Kafka / Consumer Interview Questions 2026

How does Kafka guarantee message durability? What configuration ensures messages are not lost even if a broker crashes mid-write?

Kafka consumer is falling behind (lag growing). Processing is in a tight loop with no I/O. How do you increase throughput?

You need to process Kafka messages in exactly the order they were produced for each user, but process different users in parallel. How do you configure this?

Your Kafka topic has 10 partitions. You scale your consumer group from 10 to 20 instances. What happens?

Explain Kafka's consumer group rebalancing. What triggers it and what are the performance implications?

You're consuming from a Kafka topic with 50 million messages backlog. Your consumer needs to catch up without overwhelming your database. How do you design the catch-up strategy?

You need Kafka to guarantee exactly-once message processing from producer to consumer, including writing to a database. Explain what you need to configure at each layer.

Your Kafka consumer group has 10 consumers and 10 partitions. One consumer is slower than others and keeps getting rebalances, causing other consumers to pause processing. How do you fix this?

What is Kafka log compaction and when should you use it?

How does Kafka handle message ordering? Can you guarantee order across a topic with multiple partitions?

Explain Kafka's message delivery guarantees: at-most-once, at-least-once, exactly-once. How does each affect your consumer design?

Your Kafka consumer is processing messages slowly. Messages pile up and consumer lag grows. How do you fix this?

Explain Kafka consumer groups and how they enable parallelism. What happens if you have more consumers than partitions?

Kafka guarantees message order within a partition but not across partitions. For a payments system, why does this matter, and how would you key messages to preserve per-account ordering?

You have a Kafka topic with 6 partitions and a consumer group with 10 consumer instances. What happens to the 4 extra consumers, and why is over-provisioning consumers like this a mistake?

A Kafka consumer processing payment events crashes and restarts, re-reading the last uncommitted batch. How do you make the payment-processing logic itself idempotent, not just rely on Kafka?

Kafka topics have a retention period, but consumer offsets also have their own retention. What happens if a consumer group is offline longer than the offset retention period?

Explain the difference between at-most-once, at-least-once, and exactly-once delivery semantics in Kafka. Which one requires the consumer to handle duplicate processing?