AI-assisted — how our quizzes are made
Core Kafka model and semantics: topics, partitions, offsets, records, brokers, replication, leaders/followers, keys, ordering, retention, compaction, and the architectural trade-offs backend engineers need to understand.
25 questionsProducer batching and acks, retries, idempotent producers, duplicates, ordering under retries, Kafka transactions, exactly-once semantics, transactional.id, fencing, and integrating Kafka writes with database workflows.
25 questionsOffsets and commits, consumer groups, partition assignment, rebalances, cooperative rebalancing, max.poll.interval, session timeouts, lag, parallelism, ordering constraints, retry behavior, and robust consumer design.
25 questionsEvent design, keys and partitioning strategy, schema compatibility, Avro/Protobuf/JSON trade-offs, Schema Registry concepts, tombstones, compaction-aware models, event versioning, and contract evolution across services.
25 questionsPoison messages, retry topics, backoff, DLQs, idempotent consumers, duplicate handling, consumer lag, throughput/latency metrics, stuck partitions, hot keys, broker/client failure symptoms, and diagnosing production incidents.
25 questionsPartition scaling, replication factor, ISR, leader election, min.insync.replicas, rack awareness, retention sizing, compaction trade-offs, throughput bottlenecks, broker/client configs, and operational knowledge expected from senior backend engineers.
25 questionsKafka guarantees record ordering at what scope?
Within a partition
Across an entire topic regardless of partitions
Across all topics in a cluster
Only within a producer process
What problem does cooperative/incremental rebalancing aim to reduce compared with eager rebalancing?
The amount of partition revocation and stop-the-world disruption during membership changes
Producer serialization cost
Topic retention size
Replication traffic between brokers
The same topic is consumed by two different consumer groups. What happens?
Each group can independently process all records and maintain its own offsets
They split each partition between the groups
Only the oldest group receives records
Kafka merges their committed offsets