears to be buried very much inside the Eclipse XSD library > code and there seems not to be much we can do about it, but there > is a workaround, which is to make sure the feature type cache is large > enough to keep all feature types definitions in memory. > > Try going in the "Global Settings" page and set the "feature type cache" > value to some integer larger than the total number of layers in your > GS configuration and try again. > If you still can reproduce the problem it would be nice to have > some way to reproduce. > See also these jira issues: > http://jira.codehaus.org/browse/GEOS-3534 > http://jira.codehaus.org/browse/GEOS-3653 > http://jira.codehaus.org/browse/GEOS-3907 > > Cheers > Andrea > > ----------------------------------------------------- > Ing. Andrea Aime > Senior Software Engineer > > GeoSolutions S.A.S. > Via Poggio alle Viti 1187 > 55054 Massarosa (LU) > Italy > > phone: +39 0584962313 > fax: +39 0584962313 > > http://www.geo-solutions.it > http://geo-solutions.blogspot.com/ > http://www.linkedin.com/in/andreaaime > http://twitter.com/geowolf > > ----------------------------------------------------- > ------------------------------------------------------------------------------ Learn how Oracle Real Application Clusters (RAC) One Node allows customers to consolidate database storage, standardize their database environment, and, should the need arise, upgrade to a full multi-node Oracle RAC database without downtime or disruption http://p.sf.net/sfu/oracle-sfdevnl _______________________________________________ Geoserver-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/geoserver-users
