I've fixed the problem now. Incidently I am using version 4.0.1, not 4.0.2.
I modified java file org.apache.tapestry.services.impl.ObjectPoolImpl and changed private Map _pool = new HashMap(); to private Map _pool = new WeakHashMap(); Now when the JVM is low on memory the garbage collector will collect the keys and the values will also be collected. Yes I know there will be performance implications with this, but it's still fast enough. 2009/12/16 Jabbar <aja...@gmail.com>: > Hello all, > > I am currently using tapestry 4.0.2 and have found a memory leak. For > some reason org.apache.tapestry.services.impl.ObjectPoolImp is 84MB in > size, which is 67% of the heap space. It hold hundreds of page class > instances. I've been using Apache Jmeter to simulate 14 concurrent > users, accessing different pages in the application. > > Is there a way to specify how many page class instances to cache? How > compatible is tapestry 4.1.6? Can I drop the 4.1.6 binaries in place > of 4.0.2? > > I know that some work was done with page class instance caching in the > 4.1.x branch. > > > -- > Thanks > > A Jabbar Azam > -- Thanks A Jabbar Azam --------------------------------------------------------------------- To unsubscribe, e-mail: users-unsubscr...@tapestry.apache.org For additional commands, e-mail: users-h...@tapestry.apache.org
