Occasionally our BlackTie CI test run is failing when marking a transaction rollback only because it is getting the wrong exception back from the transaction manager

16:11:10,296 WARN [arjLoggerI18N] [com.arjuna.ats.arjuna.coordinator.TransactionReaper_18] - TransactionReaper::check timeout for TX -53ee7cfb:10e2:4c0fa834:22ec in state RUN
[exec] 2010-06-09 16:11:10,296 [0x0000123c] WARN (TxLogControl :153 ) - rollback_only: INVALID_TRANSACTION minor: 20001
16:11:10,328 WARN [arjLoggerI18N] [com.arjuna.ats.arjuna.coordinator.TransactionReaper_7] - TransactionReaper::doCancellations worker Thread[Thread-18,5,jboss] successfully canceled TX -53ee7cfb:10e2:4c0fa834:22ec

The second log message comes from the C client (the other two from the reaper). The CORBA System exception major/minor code represents INVALID_TRANSACTION/INACTIVE_TRANSACTION which according to the OTS spec is incorrect.

A look at the code shows a race condition between a BlackTie C client marking a transaction abort only and the Transaction Reaper cancelling it (due to a timeout being reached). The reaper eventually winds up in ControlImple.destroy() where the CORBA object representing the transaction is destroyed. There is a window where the C client request reaches JacORB just before the reaper has had a chance to destroy the CORBA object resulting in the wrong exception being returned.