• B sends a unicast message M to A
• A drops the ACK(M) to B
• B sends no further messages to A
  --> B will keep retransmitting M to A until it gets an ACK from A, or A leaves the cluster
  --> However, A will not send an ACK(M) to B, as the logic for acking only sends an ack if we did receive a message from B and it was successfully added to the retransmit table
  --> However, the retransmitted M will not get added to A's table, as M has already been received

SOLUTION:

• Always ack a message frpm B with the highest delivered seqno for B's table