Which of the following statements about slots is incorrect?

Question

Options

• ASlots are the most granular level of execution in the Spark execution hierarchy.
• BSlots are resources for parallelization within an executor.
• CTasks are assigned to slots for computation.
• DThere can be more slots than tasks.
• EThere must be at least as many slots as there are executors.

• A
  90% (36)
• B
  3% (1)
• C
  3% (1)
• E
  5% (2)

Explanation

Option A is incorrect because tasks, not slots, are the most granular level of execution in Spark's hierarchy (Job → Stage → Task). Slots are not execution units themselves - they are parallel execution resources (threads/cores) within an executor that tasks are assigned to.

Why the distractors are correct statements (and thus wrong answer choices):

• B is true: slots represent parallelism capacity within a single executor - an executor with 4 cores has 4 slots.
• C is true: Spark's scheduler assigns individual tasks to available slots, which then perform the computation.
• D is true: if you provision 16 slots but a stage only has 3 tasks, 13 slots sit idle - slots can outnumber tasks.
• E is true: since each executor has at least 1 slot (set by spark.executor.cores), the slot count must be ≥ the executor count.

Memory tip: Think of slots as parking spaces and tasks as cars. The parking spaces (slots) belong to a garage (executor) and enable parallel parking (parallelism), but the cars (tasks) are the things actually doing the driving - they're the granular units of work. A car needs a space, not the other way around.

No community discussion yet for this question.