[ https://issues.jboss.org/browse/JBJCA-1138?page=com.atlassian.jira.plugin... ] Andrey Smirnov updated JBJCA-1138: ---------------------------------- Description: Seems like TransactionSynchronizer has been rolled back to state it was in August. I think this was caused by little typo in commit 469b4c303dd0d5c39f3476a5972556d8b9537b44 296: Record record = records.get(tx); But 'records' map uses TransactionKey as key and above line should be written as 296:TransactionKey key = new TransactionKey(tx); 297: Record record = records.get(key); I've checked fixed 1.0.22 and mem-leaks and deadlocks are all gone was: Seems like TransactionSynchronizer has been rolled back to state it was in August. I think this was caused by little typo in commit 469b4c303dd0d5c39f3476a5972556d8b9537b44 296: Record record = records.get(tx); But 'records' map uses TransactionKey as key and above line should be written as 296:TransactionKey key = new TransactionKey(tx); 297: Record record = records.get(key); -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira