diff options
Diffstat (limited to 'yarn')
-rw-r--r-- | yarn/src/main/scala/org/apache/spark/deploy/yarn/YarnAllocator.scala | 7 | ||||
-rw-r--r-- | yarn/src/main/scala/org/apache/spark/deploy/yarn/YarnSparkHadoopUtil.scala | 4 | ||||
-rw-r--r-- | yarn/src/main/scala/org/apache/spark/scheduler/cluster/YarnClusterScheduler.scala | 18 | ||||
-rw-r--r-- | yarn/src/main/scala/org/apache/spark/scheduler/cluster/YarnScheduler.scala (renamed from yarn/src/main/scala/org/apache/spark/scheduler/cluster/YarnClientClusterScheduler.scala) | 12 |
4 files changed, 16 insertions, 25 deletions
diff --git a/yarn/src/main/scala/org/apache/spark/deploy/yarn/YarnAllocator.scala b/yarn/src/main/scala/org/apache/spark/deploy/yarn/YarnAllocator.scala index d00f29665a..3849586c61 100644 --- a/yarn/src/main/scala/org/apache/spark/deploy/yarn/YarnAllocator.scala +++ b/yarn/src/main/scala/org/apache/spark/deploy/yarn/YarnAllocator.scala @@ -32,6 +32,8 @@ import org.apache.hadoop.yarn.client.api.AMRMClient import org.apache.hadoop.yarn.client.api.AMRMClient.ContainerRequest import org.apache.hadoop.yarn.util.RackResolver +import org.apache.log4j.{Level, Logger} + import org.apache.spark.{Logging, SecurityManager, SparkConf} import org.apache.spark.deploy.yarn.YarnSparkHadoopUtil._ import org.apache.spark.scheduler.cluster.CoarseGrainedSchedulerBackend @@ -60,6 +62,11 @@ private[yarn] class YarnAllocator( import YarnAllocator._ + // RackResolver logs an INFO message whenever it resolves a rack, which is way too often. + if (Logger.getLogger(classOf[RackResolver]).getLevel == null) { + Logger.getLogger(classOf[RackResolver]).setLevel(Level.WARN) + } + // Visible for testing. val allocatedHostToContainersMap = new HashMap[String, collection.mutable.Set[ContainerId]] diff --git a/yarn/src/main/scala/org/apache/spark/deploy/yarn/YarnSparkHadoopUtil.scala b/yarn/src/main/scala/org/apache/spark/deploy/yarn/YarnSparkHadoopUtil.scala index 4bff846123..4e39c1d580 100644 --- a/yarn/src/main/scala/org/apache/spark/deploy/yarn/YarnSparkHadoopUtil.scala +++ b/yarn/src/main/scala/org/apache/spark/deploy/yarn/YarnSparkHadoopUtil.scala @@ -17,12 +17,9 @@ package org.apache.spark.deploy.yarn -import java.lang.{Boolean => JBoolean} import java.io.File -import java.util.{Collections, Set => JSet} import java.util.regex.Matcher import java.util.regex.Pattern -import java.util.concurrent.ConcurrentHashMap import scala.collection.mutable.HashMap @@ -32,7 +29,6 @@ import org.apache.hadoop.security.Credentials import org.apache.hadoop.security.UserGroupInformation import org.apache.hadoop.yarn.conf.YarnConfiguration import org.apache.hadoop.yarn.api.records.{Priority, ApplicationAccessType} -import org.apache.hadoop.yarn.util.RackResolver import org.apache.hadoop.conf.Configuration import org.apache.spark.{SecurityManager, SparkConf} diff --git a/yarn/src/main/scala/org/apache/spark/scheduler/cluster/YarnClusterScheduler.scala b/yarn/src/main/scala/org/apache/spark/scheduler/cluster/YarnClusterScheduler.scala index be55d26f1c..72ec4d6b34 100644 --- a/yarn/src/main/scala/org/apache/spark/scheduler/cluster/YarnClusterScheduler.scala +++ b/yarn/src/main/scala/org/apache/spark/scheduler/cluster/YarnClusterScheduler.scala @@ -17,33 +17,17 @@ package org.apache.spark.scheduler.cluster -import org.apache.hadoop.yarn.util.RackResolver - import org.apache.spark._ import org.apache.spark.deploy.yarn.ApplicationMaster -import org.apache.spark.scheduler.TaskSchedulerImpl -import org.apache.spark.util.Utils /** * This is a simple extension to ClusterScheduler - to ensure that appropriate initialization of * ApplicationMaster, etc is done */ -private[spark] class YarnClusterScheduler(sc: SparkContext) extends TaskSchedulerImpl(sc) { +private[spark] class YarnClusterScheduler(sc: SparkContext) extends YarnScheduler(sc) { logInfo("Created YarnClusterScheduler") - // Nothing else for now ... initialize application master : which needs a SparkContext to - // determine how to allocate. - // Note that only the first creation of a SparkContext influences (and ideally, there must be - // only one SparkContext, right ?). Subsequent creations are ignored since executors are already - // allocated by then. - - // By default, rack is unknown - override def getRackForHost(hostPort: String): Option[String] = { - val host = Utils.parseHostPort(hostPort)._1 - Option(RackResolver.resolve(sc.hadoopConfiguration, host).getNetworkLocation) - } - override def postStartHook() { ApplicationMaster.sparkContextInitialized(sc) super.postStartHook() diff --git a/yarn/src/main/scala/org/apache/spark/scheduler/cluster/YarnClientClusterScheduler.scala b/yarn/src/main/scala/org/apache/spark/scheduler/cluster/YarnScheduler.scala index 2fa24cc433..4ebf3af12b 100644 --- a/yarn/src/main/scala/org/apache/spark/scheduler/cluster/YarnClientClusterScheduler.scala +++ b/yarn/src/main/scala/org/apache/spark/scheduler/cluster/YarnScheduler.scala @@ -19,14 +19,18 @@ package org.apache.spark.scheduler.cluster import org.apache.hadoop.yarn.util.RackResolver +import org.apache.log4j.{Level, Logger} + import org.apache.spark._ import org.apache.spark.scheduler.TaskSchedulerImpl import org.apache.spark.util.Utils -/** - * This scheduler launches executors through Yarn - by calling into Client to launch the Spark AM. - */ -private[spark] class YarnClientClusterScheduler(sc: SparkContext) extends TaskSchedulerImpl(sc) { +private[spark] class YarnScheduler(sc: SparkContext) extends TaskSchedulerImpl(sc) { + + // RackResolver logs an INFO message whenever it resolves a rack, which is way too often. + if (Logger.getLogger(classOf[RackResolver]).getLevel == null) { + Logger.getLogger(classOf[RackResolver]).setLevel(Level.WARN) + } // By default, rack is unknown override def getRackForHost(hostPort: String): Option[String] = { |