Back to bug 1019378
| Who | When | What | Removed | Added |
|---|---|---|---|---|
| Justin Bertram | 2013-10-15 18:12:59 UTC | Assignee | csuconic | jbertram |
| Justin Bertram | 2013-10-15 18:27:04 UTC | Status | NEW | ASSIGNED |
| Justin Bertram | 2013-10-21 16:20:57 UTC | Assignee | jbertram | csuconic |
| mark yarborough | 2013-10-25 13:43:34 UTC | CC | myarboro | |
| Clebert Suconic | 2013-11-04 23:18:03 UTC | Summary | Kill/Shutdown of server in cluster leads to lost messages | Message Redistribution could lead to loss of messages if paging and reading with batched Transactions |
| Clebert Suconic | 2013-11-04 23:20:11 UTC | Doc Text | Cause: In a rare race, if messages are being acked too fast with big chunks, redistribution could read a record before the Transaction was instantiated on the page system. Consequence: A message would be lost. Fix: We made sure the page system would correctly instantiate the page transaction, only writing the file after the page transaction is installed. Result: No messages were lost after the fix. |
|
| Clebert Suconic | 2013-11-04 23:20:56 UTC | Target Milestone | --- | CR1 |
| Clebert Suconic | 2013-11-04 23:21:23 UTC | Target Release | --- | EAP 6.2.0 |
| Brian Stansberry | 2013-11-05 03:15:04 UTC | Depends On | 1026553 | |
| Brian Stansberry | 2013-11-06 02:29:21 UTC | Status | ASSIGNED | MODIFIED |
| CC | brian.stansberry | |||
| Paul Gier | 2013-11-11 03:18:42 UTC | Status | MODIFIED | ON_QA |
| Martin Svehla | 2013-11-11 12:47:08 UTC | Status | ON_QA | VERIFIED |
| Lucas Costi | 2013-11-15 05:56:50 UTC | CC | lcosti | |
| Doc Text | Cause: In a rare race, if messages are being acked too fast with big chunks, redistribution could read a record before the Transaction was instantiated on the page system. Consequence: A message would be lost. Fix: We made sure the page system would correctly instantiate the page transaction, only writing the file after the page transaction is installed. Result: No messages were lost after the fix. | In a rare circumstances, if messages were being acknowledged too fast with big chunks on a HornetQ server, a message redistribution could read a record before the transaction was instantiated on the page system. This situation would result in message loss. This issue has been fixed in this release of JBoss EAP 6 by making sure the paging system will correctly instantiate a page transaction, and only writing the file after the page transaction is instantiated. As a result of this fix, under the same circumstances there will be no lost messages. |
||
| mark yarborough | 2013-12-15 16:17:00 UTC | Status | VERIFIED | CLOSED |
| Resolution | --- | CURRENTRELEASE | ||
| Last Closed | 2013-12-15 11:17:00 UTC |
Back to bug 1019378