ZooKeeper role (and KRaft replacement) in Kafka - Practice Problems & Coding Challenges

Challenge - 5 Problems

❓ Predict Output

intermediate

What is the output of this Kafka cluster state query?

Consider a Kafka cluster running with ZooKeeper. You run the command to describe the cluster state and get the following output snippet:

Controller ID: 1
Broker IDs: 1, 2, 3
ZooKeeper connected: true

What does this output tell you about the cluster?

A The cluster has three brokers, broker 3 is the controller, and ZooKeeper is connected.

B The cluster has one broker, ZooKeeper is disconnected, and broker 1 is the controller.

C The cluster has three brokers, broker 1 is the controller, and ZooKeeper is connected.

D The cluster has three brokers, no controller is elected, and ZooKeeper is disconnected.

💡 Hint

Look for the Controller ID and ZooKeeper connection status in the output.

The output shows Controller ID as 1, meaning broker 1 is the controller. Broker IDs list three brokers. ZooKeeper connected is true, so ZooKeeper is active.

🧠 Conceptual

intermediate

What is the main role of ZooKeeper in a Kafka cluster?

In Kafka clusters that use ZooKeeper, what is ZooKeeper primarily responsible for?

A Storing all Kafka messages permanently.

B Managing cluster metadata, leader election, and configuration storage.

C Handling client requests and producing messages.

D Running Kafka Connect and Kafka Streams applications.

💡 Hint

Think about what coordination tasks are needed in a distributed system.

ZooKeeper manages metadata like broker info, topic configurations, and elects leaders for partitions. It does not store messages or handle client requests directly.

❓ Predict Output

advanced

What error occurs when Kafka tries to start without ZooKeeper in legacy mode?

You start a Kafka broker configured to use ZooKeeper but the ZooKeeper service is down. The broker logs show:

ERROR Fatal error during KafkaServer startup. Exiting.
Caused by: org.apache.zookeeper.KeeperException$ConnectionLossException: KeeperErrorCode = ConnectionLoss

What does this error mean?

A Kafka broker cannot connect to ZooKeeper, so it fails to start.

B Kafka broker started successfully without ZooKeeper.

C Kafka broker is running but cannot produce messages.

D Kafka broker lost connection to clients.

💡 Hint

Look at the exception type and message about connection loss.

The error shows Kafka cannot connect to ZooKeeper, which is required in legacy mode, so startup fails.

🧠 Conceptual

advanced

How does KRaft mode replace ZooKeeper in Kafka?

Kafka introduced KRaft mode to remove the dependency on ZooKeeper. What is the key difference in how KRaft manages metadata compared to ZooKeeper?

A KRaft uses an internal quorum of Kafka brokers to manage metadata without an external service.

B KRaft requires an external ZooKeeper cluster to run alongside Kafka brokers.

C KRaft stores metadata in a separate database outside Kafka and ZooKeeper.

D KRaft disables metadata management and relies on clients to handle it.

💡 Hint

Think about how Kafka can manage metadata internally without ZooKeeper.

KRaft mode replaces ZooKeeper by using a built-in consensus quorum among Kafka brokers to manage metadata internally.

🚀 Application

expert

What happens to Kafka cluster availability during KRaft controller failure?

In a Kafka cluster running in KRaft mode with 3 controller nodes, one controller node fails unexpectedly. What is the expected behavior regarding cluster availability and metadata management?

A Kafka automatically switches back to ZooKeeper mode to maintain availability.

B All brokers shut down because metadata is lost.

C The entire Kafka cluster becomes unavailable until the failed controller is restored.

D The remaining controllers continue managing metadata, and the cluster remains available without interruption.

💡 Hint

Consider how consensus and fault tolerance work in KRaft mode with multiple controllers.

KRaft uses a quorum of controllers. Losing one controller in a 3-node quorum still leaves a majority to manage metadata and keep the cluster running.