Node X Disconnected

[Scott Tully]

I am running 4.1.14 (3 replicas with 6 ndb's on 3 hosts, and 3 api nodes) and have a question about something i see in my manager log. What causes a Node to be Disconected without warning or missed heartbeats?

Here's some log (node 11 is the Master data node, but it happens with api nodes too):

2005-10-02 13:27:23 [MgmSrvr] INFO -- Node 11: Local checkpoint 125 started. Keep GCI = 112153 oldest restorable GCI = 112168
2005-10-02 13:53:02 [MgmSrvr] ALERT -- Node 1: Node 11 Disconnected
2005-10-02 13:53:02 [MgmSrvr] ALERT -- Node 13: Node 11 Disconnected
2005-10-02 13:53:02 [MgmSrvr] ALERT -- Node 21: Node 11 Disconnected
2005-10-02 13:53:02 [MgmSrvr] ALERT -- Node 41: Node 11 Disconnected
2005-10-02 13:53:02 [MgmSrvr] ALERT -- Node 42: Node 11 Disconnected
2005-10-02 13:53:02 [MgmSrvr] ALERT -- Node 23: Node 11 Disconnected
2005-10-02 13:53:02 [MgmSrvr] INFO -- Node 23: Communication to Node 11 closed
2005-10-02 13:53:02 [MgmSrvr] INFO -- Node 13: Communication to Node 11 closed
2005-10-02 13:53:02 [MgmSrvr] INFO -- Node 23: Communication to Node 11 closed
2005-10-02 13:53:02 [MgmSrvr] INFO -- Node 41: Communication to Node 11 closed
2005-10-02 13:53:02 [MgmSrvr] INFO -- Node 42: Communication to Node 11 closed
2005-10-02 13:53:02 [MgmSrvr] ALERT -- Node 13: Arbitration check won - node group majority
2005-10-02 13:53:02 [MgmSrvr] INFO -- Node 13: President restarts arbitration thread [state=6]
2005-10-02 13:53:02 [MgmSrvr] INFO -- Node 13: GCP Take over started
2005-10-02 13:53:02 [MgmSrvr] INFO -- Node 21: Communication to Node 11 closed
2005-10-02 13:53:02 [MgmSrvr] INFO -- Node 13: LCP Take over started

..........


Node 11 is fine... why is it sending a disconnect signal? Is this a bug, or a configuration issue?

Thanks!
Scott