InboundBridgeRecoveryManager.checkXid does not correctly check nodename of the xid. Thus where multiple app server nodes share a db, bottom-up recovery can abort branches that don't belong to the node. We miss more information in server logs for investigation of XTS processing in case of troubles. There should be added more trace logging messages to understand what SOAP messages are processed and what are tasks pushed to TaskManager queue.
Verified with EAP 6.4.24.CP.CR1
This product has been discontinued or is no longer tracked in Red Hat Bugzilla.