This applies to kafka and kinesis connections but is also used when Kadeck nodes talk to each other. It's possible that in slow networks and with a lot of data coming from the brokers that 60 secs might not be sufficient. Adjust accordingly.

While the above is mostly about connecting, this one is about transferring record data from the broker.

Max time waited by the client to the broker until it gives up trying to poll "Data Poll Max Record (see below)" number of records.

At which interval does the client poll the broker for data.

This setting allowed to set a TTL to the cache of streams returned by Kinesis. Until the TTL elapses, loading the list of streams will return the value cached.