Github user steveloughran commented on a diff in the pull request:
https://github.com/apache/spark/pull/9182#discussion_r43377401
--- Diff:
yarn/src/main/scala/org/apache/spark/scheduler/cluster/SchedulerExtensionService.scala
---
@@ -0,0 +1,155 @@
+/*
+ * 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.cluster
+
+import java.util.concurrent.atomic.AtomicBoolean
+
+import org.apache.hadoop.yarn.api.records.{ApplicationAttemptId,
ApplicationId}
+
+import org.apache.spark.util.Utils
+import org.apache.spark.{Logging, SparkContext}
+
+/**
+ * An extension service that can be loaded into a Spark YARN scheduler.
+ * A Service that can be started and stopped.
+ *
+ * 1. For implementations to be loadable by [[SchedulerExtensionServices]],
+ * they must provide an empty constructor.
+ * 2. The `stop()` operation MUST be idempotent, and succeed even if
`start()` was
+ * never invoked.
+ */
+trait SchedulerExtensionService {
+
+ /**
+ * Start the extension service. This should be a no-op if
+ * called more than once.
+ * @param binding binding to the spark application and YARN
+ */
+ def start(binding: SchedulerExtensionServiceBinding): Unit
+
+ /**
+ * Stop the service
+ * The `stop()` operation MUST be idempotent, and succeed even if
`start()` was
+ * never invoked.
+ */
+ def stop(): Unit
+}
+
+/**
+ * Binding information for a [[SchedulerExtensionService]].
+ *
+ * The attempt ID will be set if the service is started within a YARN
application master;
+ * there is then a different attempt ID for every time that AM is
restarted.
+ * When the service binding is instantiated on a client, there's no
attempt ID, as it lacks
+ * this information.
+ * @param sparkContext current spark context
+ * @param applicationId YARN application ID
+ * @param attemptId YARN attemptID -if known.
+ */
+case class SchedulerExtensionServiceBinding(
+ sparkContext: SparkContext,
+ applicationId: ApplicationId,
+ attemptId: Option[ApplicationAttemptId] = None)
+
+/**
+ * Container for [[SchedulerExtensionService]] instances.
+ *
+ * Loads Extension Services from the configuration property
+ * `"spark.yarn.services"`, instantiates and starts them.
+ * When stopped, it stops all child entries.
+ *
+ * The order in which child extension services are started and stopped
+ * is undefined.
+ *
+ */
+private[spark] class SchedulerExtensionServices extends
SchedulerExtensionService
+ with Logging {
+ private var services: List[SchedulerExtensionService] = Nil
+ private val started = new AtomicBoolean(false)
+ private var binding: SchedulerExtensionServiceBinding = _
+
+ /**
+ * Binding operation will load the named services and call bind on them
too; the
+ * entire set of services are then ready for `init()` and `start()`
calls.
+ *
+ * @param binding binding to the spark application and YARN
+ */
+ def start(binding: SchedulerExtensionServiceBinding): Unit = {
+ if (started.getAndSet(true)) {
+ logWarning("Ignoring re-entrant start operation")
+ return
+ }
+ require(binding.sparkContext != null, "Null context parameter")
+ require(binding.applicationId != null, "Null appId parameter")
+ this.binding = binding
+ val sparkContext = binding.sparkContext
+ val appId = binding.applicationId
+ val attemptId = binding.attemptId
+ logInfo(s"Starting Yarn extension services with app
${binding.applicationId}" +
+ s" and attemptId $attemptId")
+
+ services =
sparkContext.getConf.getOption(SchedulerExtensionServices.SPARK_YARN_SERVICES)
+ .map { s =>
+ s.split(",").map(_.trim()).filter(!_.isEmpty)
+ .map { sClass =>
+ val instance = Utils.classForName(sClass)
+ .newInstance()
+ .asInstanceOf[SchedulerExtensionService]
+ // bind this service
+ instance.start(binding)
+ logInfo(s"Service $sClass started")
+ instance
+ }.toList
+ }.getOrElse(Nil)
+ }
+
+ /**
+ * Get the list of services.
+ *
+ * @return a list of services; Nil until the service is started
+ */
+ def getServices: List[SchedulerExtensionService] = {
+ services
+ }
+
+ /**
+ * Stop the services; idempotent.
+ *
+ * Any
+ */
+ override def stop(): Unit = {
+ if (started.getAndSet(false)) {
+ logInfo(s"Stopping $this")
--- End diff --
added one
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
