Yes. We first noticed that from the Windows Task Manager, where the process javaw.exe keeps increasing its memory usage and it never gets released, until the Nuxeo server is stopped. We also noticed that the Nuxeo DM starts slowing down and eventually, when it reaches the max. memory size allocated by jboss startup option -Xmx, Nuxeo does not allow any users to log in and existing users are logged out. The server console keeps listing java errors and the only way to resolve this is to stop (and sometimes having to terminate) the Nuxeo server.
The objects that remain in memory include the classes of our custom beans. If I take a heap dump after cancelling the creation of a document, our bean objects are still in memory. If I create a button on the document which destroys the objects, our bean objects are no longer in memory when I take a heap dump after cancelling the creation of a document. This clearly shows that for some reason our class objects are not cleared. We do have a destroy method in the class, but this is never called (I put warning messages in the log to double confirm this). I don't know whether this is related, but the Nuxeo console prints this message when a new document (containg the data table) is created: [[/nuxeo]] Queue with name 'document_create' has already been registered. Any further help/suggestions would be appreciated. --- Mailing list: [email protected] Forum: http://forum.nuxeo.org/f/1/
