[ https://jira.jboss.org/jira/browse/GTNPORTAL-855?page=com.atlassian.jira.... ] Marek Posolda updated GTNPORTAL-855: ------------------------------------ Environment: GateIn trunk (revision 2166) deployed on EAP5, Sun JDK 1.6, MySQL 5.1 shared database (between two GateIn nodes) was: GateIn trunk (revision 2083) deployed on EAP5, Sun JDK 1.6, MySQL 5.1 shared database (between two GateIn nodes) Description: I have a problem when having two GateIn instances on different servers running in cluster. Steps to simulate: 1) Run node1 with -Dexo.profiles=cluster 2) Run node2 with -Dexo.profiles=cluster 3) Login as root in node1 and go to portalnavigation -> Edit navigations -> Add node -> Use existing page in page selector (portal::classic::homepage) -> Fill node name and title -> click to Save 4) Now I am not able to see my new node. But I am not able to see it after logout. I can see exceptions in server logs in both node1 and node2. Exception on node2 is "[TxInterceptor] prepare failed! java.lang.NullPointerException". Exception on node1 is "[TxIsolatedOperation] Transaction has been rolled back javax.transaction.RollbackException: [com.arjuna.ats.internal.jta.transaction.arjunacore.commitwhenaborted] [com.arjuna.ats.internal.jta.transaction.arjunacore.commitwhenaborted] Could not commit transaction." Exception on node2 (NullPointerException) occurs first and exception "RollbackExcption" on node1, where new navigations were created, occurs after it. And I can see only when having cluster with more nodes. It doesn't occur in single node. For completeness: I am using shared database MySQL with JBoss datasources in server/production/deploy directory and "no-tx-datasource" datasource. Both nodes are executed with "-Dexo.profiles=cluster". I am attaching my server/production/conf/gatein/configuration.xml. was: I have a problem when having two GateIn instances on different servers running in cluster. Steps to simulate: 1) Run node1 with -Dexo.profiles=cluster 2) Run node2 with -Dexo.profiles=cluster 3) Login as root in node1 and go to portalnavigation -> Edit navigations -> Add node -> Use existing page in page selector (portal::classic::homepage) -> Fill node name and title -> click to Save 4) Now I am able to see my new node. But I can see exceptions in server logs in both node1 and node2. Exception on node2 is "[TxInterceptor] prepare failed! java.lang.NullPointerException". Exception on node1 is "[TxIsolatedOperation] Transaction has been rolled back javax.transaction.RollbackException: [com.arjuna.ats.internal.jta.transaction.arjunacore.commitwhenaborted] [com.arjuna.ats.internal.jta.transaction.arjunacore.commitwhenaborted] Could not commit transaction." Full stacktraces of both exceptions are attached. Interesting is that I am also able to see the page on second node. So it seems that nothing happens in application itself, only exceptions in server logs are strange. Exception on node2 (NullPointerException) occurs first and exception "RollbackExcption" on node1, where new navigations were created, occurs after it. And I can see only when having cluster with more nodes. It doesn't occur in single node. For completeness: I am using shared database MySQL with JBoss datasources in server/production/deploy directory and "no-tx-datasource" datasource. Both nodes are executed with "-Dexo.profiles=cluster". I am attaching my server/production/conf/gatein/configuration.xml. -- 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/GTNPORTAL-855?page=com.atlassian.jira.... ] Marek Posolda commented on GTNPORTAL-855: ----------------------------------------- New navigation node can be seen on both servers after restart of GateIn. -- 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/GTNPORTAL-855?page=com.atlassian.jira.... ] Marek Posolda commented on GTNPORTAL-855: ----------------------------------------- I am seeing similar problem to this after changing access permission of some portal (like classic portal). After changing permissions and click to "Save" are the same exceptions in server log on both cluster nodes. -- 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/GTNPORTAL-855?page=com.atlassian.jira.... ] Julien Viet edited comment on GTNPORTAL-855 at 4/20/10 3:09 PM: ---------------------------------------------------------------- I was able to reproduce the issue but with a different exception on one of the two nodes (the