This issue is branched from
MB-45869, please take logs from that ticket.
Rebalance at 2021-04-21T04:04:51.677-07:00 fails because of PrepareTopologyChange timeout:
Which is a call to eventing on 172.23.123.27
On 172.23.123.27 I see that eventing created a lot (~2000) of streaming connections against ns_server around that time, which I believe could lead to a timeout.
Please investigate why there are so many streaming connections.