holdenk commented on a change in pull request #28370:
URL: https://github.com/apache/spark/pull/28370#discussion_r419761212



##########
File path: 
core/src/main/scala/org/apache/spark/storage/BlockManagerMasterEndpoint.scala
##########
@@ -299,6 +310,39 @@ class BlockManagerMasterEndpoint(
     blockManagerIdByExecutor.get(execId).foreach(removeBlockManager)
   }
 
+  /**
+   * Decommission the given Seq of blockmanagers
+   *    - Adds these block managers to decommissioningBlockManagerSet Set
+   *    - Sends the DecommissionBlockManager message to each of the 
[[BlockManagerSlaveEndpoint]]
+   */
+  def decommissionBlockManagers(blockManagerIds: Seq[BlockManagerId]): 
Future[Seq[Unit]] = {
+    val newBlockManagersToDecommission = 
blockManagerIds.toSet.diff(decommissioningBlockManagerSet)
+    val futures = newBlockManagersToDecommission.map { blockManagerId =>
+      decommissioningBlockManagerSet.add(blockManagerId)
+      val info = blockManagerInfo(blockManagerId)
+      info.slaveEndpoint.ask[Unit](DecommissionBlockManager)
+    }
+    Future.sequence{ futures.toSeq }
+  }
+
+  /**
+   * Returns a Seq of ReplicateBlock for each RDD block stored by given 
blockManagerId
+   * @param blockManagerId - block manager id for which ReplicateBlock info is 
needed
+   * @return Seq of ReplicateBlock
+   */
+  private def getReplicateInfoForRDDBlocks(blockManagerId: BlockManagerId): 
Seq[ReplicateBlock] = {
+    val info = blockManagerInfo(blockManagerId)
+
+    val rddBlocks = info.blocks.keySet().asScala.filter(_.isRDD)

Review comment:
       I think storing a separate list of RDD blocks would be more expensive, 
we also only this call occasionally compared to the other operations in the 
block manager so I think it's ok if filtering the keyset is expensive (although 
it probably is not  very expensive).




----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

For queries about this service, please contact Infrastructure at:
us...@infra.apache.org



---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org

Reply via email to