holdenk commented on a change in pull request #29014:
URL: https://github.com/apache/spark/pull/29014#discussion_r456701507



##########
File path: 
core/src/main/scala/org/apache/spark/scheduler/ExecutorLossReason.scala
##########
@@ -54,9 +54,14 @@ private [spark] object LossReasonPending extends 
ExecutorLossReason("Pending los
 /**
  * @param _message human readable loss reason
  * @param workerLost whether the worker is confirmed lost too (i.e. including 
shuffle service)
+ * @param causedByApp whether the loss of the slave is the fault of the 
running app. (assumed true

Review comment:
       replace slave with executor

##########
File path: core/src/main/scala/org/apache/spark/scheduler/DAGScheduler.scala
##########
@@ -1767,8 +1767,13 @@ private[spark] class DAGScheduler(
 
           // TODO: mark the executor as failed only if there were lots of 
fetch failures on it
           if (bmAddress != null) {
-            val hostToUnregisterOutputs = if 
(env.blockManager.externalShuffleServiceEnabled &&
-              unRegisterOutputOnHostOnFetchFailure) {
+            val externalShuffleServiceEnabled = 
env.blockManager.externalShuffleServiceEnabled
+            val isHostDecommissioned = taskScheduler
+              .getExecutorDecommissionInfo(bmAddress.executorId)
+              .exists(_.isHostDecommissioned)

Review comment:
       So would we want to do this unless local fetch is disabled (which since 
it isn't merged yet it will be) and the executor is decommissioned or the the 
host is decommissioned? Should we use isShuffleLost here instead?

##########
File path: 
core/src/main/scala/org/apache/spark/scheduler/ExecutorDecommissionInfo.scala
##########
@@ -0,0 +1,34 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements.  See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License.  You may obtain a copy of the License at
+ *
+ *    http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.spark.scheduler
+
+/**
+ * Provides more detail when an executor is being decommissioned.
+ * @param message Human readable reason for why the decommissioning is 
happening.
+ * @param isHostDecommissioned Whether the host (aka the `node` or `worker` in 
other places) is
+ *                             being decommissioned too. Used to infer if the 
shuffle data might
+ *                             be lost if external shuffle service is enabled.
+ */
+private[spark]
+case class ExecutorDecommissionInfo(message: String, isHostDecommissioned: 
Boolean) {
+  // Shuffle files are lost when an extecutor is decommissioned if:
+  // - No external shuffle service is present and executor is serving the 
shuffle data
+  // - It's known that the node (and thus the resident external shuffle 
service) is also lost
+  def isShuffleLost(externalShuffleServiceEnabled: Boolean): Boolean =
+    !externalShuffleServiceEnabled || isHostDecommissioned

Review comment:
       I think we want to add the check here for the local shuffle block 
serving as well (once that's integrated). Maybe file a follow up JIRA that 
depends on both of those JIRAs for tracking.




----------------------------------------------------------------
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:
[email protected]



---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to