[ http://jira.jboss.com/jira/browse/JBSEAM-2436?page=all ] Pete Muir updated JBSEAM-2436: ------------------------------ Component/s: EJB3 Fix Version/s: 2.1.0.GA -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://jira.jboss.com/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://jira.jboss.com/jira/browse/JBSEAM-2436?page=all ] Pete Muir closed JBSEAM-2436. ----------------------------- Resolution: Incomplete Description Assignee: Pete Muir I don't understand the bug report. If you are reporting a bug in Seam, please state explicitly what the problem is. Seam JIRA is not the right place to address production issues in your software. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://jira.jboss.com/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://jira.jboss.com/jira/browse/JBSEAM-2436?page=comments#action_12402956 ] Pete Muir commented on JBSEAM-2436: ----------------------------------- Please reopen if you can provide a clearer description. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://jira.jboss.com/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://jira.jboss.com/jira/browse/JBSEAM-2436?page=comments#action_12402962 ] Antoine Sabot-Durand commented on JBSEAM-2436: ---------------------------------------------- Ok, sorry for the mis-understanding : I'm not Jira-certified :-). I'll try to provide a use case for this problem or at least a clear description next week. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://jira.jboss.com/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://jira.jboss.com/jira/browse/JBSEAM-2436?page=comments#action_12402994 ] Susanne Jarl commented on JBSEAM-2436: -------------------------------------- This is not a production issue from my own written code. I would never post them in JIRA, I really investigate and think a lot before I post here. I also posted in the forum first but no one answered. But maybe I could have stripped my code a bit from stuff that is not interesting but the first time I posted I didn't know exactly what was the problem. I think I discovered a very general serious bug and i reported here in JIRA. If I try to explain it again it goes like this. If you have a session-scoped SFSB which holds entites, the entities does not get destroyed. So in production where many instances of the SFSB is created this leads to OutOfMemoryException IF the session timeout is not set to 2 minutes, because then the instances will be destroyed(!). I have investigated this for many many hours and watched the number of entities just growing (although I saw some instances got removed if I checked the number of instances really often but just a few ). I tried different solutions in my code but it was the exactly same result as long as I had the SFSB in session scope and the setting for Session Timeout more than 2 minutes. I really think you should look into this because it is a really serious bug. I hade to rewrite all my code not to use SFSB in session scope. Hopefully Antoine Sabot-Durand will provide you some useful code. Otherwise you could use mine, but you may strip it from all stuff that is not intresting. I could help with that as well but I can't really test this since I don't have any environment set up for load testing. If there is any other information or help I can provide please tell me. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://jira.jboss.com/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://jira.jboss.com/jira/browse/JBSEAM-2436?page=comments#action_12403006 ] Susanne Jarl commented on JBSEAM-2436: -------------------------------------- I expect my entities to be destroyed when the SFSB that holds the entities is destroyed. When I'm writing about number of entities I mean the number being held by the JVM seen through a profiler. I used the profiler that comes with Java 6. Always when I write "number of instances" I mean the number being held by the JVM. Maybe this is not correct way to express that. Is the correct expression "number of instances held by the JVM"? If I set the session timeout to 2 minutes I will not get a OOME. You can see that garbage collection works in the profiler - the number of instances is held at a rather constant level that varies with load during the hours of the day. But if you set the session timeout higher the garbage collection does not work - number of instances in the JVM is increasing. The higher I set the session timout the sooner I get the OOME. I am sure of that the problem is not that the session has not timed out yet, that the number of instances are growing. I have followed the profiler closely and I believe that the correct behaviour would be if I set the session timeout to 10 minutes, after 10 minutes the number of instances should be around the same value as long as the load on the web page is about the same. But the number of instances has increased regardless of the load is the same or decreasing. I followed this for several hours with a 10 minutes session timeout and even after 2 hours with decreasing load the number of instances continued to grow. I am not 100% sure but I think that the SFSB do not get destroyed either. But why would that not be a bug? The garbage collection seems to stop working for SFSB in session scope which holds entities if the session timeout is set higher than 2 minutes. Something is magic about this 2 minutes limit. There must be a bug somewhere. But like Antoine Sabot-Durand I also upgraded to JBoss 4.2.2 so the bug might be there and not in Seam 2.0.0? -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://jira.jboss.com/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://jira.jboss.com/jira/browse/JBSEAM-2436?page=comments#action_12403085 ] Susanne Jarl commented on JBSEAM-2436: -------------------------------------- If I would test this I would have to do it in my production environment which I am not so keen on. I hope that Antoine Sabot-Durand will provide you with the test code you need. If there is no other alternative I could test it but I hope I will not have to do that. I will follow the conversation of this bug report. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://jira.jboss.com/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://jira.jboss.com/jira/browse/JBSEAM-2436?page=all ] Pete Muir reopened JBSEAM-2436: ------------------------------- Assignee: Jay Balunas (was: Pete Muir) Jay, I'm assigning to you as this is related to scalability. Can you investigate when you have any time? -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://jira.jboss.com/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://jira.jboss.com/jira/browse/JBSEAM-2436?page=all ] Pete Muir updated JBSEAM-2436: ------------------------------ Fix Version/s: 2.1.0.GA (was: 2.1.0.BETA1) -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://jira.jboss.com/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ https://jira.jboss.org/jira/browse/JBSEAM-2436?page=com.atlassian.jira.pl... ] Michael Kostadinovich commented on JBSEAM-2436: ----------------------------------------------- We were plagued by the same problem, after some investigation we found a fix/workaround: session scoped SFSBs that get passivated, never get removed from the memory nor get released by SimpleStatefulCache. Workaround is to switch to NoPassivationCache, either by using the @Cache(org.jboss.ejb3.cache.NoPassivationCache.class) annotation on your class or change the default cache implementation to NoPassivationCache (see http://www.jboss.org/community/docs/DOC-9656 ). Additionally (still unconfirmed, I will have to do more testing to confirm this), if you have an abstract class which defines a @Remove @Destroy method and your SFSBs extend this class, Seam might/will not call the @Destroy at the end of the session. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: https://jira.jboss.org/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ https://jira.jboss.org/jira/browse/JBSEAM-2436?page=com.atlassian.jira.pl... ] Pete Muir commented on JBSEAM-2436: ----------------------------------- Michael, if you can provide an example to reproduce, we can take a look. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: https://jira.jboss.org/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira