Two brokers can go down, and one replica will still be able to receive and serve data

First local cache is checked for the message schema. In case of cache miss, schema is pulled from the schema registry. An exception will be thrown in the Schema Registry does not have the schema (which should never happen if you set it up properly)

Seehttps://docs.confluent.io/current/streams/developer-guide/dsl-api.html#joining

The Confluent Schema Registry is your safeguard against incompatible schema changes and will be the component that ensures no breaking schema evolution will be possible. Kafka Brokers do not look at your payload and your payload schema, and therefore will not reject data

Calling close() on consumer immediately triggers a partition rebalance as the consumer will not be available anymore.