[SPARK-11206] Support SQL UI on the history server (resubmit)

Resubmit #9297 and #9991
On the live web UI, there is a SQL tab which provides valuable information for the SQL query. But once the workload is finished, we won't see the SQL tab on the history server. It will be helpful if we support SQL UI on the history server so we can analyze it even after its execution.

To support SQL UI on the history server:
1. I added an onOtherEvent method to the SparkListener trait and post all SQL related events to the same event bus.
2. Two SQL events SparkListenerSQLExecutionStart and SparkListenerSQLExecutionEnd are defined in the sql module.
3. The new SQL events are written to event log using Jackson.
4. A new trait SparkHistoryListenerFactory is added to allow the history server to feed events to the SQL history listener. The SQL implementation is loaded at runtime using java.util.ServiceLoader.

Author: Carson Wang <[email protected]>

Closes #10061 from carsonwang/SqlHistoryUI.

Author: carsonwang

.rat-excludes

@@ -82,4 +82,5 @@ INDEX
 gen-java.*
 .*avpr
 org.apache.spark.sql.sources.DataSourceRegister
+org.apache.spark.scheduler.SparkHistoryListenerFactory
 .*parquet

core/src/main/java/org/apache/spark/JavaSparkListener.java

@@ -82,4 +82,7 @@ public void onExecutorRemoved(SparkListenerExecutorRemoved executorRemoved) { }
   @Override
   public void onBlockUpdated(SparkListenerBlockUpdated blockUpdated) { }
 
+  @Override
+  public void onOtherEvent(SparkListenerEvent event) { }
+
 }

core/src/main/java/org/apache/spark/SparkFirehoseListener.java

@@ -118,4 +118,8 @@ public void onBlockUpdated(SparkListenerBlockUpdated blockUpdated) {
         onEvent(blockUpdated);
     }
 
+    @Override
+    public void onOtherEvent(SparkListenerEvent event) {
+        onEvent(event);
+    }
 }

core/src/main/scala/org/apache/spark/scheduler/EventLoggingListener.scala

@@ -207,6 +207,10 @@ private[spark] class EventLoggingListener(
   // No-op because logging every update would be overkill
   override def onExecutorMetricsUpdate(event: SparkListenerExecutorMetricsUpdate): Unit = { }
 
+  override def onOtherEvent(event: SparkListenerEvent): Unit = {
+    logEvent(event, flushLogger = true)
+  }
+
   /**
    * Stop logging events. The event log file will be renamed so that it loses the
    * ".inprogress" suffix.

core/src/main/scala/org/apache/spark/scheduler/SparkListener.scala

@@ -22,15 +22,19 @@ import java.util.Properties
 import scala.collection.Map
 import scala.collection.mutable
 
-import org.apache.spark.{Logging, TaskEndReason}
+import com.fasterxml.jackson.annotation.JsonTypeInfo
+
+import org.apache.spark.{Logging, SparkConf, TaskEndReason}
 import org.apache.spark.annotation.DeveloperApi
 import org.apache.spark.executor.TaskMetrics
 import org.apache.spark.scheduler.cluster.ExecutorInfo
 import org.apache.spark.storage.{BlockManagerId, BlockUpdatedInfo}
 import org.apache.spark.util.{Distribution, Utils}
+import org.apache.spark.ui.SparkUI
 
 @DeveloperApi
-sealed trait SparkListenerEvent
+@JsonTypeInfo(use = JsonTypeInfo.Id.CLASS, include = JsonTypeInfo.As.PROPERTY, property = "Event")
+trait SparkListenerEvent
 
 @DeveloperApi
 case class SparkListenerStageSubmitted(stageInfo: StageInfo, properties: Properties = null)
@@ -130,6 +134,17 @@ case class SparkListenerApplicationEnd(time: Long) extends SparkListenerEvent
  */
 private[spark] case class SparkListenerLogStart(sparkVersion: String) extends SparkListenerEvent
 
+/**
+ * Interface for creating history listeners defined in other modules like SQL, which are used to
+ * rebuild the history UI.
+ */
+private[spark] trait SparkHistoryListenerFactory {
+  /**
+   * Create listeners used to rebuild the history UI.
+   */
+  def createListeners(conf: SparkConf, sparkUI: SparkUI): Seq[SparkListener]
+}
+
 /**
  * :: DeveloperApi ::
  * Interface for listening to events from the Spark scheduler. Note that this is an internal
@@ -223,6 +238,11 @@ trait SparkListener {
    * Called when the driver receives a block update info.
    */
   def onBlockUpdated(blockUpdated: SparkListenerBlockUpdated) { }
+
+  /**
+   * Called when other events like SQL-specific events are posted.
+   */
+  def onOtherEvent(event: SparkListenerEvent) { }
 }
 
