Auto-commit vs manual commit in Kafka - Practice Questions

Challenge - 5 Problems

❓ Predict Output

intermediate

What is the output when auto-commit is enabled?

Consider a Kafka consumer configured with enable.auto.commit=true and auto.commit.interval.ms=1000. The consumer reads messages continuously. What happens if the consumer crashes after processing some messages but before the next auto-commit?

Kafka

Properties props = new Properties();
props.put("bootstrap.servers", "localhost:9092");
props.put("group.id", "test-group");
props.put("enable.auto.commit", "true");
props.put("auto.commit.interval.ms", "1000");
props.put("key.deserializer", "org.apache.kafka.common.serialization.StringDeserializer");
props.put("value.deserializer", "org.apache.kafka.common.serialization.StringDeserializer");

KafkaConsumer<String, String> consumer = new KafkaConsumer<>(props);
consumer.subscribe(Arrays.asList("my-topic"));

while (true) {
    ConsumerRecords<String, String> records = consumer.poll(Duration.ofMillis(100));
    for (ConsumerRecord<String, String> record : records) {
        System.out.printf("offset = %d, key = %s, value = %s\n", record.offset(), record.key(), record.value());
        // Process message
    }
    // No manual commit here
}

A The consumer will throw an exception if it crashes before committing offsets.

B Offsets are committed manually by the user after processing each message.

C All processed messages are guaranteed committed immediately after processing each message.

D Messages processed before the crash may be reprocessed after restart because the offset was not committed yet.

💡 Hint

Think about how auto-commit interval affects when offsets are saved.

With auto-commit enabled, offsets are committed periodically (every 1000 ms here). If the consumer crashes before the next commit, some processed messages may not have their offsets saved, causing reprocessing after restart.

❓ Predict Output

intermediate

What happens when manual commit is used incorrectly?

A Kafka consumer is configured with enable.auto.commit=false. The code processes messages but never calls commitSync() or commitAsync(). What is the effect on message processing?

Kafka

Properties props = new Properties();
props.put("bootstrap.servers", "localhost:9092");
props.put("group.id", "test-group");
props.put("enable.auto.commit", "false");
props.put("key.deserializer", "org.apache.kafka.common.serialization.StringDeserializer");
props.put("value.deserializer", "org.apache.kafka.common.serialization.StringDeserializer");

KafkaConsumer<String, String> consumer = new KafkaConsumer<>(props);
consumer.subscribe(Arrays.asList("my-topic"));

while (true) {
    ConsumerRecords<String, String> records = consumer.poll(Duration.ofMillis(100));
    for (ConsumerRecord<String, String> record : records) {
        System.out.printf("offset = %d, key = %s, value = %s\n", record.offset(), record.key(), record.value());
        // Process message
    }
    // No commit called here
}

A Messages will be reprocessed after restart because offsets are never committed.

B Offsets are committed asynchronously by default even if auto-commit is false.

C The consumer will throw a runtime exception due to missing commit calls.

D Messages will be processed exactly once because offsets are committed automatically.

💡 Hint

Consider what happens if offsets are never saved.

When auto-commit is disabled and no manual commit is done, offsets are never saved. After restart, the consumer will re-read messages from the last committed offset, causing duplicates.

🧠 Conceptual

advanced

Why choose manual commit over auto-commit in Kafka consumers?

Which of the following is the best reason to use manual commit instead of auto-commit in Kafka consumers?

A To ensure offsets are committed only after successful processing of messages, avoiding data loss or duplication.

B To reduce network traffic by committing offsets less frequently.

C Because auto-commit is deprecated and no longer supported in Kafka.

D To automatically retry failed message processing without manual intervention.

💡 Hint

Think about reliability and message processing guarantees.

Manual commit allows the consumer to commit offsets only after messages are fully processed, which helps avoid data loss or duplicate processing. Auto-commit commits offsets periodically regardless of processing success.

❓ Predict Output

advanced

What error occurs if commitSync() is called outside poll loop?

Given a Kafka consumer with enable.auto.commit=false, what happens if commitSync() is called before any poll() call?

Kafka

Properties props = new Properties();
props.put("bootstrap.servers", "localhost:9092");
props.put("group.id", "test-group");
props.put("enable.auto.commit", "false");
props.put("key.deserializer", "org.apache.kafka.common.serialization.StringDeserializer");
props.put("value.deserializer", "org.apache.kafka.common.serialization.StringDeserializer");

KafkaConsumer<String, String> consumer = new KafkaConsumer<>(props);

// Calling commitSync before poll
consumer.commitSync();

A Throws TimeoutException because commitSync waits for poll to assign partitions.

B No error; commitSync commits the last committed offsets successfully.

C Throws IllegalStateException because no offsets have been fetched yet.

D Throws NullPointerException due to missing consumer records.

💡 Hint

Consider what commitSync needs before it can commit offsets.

commitSync requires the consumer to have assigned partitions and fetched offsets via poll. Calling it before poll causes IllegalStateException because no offsets are known yet.

🚀 Application

expert

How to implement exactly-once processing with manual commit in Kafka consumer?

You want to process Kafka messages exactly once with manual commit. Which approach below correctly ensures that offsets are committed only after successful processing?

A Process messages, then call <code>commitSync()</code> immediately after processing each message inside the for-loop.

B Process all messages in the batch, then call <code>commitSync()</code> once after the for-loop completes successfully.

C Call <code>commitSync()</code> before processing messages to reserve offsets.

D Enable auto-commit and rely on Kafka to commit offsets after processing.

💡 Hint

Think about committing offsets only after all messages are processed successfully.

Committing offsets once after processing the entire batch ensures that no message is marked as processed before it actually is. Committing inside the loop risks committing offsets for unprocessed messages if a failure occurs.