Skip to content

ShuffleBlockResolver

= [[ShuffleBlockResolver]] ShuffleBlockResolver

ShuffleBlockResolver is an <> of <> that storage:BlockManager.md[BlockManager] uses to <> for a logical shuffle block identifier (i.e. map, reduce, and shuffle).

NOTE: Shuffle block data files are often referred to as map outputs files.

[[implementations]] NOTE: shuffle:IndexShuffleBlockResolver.md[IndexShuffleBlockResolver] is the default and only known ShuffleBlockResolver in Apache Spark.

[[contract]] .ShuffleBlockResolver Contract [cols="1m,3",options="header",width="100%"] |=== | Method | Description

| getBlockData a| [[getBlockData]]

[source, scala]

getBlockData( blockId: ShuffleBlockId): ManagedBuffer


Retrieves the data (as a ManagedBuffer) for the given storage:BlockId.md#ShuffleBlockId[block] (a tuple of shuffleId, mapId and reduceId).

Used when BlockManager is requested to retrieve a storage:BlockManager.md#getLocalBytes[block data from a local block manager] and storage:BlockManager.md#getBlockData[block data]

| stop a| [[stop]]

[source, scala]

stop(): Unit

Stops the ShuffleBlockResolver

Used when SortShuffleManager is requested to SortShuffleManager.md#stop[stop]

|===