 /**

core/src/main/scala/org/apache/spark/scheduler/SparkListenerBus.scala

@@ -61,6 +61,7 @@ private[spark] trait SparkListenerBus extends ListenerBus[SparkListener, SparkLi
       case blockUpdated: SparkListenerBlockUpdated =>
         listener.onBlockUpdated(blockUpdated)
       case logStart: SparkListenerLogStart => // ignore event log metadata
+      case _ => listener.onOtherEvent(event)
     }
   }
 

core/src/main/scala/org/apache/spark/ui/SparkUI.scala

@@ -17,10 +17,13 @@
 
 package org.apache.spark.ui
 
-import java.util.Date
+import java.util.{Date, ServiceLoader}
+
+import scala.collection.JavaConverters._
 
 import org.apache.spark.status.api.v1.{ApiRootResource, ApplicationAttemptInfo, ApplicationInfo,
   UIRoot}
+import org.apache.spark.util.Utils
 import org.apache.spark.{Logging, SecurityManager, SparkConf, SparkContext}
 import org.apache.spark.scheduler._
 import org.apache.spark.storage.StorageStatusListener
@@ -154,7 +157,16 @@ private[spark] object SparkUI {
       appName: String,
       basePath: String,
       startTime: Long): SparkUI = {
-    create(None, conf, listenerBus, securityManager, appName, basePath, startTime = startTime)
+    val sparkUI = create(
+      None, conf, listenerBus, securityManager, appName, basePath, startTime = startTime)
+
+    val listenerFactories = ServiceLoader.load(classOf[SparkHistoryListenerFactory],
+      Utils.getContextOrSparkClassLoader).asScala
+    listenerFactories.foreach { listenerFactory =>
+      val listeners = listenerFactory.createListeners(conf, sparkUI)
+      listeners.foreach(listenerBus.addListener)
+    }
+    sparkUI
   }
 
   /**

core/src/main/scala/org/apache/spark/util/JsonProtocol.scala

@@ -19,19 +19,21 @@ package org.apache.spark.util
 
 import java.util.{Properties, UUID}
 
-import org.apache.spark.scheduler.cluster.ExecutorInfo
-
 import scala.collection.JavaConverters._
 import scala.collection.Map
 
+import com.fasterxml.jackson.databind.ObjectMapper
+import com.fasterxml.jackson.module.scala.DefaultScalaModule
 import org.json4s.DefaultFormats
 import org.json4s.JsonDSL._
 import org.json4s.JsonAST._
+import org.json4s.jackson.JsonMethods._
 
 import org.apache.spark._
 import org.apache.spark.executor._
 import org.apache.spark.rdd.RDDOperationScope
 import org.apache.spark.scheduler._
+import org.apache.spark.scheduler.cluster.ExecutorInfo
 import org.apache.spark.storage._
 
 /**
@@ -54,6 +56,8 @@ private[spark] object JsonProtocol {
 
   private implicit val format = DefaultFormats
 
+  private val mapper = new ObjectMapper().registerModule(DefaultScalaModule)
+
   /** ------------------------------------------------- *
    * JSON serialization methods for SparkListenerEvents |
    * -------------------------------------------------- */
@@ -96,6 +100,7 @@ private[spark] object JsonProtocol {
         executorMetricsUpdateToJson(metricsUpdate)
       case blockUpdated: SparkListenerBlockUpdated =>
         throw new MatchError(blockUpdated)  // TODO(ekl) implement this
+      case _ => parse(mapper.writeValueAsString(event))
     }
   }
 
@@ -506,6 +511,8 @@ private[spark] object JsonProtocol {
       case `executorRemoved` => executorRemovedFromJson(json)
       case `logStart` => logStartFromJson(json)
       case `metricsUpdate` => executorMetricsUpdateFromJson(json)
+      case other => mapper.readValue(compact(render(json)), Utils.classForName(other))
+        .asInstanceOf[SparkListenerEvent]
     }
   }
 

sql/core/src/main/resources/META-INF/services/org.apache.spark.scheduler.SparkHistoryListenerFactory

@@ -0,0 +1 @@
+org.apache.spark.sql.execution.ui.SQLHistoryListenerFactory

sql/core/src/main/scala/org/apache/spark/sql/SQLContext.scala

@@ -1245,6 +1245,7 @@ class SQLContext private[sql](
   sparkContext.addSparkListener(new SparkListener {
     override def onApplicationEnd(applicationEnd: SparkListenerApplicationEnd): Unit = {
       SQLContext.clearInstantiatedContext()
+      SQLContext.clearSqlListener()
     }
   })
 
@@ -1272,6 +1273,8 @@ object SQLContext {
    */
   @transient private val instantiatedContext = new AtomicReference[SQLContext]()
 
+  @transient private val sqlListener = new AtomicReference[SQLListener]()
+
   /**
    * Get the singleton SQLContext if it exists or create a new one using the given SparkContext.
    *
@@ -1316,6 +1319,10 @@ object SQLContext {
     Option(instantiatedContext.get())
   }
 
+  private[sql] def clearSqlListener(): Unit = {
+    sqlListener.set(null)
+  }
+
   /**
    * Changes the SQLContext that will be returned in this thread and its children when
    * SQLContext.getOrCreate() is called. This can be used to ensure that a given thread receives
@@ -1364,9 +1371,13 @@ object SQLContext {
    * Create a SQLListener then add it into SparkContext, and create an SQLTab if there is SparkUI.
    */
   private[sql] def createListenerAndUI(sc: SparkContext): SQLListener = {
-    val listener = new SQLListener(sc.conf)
-    sc.addSparkListener(listener)
-    sc.ui.foreach(new SQLTab(listener, _))
-    listener
+    if (sqlListener.get() == null) {
+      val listener = new SQLListener(sc.conf)
+      if (sqlListener.compareAndSet(null, listener)) {
+        sc.addSparkListener(listener)
+        sc.ui.foreach(new SQLTab(listener, _))
+      }
+    }
+    sqlListener.get()
   }
 }