Atomikos Forum
Can you help spot why it rollbacks with code A and not code B |
Hello, i'm using Atomikos along with Spring for declarative transactions, and the transactional resource is JBossPOJOCache.
I have a unit test that verifies that failed operations are rolled back; the test exercises our application, which has two different versions A and B (B is a refactor of A + some new functionality).
The test passes fine with code A but not B (it doesn't rollback with B). The problem is that the A and B codes are quite different so i'm having a hard time spotting what's wrong.
Using the logs, i see something very suspicious:
-on node A, there are calls to
1/"afterCompletion(STATUS_ROLLEDBACK) called on Synchronization:org.jboss.cache.pojo.interceptors.PojoTxSynchronizationHandler"
2/"afterCompletion(STATUS_ROLLEDBACK) called on Synchronization:tx=com.atomikos.icatch.jta.TransactionImp@nnnn, handlers=........"
-on node B, there are only calls to 1 but not 2.
That sound like a significant difference, so maybe someone here can give me a clue?
I have a unit test that verifies that failed operations are rolled back; the test exercises our application, which has two different versions A and B (B is a refactor of A + some new functionality).
The test passes fine with code A but not B (it doesn't rollback with B). The problem is that the A and B codes are quite different so i'm having a hard time spotting what's wrong.
Using the logs, i see something very suspicious:
-on node A, there are calls to
1/"afterCompletion(STATUS_ROLLEDBACK) called on Synchronization:org.jboss.cache.pojo.interceptors.PojoTxSynchronizationHandler"
2/"afterCompletion(STATUS_ROLLEDBACK) called on Synchronization:tx=com.atomikos.icatch.jta.TransactionImp@nnnn, handlers=........"
-on node B, there are only calls to 1 but not 2.
That sound like a significant difference, so maybe someone here can give me a clue?
chtimi 
Tuesday, June 15, 2010
Tuesday, June 15, 2010
Other recent topics