[ https://issues.jboss.org/browse/WFLY-6215?page=com.atlassian.jira.plugin.... ] Ryan Emerson updated WFLY-6215: ------------------------------- Description: Issue first encountered by customer in EAP5 (JBPAPP-11239). Assume a local SFSB, B1, with container managed transactions and a calling SFSB, B2, which has bean managed transactions. B1 implements the SessionSynchronization interface. When B2's bean managed transaction timesout whilst executing B1.exampleMethod(), B2's transaction aborts and the transaction reaper thread calls B1.afterCompletion(). However, the original worker thread that was executing B1.exampleMethod() continues to execute B1.exampleMethod() until it has completed. Hence it is possible for the B1 SFSB to be accessed concurrently. It is also possible for B1.afterCompletion() to finish executing before B1.exampleMethod(). was: Issue first encountered by customer in EAP5 (JBPAPP-11239). Assume a local SFSB, B1, with container managed transactions and a calling SFSB, B2, which has bean managed transactions. B1 implements the SessionSynchronization interface. When B2's bean managed transaction timesout whilst executing B1.exampleMethod(), B2's transaction aborts and the transaction reaper thread calls B1.afterCompletion(). However, the original worker thread that was executing B1.exampleMethod() continues to execute B1.exampleMethod() until it has completed. Hence it is possible for the B1 SFSB to be accessed concurrently. -- This message was sent by Atlassian JIRA (v6.4.11#64026)