Merge pull request #23 from markhamstra/master-csd

SPY-350 and catching up to Apache branch-0.9

Author: markhamstra

CHANGES.txt

@@ -1,6 +1,214 @@
 Spark Change Log
 ----------------
 
+Release 0.9.2
+
+  [branch-0.9] bump versions for v0.9.2 release candidate
+  Xiangrui Meng <[email protected]>
+  2014-07-16 23:53:40 -0700
+  Commit: c9a22e8, github.com/apache/spark/pull/1458
+
+  [branch-0.9] Fix github links in docs
+  Xiangrui Meng <[email protected]>
+  2014-07-16 23:39:02 -0700
+  Commit: 60f4b3b, github.com/apache/spark/pull/1456
+
+  [SPARK-1112, 2156] (0.9 edition) Use correct akka frame size and overhead amounts.
+  Patrick Wendell <[email protected]>
+  2014-07-16 21:30:50 -0700
+  Commit: 7edee34, github.com/apache/spark/pull/1455
+
+  [SPARK-2433][MLLIB] fix NaiveBayesModel.predict
+  Xiangrui Meng <[email protected]>
+  2014-07-16 20:12:09 -0700
+  Commit: 0116dee, github.com/apache/spark/pull/1453
+
+  [SPARK-2362] Fix for newFilesOnly logic in file DStream
+  Gabriele Nizzoli <[email protected]>
+  2014-07-08 14:23:38 -0700
+  Commit: 8e5604b, github.com/apache/spark/pull/1077
+
+  SPARK-2282: Reuse PySpark Accumulator sockets to avoid crashing Spark
+  Aaron Davidson <[email protected]>
+  2014-07-03 23:02:36 -0700
+  Commit: 57873ef, github.com/apache/spark/pull/1220
+
+  [SPARK-2350] Don't NPE while launching drivers
+  Aaron Davidson <[email protected]>
+  2014-07-03 22:31:41 -0700
+  Commit: c37e9ed, github.com/apache/spark/pull/1289
+
+  [SPARK-1516]Throw exception in yarn client instead of run system.exit
+  John Zhao <[email protected]>
+  2014-07-03 15:17:51 -0700
+  Commit: 0d3d5ce, github.com/apache/spark/pull/1099
+
+  HOTFIX: Removing out dated python path in testing tool.
+  Patrick Wendell <[email protected]>
+  2014-06-27 18:19:16 -0700
+  Commit: b3f4245
+
+  [SPARK-1912] fix compress memory issue during reduce
+  Wenchen Fan(Cloud) <[email protected]>
+  2014-06-03 13:18:20 -0700
+  Commit: 9509819, github.com/apache/spark/pull/860
+
+  SPARK-2241: quote command line args in ec2 script
+  Ori Kremer <[email protected]>
+  2014-06-22 20:21:23 -0700
+  Commit: ef8501d, github.com/apache/spark/pull/1169
+
+  HOTFIX: bug caused by #941
+  Patrick Wendell <[email protected]>
+  2014-06-17 15:09:24 -0700
+  Commit: 2a2eace, github.com/apache/spark/pull/1108
+
+  SPARK-1990: added compatibility for python 2.6 for ssh_read command
+  Anant <[email protected]>
+  2014-06-16 23:42:27 -0700
+  Commit: 8e9f479, github.com/apache/spark/pull/941
+
+  [SPARK-1998] SparkFlumeEvent with body bigger than 1020 bytes are not re...
+  joyyoj <[email protected]>
+  2014-06-10 17:26:17 -0700
+  Commit: 706e38f, github.com/apache/spark/pull/951
+
+  Spark 1384 - Fix spark-shell on yarn access to secure hdfs - branch-0.9 only
+  Thomas Graves <[email protected]>
+  2014-06-09 23:07:25 -0700
+  Commit: cc95d97, github.com/apache/spark/pull/287
+
+  [SPARK-1870] Made deployment with --jars work in yarn-standalone mode.
+  DB Tsai <[email protected]>
+  2014-06-09 22:56:24 -0700
+  Commit: 1d3aab9, github.com/apache/spark/pull/1013
+
+  SPARK-2043: ExternalAppendOnlyMap doesn't always find matching keys
+  Matei Zaharia <[email protected]>
+  2014-06-05 23:01:48 -0700
+  Commit: 51f677e, github.com/apache/spark/pull/986
+
+  SPARK-1790: Update EC2 scripts to support r3 instance types
+  Varakhedi Sujeet <[email protected]>
