-
Bug
-
Resolution: Done
-
Blocker
-
7.1.0.DR18
-
None
I was investigating on the use case described at https://developer.jboss.org/message/972213 and from what I can see this is still an issue for WFTC.
The case is following
- server 1 starts transaction
- server 1 calls server 2 via ejb call, the context is propagated
- server 2 does a business logic (db insertion, jms call)
- server 2 returns to server 1
- server 1 continue with long time processing
- meanwhile server 2 is restarted. that leads to the cleaning of the notion about the propagated transaction
- server 1 ends its processing and tries to commit (prepare)
- server 1 should fail as server 2 does not know the Xid to commit/prepare
The current WFTC implementation is not failing and leaving the execution to proceed with success. It shows no error. For one phase commit the return of the commit is TwoPhaseOutcome.FINISH_OK, for two-phase commit the return is TwoPhaseOutcome.PREPARE_READONLY.
- is cloned by
-
WFTC-29 Transaction server channel should not create a new transaction for before completion phase
- Resolved
- is incorporated by
-
JBEAP-11718 Upgrade WildFly Transaction Client to 1.0.0.CR2
- Closed
- relates to
-
JBEAP-8625 Injecting of JMSContext via CF fails
- Closed