[ https://jira.jboss.org/browse/ISPN-691?page=com.atlassian.jira.plugin.sys... ] Galder Zamarreño commented on ISPN-691: --------------------------------------- Craig, do you have a unit test for this? I'm not sure I understand how when max size is hit, items might get flushed out of order. Could you explain in further detail the behaivour you're seeing? Is it because the queue might be flushed concurrently if max size is hit and timer expires at the same time? -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: https://jira.jboss.org/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ https://jira.jboss.org/browse/ISPN-691?page=com.atlassian.jira.plugin.sys... ] craig bomba commented on ISPN-691: ---------------------------------- Was thinking a separate JIRA would be needed for this. Not. Was actually thinking of ISPN-608 which requested that a separate JIRA be created for the out-of-orderness. That is this JIRA. So, last thing to do on this one is submit the code change. Will do. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: https://jira.jboss.org/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ https://jira.jboss.org/browse/ISPN-691?page=com.atlassian.jira.plugin.sys... ] craig bomba commented on ISPN-691: ---------------------------------- Picked up the latest changes from Manik on this source file (from ISPN-716) and added the synchronization to the flush method to address the out of order concerns. Attached the new source. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: https://jira.jboss.org/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ https://jira.jboss.org/browse/ISPN-691?page=com.atlassian.jira.plugin.sys... ] Galder Zamarreño commented on ISPN-691: --------------------------------------- Craig, thanks for attaching the logs and diagnosis. On top of what you're saying, it seems to me that the scheduled repl queue flush thread name is not the best? pool-*. I'm checking this as well. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: https://jira.jboss.org/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ https://jira.jboss.org/browse/ISPN-691?page=com.atlassian.jira.plugin.sys... ] Galder Zamarreño reopened ISPN-691: ----------------------------------- Reopen cos the issue has not actually been committed even though Craig might have attached a fix for it. I'll review the fix and check the thread naming too. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: https://jira.jboss.org/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ https://jira.jboss.org/browse/ISPN-691?page=com.atlassian.jira.plugin.sys... ] Galder Zamarreño commented on ISPN-691: --------------------------------------- Craig, there's something I don't understand about the logs. The background timer thread should, by default, be named as: "Scheduled-replicationQueue-thread-*", but in your logs I see none of that. All I see is "pool-*" threads. So, what configuration changes or code changes are you applying in your system? And, if all threads are called "pool-*", how do u figure out which is the background thread and which the worker thread? Another thing I don't get is the fact that you have configured replication queue like this: jgroups-infinispan-MyInfispanLooperMaster.log:34:2010-09-03 11:48:45,610 3919 TRACE [org.infinispan.remoting.ReplicationQueue] (main:) Starting replication queue, with interval 100000 and maxElements 500 So, it means that replication queue should be flushed, every 100.000 ms which is 100s and max elements of queue is 500. However, when I look at the logs, I see plenty of flushes happening when the queue is 100 elements: 2010-09-03 11:49:43,244 61553 TRACE [org.infinispan.remoting.ReplicationQueue] (pool-5-thread-2:) flush(): flushing repl queue (num elements=100) 2010-09-03 11:49:43,244 61553 TRACE [org.infinispan.remoting.ReplicationQueue] (pool-5-thread-2:) Flushing 100 elements ... 2010-09-03 11:49:43,255 61564 TRACE [org.infinispan.remoting.ReplicationQueue] (pool-5-thread-9:) flush(): flushing repl queue (num elements=100) 2010-09-03 11:49:43,256 61565 TRACE [org.infinispan.remoting.ReplicationQueue] (pool-5-thread-9:) Flushing 100 elements It seems like these flushes are based on number of elements rather than time, but they do not match the configuration settings. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: https://jira.jboss.org/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ https://jira.jboss.org/browse/ISPN-691?page=com.atlassian.jira.plugin.sys... ] craig bomba commented on ISPN-691: ---------------------------------- Galder, I am verifying the question in regard to the background timer thread name and the distinction from the worker threads. Will post another comment in a bit. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: https://jira.jboss.org/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira