After Fabric container got reconnected to Zookeeper, the Fabric discovery transport does not get updated on brokers joining or leaving the Fabric

Issue

We are running a Fabric managed environment with multiple broker instances running under the same Fabric broker group name.
Clients use the Fabric discovery transport, discovery:(fabric:xxx) to discover broker instances and their listening ports at runtime.

We have had an occasion where a Fabric managed container got disconnected and reconnected to Zookeeper (due to problems at network layer) and thereafter its Fabric discovery transport did not receive any update on brokers joining or leaving the broker group.

As a result, broker clients running on this container were at some stage unable to connect to any brokers, since all broker instances restarted on a new port in the meantime.

To resolve the problem we had to restart this client container. Upon restart it successfully discovered the new broker urls.