other is similar) on EAP 5.0 as well, but using shared HSQLDB. was (Author: julien_viet): I was able to reproduce the issue but with a different exception on one of the two nodes (the other is similar) -- 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/GTNPORTAL-855?page=com.atlassian.jira.... ] Marek Posolda commented on GTNPORTAL-855: ----------------------------------------- I am able to reproduce it with latest trunk with same scenario as described and same exceptions in both nodes in server.log. NullPointerException coming from JBoss cache on node2, RollbackException on node1. Exception occurs directly after creating new page or navigation node. I have GateIn trunk from today (revision 2716) deployed on EAP 5.0.1 CR3. AS configuration "production". JBoss cache versions: 3.2.2.GA in server/production/lib 3.2.3.CR1 in server/production/deploy/gatein.ear/lib configuration.properties are configured to use cluster (attaching newest version), MySQL 5.1 is used as shared database and JCA datasources are in server/production/deploy (so HSQLDB configuration in configuration.properties should be ignored thanks to org.exoplatform.commons.InitialContextInitializer2 ) I am able to simulate it in lab on more hosts and also on localhost with virtual hosts. I am running portal on localhost this way: ./run.sh -c production -b 127.0.1.1 -g portalFailoverPartition -u 239.10.15.18 -Djboss.messaging.ServerPeerID=0 ./run.sh -c production2 -b 127.0.1.2 -g portalFailoverPartition -u 239.10.15.18 -Djboss.messaging.ServerPeerID=1 Is more info needed? -- 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/GTNPORTAL-855?page=com.atlassian.jira.... ] Marek Posolda updated GTNPORTAL-855: ------------------------------------ Attachment: configuration.properties Attaching newest configuration.properties -- 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/GTNPORTAL-855?page=com.atlassian.jira.... ] Nicolas Filotto commented on GTNPORTAL-855: ------------------------------------------- Sorry this issue also affect 3.2.2.GA, could you then test with 3.2.1.GA? -- 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/GTNPORTAL-855?page=com.atlassian.jira.... ] Marek Posolda commented on GTNPORTAL-855: ----------------------------------------- I just retested it with JBoss Cache 3.2.3.GA from nexus repository. I have jbosscache-core-3.2.3-GA only in server/production/lib and I removed jbosscache-core jar from server/production/lib/gatein.ear/lib . But it is still failing in my environment... With same exceptions on node2 (NullPointerException) and node1 (RollbackException). I also tried to use HSQLDB as shared database instead of MySQL. But nothing changed and I am seeing the same exceptions. -- 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/GTNPORTAL-855?page=com.atlassian.jira.... ] Galder Zamarreno commented on GTNPORTAL-855: -------------------------------------------- The NPE appears to be the cause of JBCACHE-1579. -- 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/GTNPORTAL-855?page=com.atlassian.jira.... ] Thomas Heute updated GTNPORTAL-855: ----------------------------------- Attachment: GTNPORTAL-855.diff So using read committed instead of repeteable read, as proposed by Galder, has worked for me, I've attached the patch here (GTNPORTAL-855.diff). That would require a careful QA though. -- 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/GTNPORTAL-855?page=com.atlassian.jira.... ] Marek Posolda commented on GTNPORTAL-855: ----------------------------------------- So I tried it also in "original" (non-patched) environment and with newest jbosscache-core.jar build from Galder. And it also helps, issue can't be reproduced anymore on my side. -- 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/GTNPORTAL-855?page=com.atlassian.jira.... ] Marek Posolda commented on GTNPORTAL-855: ----------------------------------------- Retested with EPP5 CR1 with JBoss cache 3.2.4.GA. Issue can't be simulated anymore with JBC 3.2.4.GA (related issue https://jira.jboss.org/jira/browse/JBCACHE-1579 seems to be fixed in JBC 3.2.4.GA) -- 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/GTNPORTAL-855?page=com.atlassian.jira.... ] Thomas Heute commented on GTNPORTAL-855: ---------------------------------------- Solved with JBC 3.2.4, still need to make sure that's the one we ship -- 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