Federated queue not ack message and duplicate

Hello. I have a 3 instances of artemis Cluster (Artemis01 (Master) -> Artemis02 (Slave) and Artemis03 (standalone).

Artemis01 and Artemis02 have upstream federated queue to Artemis03. I run a JMS consumer on Artemis01 (Master) TEST.FEDERATED address (ANYCAST). I send message with producer to Artemis03 TEST.FEDERATE address (ANYCAST). This message auto delivered with federated queue to my consumer on Artemis01 (Master) and i ACK this message. After this i kill process of Artemis01 (Master) with command kill -9 artemis01_pid. My consumer auto reconnect to slave Artemis02 (Slave) and MY ACK message returned back to Artemis 03!  Why? If i received this and ack. After message returned back to Artemis03 federated queue did second delivery this message to my JMS consumer on Artemis 02 (Slave). I think this is incorrect behaviour. I add gif image with details. If i not kill artemis 01 process and use command artemis stop then all work fine and message not returned back and deliver twice! I think this is bug. Please fix it.