Unit 5: Spark Streaming with Apache Kafka - Practice Quiz

1 What is Apache Kafka primarily used for?

2 What is a topic in Apache Kafka?

3 Which description best matches an event in Kafka?

4 What is a Kafka broker?

5 Why is a Kafka topic divided into partitions?

6 What is the main purpose of replication in a Kafka cluster?

7 Which software platform is commonly required to run Apache Kafka?

8 What should usually be configured before starting a Kafka broker?

9 Which Spark feature is commonly used to process streaming data from Kafka?

10 In a Spark and Kafka integration, what role does Kafka usually play?

11 Which information does Spark need to connect to a Kafka cluster?

12 What does a Kafka producer do?

13 What does a Kafka consumer do?

14 What is the purpose of a Kafka consumer group?

15 Which sequence represents a basic Kafka data pipeline?

16 What is the role of Kafka in a data pipeline?

17 What does an offset identify in a Kafka partition?

18 Which ingestion pattern processes events soon after they are generated?

19 Which ingestion pattern collects data over a period before processing it together?

20 Which ingestion pattern is suitable for continuously arriving sensor readings?

21 An application requires all events for the same customer to be processed in the order in which they were produced. Which Kafka design best supports this requirement?

22 A Kafka topic has a retention period of seven days. A consumer remains offline for nine days and then resumes from its last committed offset. What is the most likely outcome?

23 A topic contains six partitions, and each partition receives approximately 500 messages per second. Assuming an even distribution, what is the approximate total ingestion rate?

24 A partition has replication factor 3. Its leader broker fails, but two replicas remain in the in-sync replica set. What should the cluster normally do?

25 A Kafka topic has 12 partitions and is consumed by one consumer group containing 15 active consumers. How many consumers can process records simultaneously at most?

26 Why are partition replicas usually distributed across different Kafka brokers?

27 When configuring a multi-broker Kafka cluster, which combination is required for the brokers to operate as members of the same cluster?

28 Kafka starts successfully, but remote clients receive broker addresses such as localhost:9092 and cannot connect. Which configuration should be checked first?

29 After installing Kafka, an administrator wants to verify end-to-end messaging before connecting Spark. Which test is most appropriate?

30 A Spark Structured Streaming application must resume from the correct Kafka offsets after a driver restart. Which feature should be configured?

31 A Kafka topic has eight partitions, but a Spark streaming job is configured with 20 executor cores. What primarily limits the initial parallelism of Kafka reads?

32 In Spark Structured Streaming, Kafka records expose key and value as binary columns. What should the application normally do before applying string or JSON operations?

33 A Spark streaming query reads Kafka events containing event timestamps and performs a windowed aggregation. Which feature helps limit state for events arriving late?

34 A producer sets acks=all for a replicated Kafka topic. What is the main effect?

35 A consumer processes a record and commits its offset before writing the result to a database. The database write then fails. What delivery problem can occur?

36 Two independent applications must each receive every event from the same Kafka topic. How should their consumers be configured?

37 In a pipeline producers → Kafka → Spark → database, the database becomes temporarily slow. What Kafka capability most directly helps absorb the slowdown?

38 Several producers send JSON events to a shared Kafka topic, but they use incompatible field types for the same attribute. Which pipeline practice best prevents processing failures?

39 A company must ingest continuous sensor readings with low latency while also importing one large historical file each night. Which design is most appropriate?

40 A corrected Spark transformation must be applied to the last three days of Kafka events without asking producers to resend them. Which ingestion pattern enables this?

41 A producer publishes records with keys A, B, and C to a topic with six partitions using Kafka's default key-based partitioning. What property is guaranteed if the number of partitions remains unchanged?

42 A Kafka topic has retention.ms=86400000 and retention.bytes=10737418240. The topic contains 20 GB of log data, and all segments are older than the retention time. Which statement best describes retention behavior?

43 A producer uses acks=all and enables idempotence. A record is acknowledged after being written to all in-sync replicas, but the leader fails before the producer receives the response. What is the safest producer behavior?

44 A topic has replication factor 3 and min.insync.replicas=2. The producer uses acks=all. One replica becomes unavailable, but the leader and one follower remain in the ISR. What is the expected result for new writes?

45 A topic's partitions are distributed across three brokers, but each partition's leader and its first replica are placed in the same rack. What is the main unresolved availability risk?

46 A consumer group has eight consumers assigned to a topic with five partitions. Assuming no idle or standby assignment behavior, what is the most accurate result?

47 A Kafka installation is configured with a listener advertised as PLAINTEXT://localhost:9092, and remote Spark executors attempt to connect using the broker metadata. The TCP port is reachable, but clients still fail after metadata retrieval. What is the most likely cause?

48 An administrator starts a Kafka broker in KRaft mode but supplies only a broker process role and no controller quorum configuration. Which outcome is expected in a standard combined-node deployment?

49 A broker is configured with a log directory on a nearly full disk. Kafka begins rejecting new records even though topic-level retention appears correct. Which operational explanation is most plausible?

50 A Structured Streaming query reads Kafka records, performs a deterministic transformation, and writes to an idempotent external sink. The query fails after the sink write but before the Spark checkpoint is updated. What behavior should the sink support?

51 A Spark Structured Streaming job reads from Kafka with startingOffsets=latest, but the query is restarted after its checkpoint directory is deleted. New records arrive during startup. Which result is expected?

52 A Kafka topic has 100 partitions, but a Spark streaming query is limited by maxOffsetsPerTrigger to 10,000 records. If partition rates are highly skewed, what limitation can this setting introduce?

53 A Spark job reads Kafka data, applies a stateful aggregation, and is configured with a short trigger interval. Processing time regularly exceeds the trigger interval. What is the most likely consequence?

54 Two consumers belong to the same Kafka consumer group and process different partitions. Consumer A commits offset 120 for partition 0, while Consumer B fails before committing offset 85 for partition 1. After rebalance, what does the group normally resume from?

55 A consumer processes a record for 40 seconds, while max.poll.interval.ms is 30 seconds. The consumer continues sending heartbeats but does not call poll during processing. What is the likely result?

56 A producer sends keyed events to a partition. Retries are enabled, but idempotence is disabled. A transient acknowledgment failure causes a retry after the first append succeeded. Which issue is possible?

57 A pipeline consumes clickstream events, enriches them, and republishes results to a second Kafka topic. The enrichment service is nondeterministic and the output topic is compacted. Which design most reliably supports replay without inconsistent final state?

58 A Kafka pipeline has a source topic, a stream processor, and a sink topic. The processor commits source offsets before publishing sink records. If it crashes immediately afterward, what failure mode is most likely?

59 A pipeline uses a compacted Kafka topic for the latest customer profile. A tombstone for customer C7 is produced, but downstream consumers start after the tombstone has been removed. What conclusion is correct?

60 A Spark streaming application must ingest a bursty Kafka workload while preventing a temporary spike from exhausting executor memory. Which strategy is most appropriate?