Spark on YARN-client is suitable for production environments because you can see the output of APP faster.

Question

Options

• ATrue
• BFalse

• A
  18% (6)
• B
  82% (28)

Explanation

B is correct because YARN-client mode is designed for development and interactive debugging, not production workloads. In client mode, the Spark driver runs on the submitting machine - if that machine disconnects or crashes, the entire job fails. This lack of resilience makes it fundamentally unsuitable for production.

Why A is wrong: While it's true that YARN-client mode does show application output more directly (since the driver runs locally), this is a debugging convenience, not a production virtue. The speed-of-output benefit is real but irrelevant to production requirements like reliability and fault tolerance.

The core production issue: In production, YARN-cluster mode is preferred because the driver runs inside YARN's ApplicationMaster on a cluster node, meaning the job lifecycle is managed entirely by the cluster - independent of the submitting client's uptime.

Memory tip: Think "Client = Console (dev), Cluster = Controlled (prod)" - client mode keeps output on your console for easy debugging; cluster mode hands control to YARN for reliable, unattended execution.

No community discussion yet for this question.