+  2014-06-04 16:01:56 -0700
+  Commit: 6634a34, github.com/apache/spark/pull/960
+
+  [SPARK-1468] Modify the partition function used by partitionBy.
+  Erik Selin <[email protected]>
+  2014-06-03 13:31:16 -0700
+  Commit: 41e7853, github.com/apache/spark/pull/371
+
+  SPARK-1917: fix PySpark import of scipy.special functions
+  Uri Laserson <[email protected]>
+  2014-05-31 14:59:09 -0700
+  Commit: e03af41, github.com/apache/spark/pull/866
+
+  SPARK-1935: Explicitly add commons-codec 1.5 as a dependency (for branch-0.9).
+  Yin Huai <[email protected]>
+  2014-05-30 22:12:17 -0700
+  Commit: 563bfe1, github.com/apache/spark/pull/912
+
+  SPARK-1188: Do not re-use objects in the EdgePartition/EdgeTriplet iterators.
+  Daniel Darabos <[email protected]>
+  2014-04-02 12:27:37 -0700
+  Commit: a92900c, github.com/apache/spark/pull/276
+
+  [SPARK-1712]: TaskDescription instance is too big causes Spark to hang
+  witgo <[email protected]>
+  2014-05-28 15:57:05 -0700
+  Commit: aef6390, github.com/apache/spark/pull/694
+
+  Spark 1916
+  David Lemieux <[email protected]>
+  2014-05-28 15:50:35 -0700
+  Commit: 234a378, github.com/apache/spark/pull/865
+
+  SPARK-1145: Memory mapping with many small blocks can cause JVM allocation failures
+  Patrick Wendell <[email protected]>
+  2014-04-27 17:40:56 -0700
+  Commit: 7633949, github.com/apache/spark/pull/43
+
+  Update version to 0.9.2-SNAPSHOT in sbt
+  Matei Zaharia <[email protected]>
+  2014-05-11 16:54:54 -0700
+  Commit: c9f40d0
+
+  SPARK-1775: Unneeded lock in ShuffleMapTask.deserializeInfo
+  Sandeep <[email protected]>
+  2014-05-08 22:30:17 -0700
+  Commit: bea2be3, github.com/apache/spark/pull/707
+
+  [SPARK-1685] Cancel retryTimer on restart of Worker or AppClient
+  Mark Hamstra <[email protected]>
+  2014-05-06 12:53:39 -0700
+  Commit: 9e2c59e, github.com/apache/spark/pull/602
+
+  [WIP] SPARK-1676: Cache Hadoop UGIs by default to prevent FileSystem leak
+  Thomas Graves <[email protected]>
+  2014-05-03 10:59:05 -0700
+  Commit: 45561cd, github.com/apache/spark/pull/621
+
+  version number fix
+  Nan Zhu <[email protected]>
+  2014-04-21 23:42:47 -0700
+  Commit: 54c3b7e, github.com/apache/spark/pull/467
+
+  Small syntax error from previous backport
+  Patrick Wendell <[email protected]>
+  2014-04-13 14:32:22 -0700
+  Commit: 9e89789
+
+  Update WindowedDStream.scala
+  baishuo(白硕) <[email protected]>
+  2014-04-11 20:33:42 -0700
+  Commit: 4a325e1, github.com/apache/spark/pull/390
+
+  Fixed typo on Spark quick-start docs.
+  Tathagata Das <[email protected]>
+  2014-04-07 18:27:46 -0700
+  Commit: 19cf2f7
+
+  SPARK-1432: Make sure that all metadata fields are properly cleaned
+  Davis Shepherd <[email protected]>
+  2014-04-07 10:02:00 -0700
+  Commit: 69fc97d, github.com/apache/spark/pull/338
+
+  SPARK-1421. Make MLlib work on Python 2.6
+  Matei Zaharia <[email protected]>
+  2014-04-05 20:52:05 -0700
+  Commit: 139fc1a, github.com/apache/spark/pull/335
+
+  Update documentation for work around for SPARK-1384
+  Thomas Graves <[email protected]>
+  2014-04-04 18:26:51 -0700
+  Commit: d4df076, github.com/apache/spark/pull/314
+
+  SPARK-1337: Application web UI garbage collects newest stages
+  Patrick Wendell <[email protected]>
+  2014-04-03 22:13:56 -0700
+  Commit: 7f727cf, github.com/apache/spark/pull/320
+
+  [SPARK-1134] Fix and document passing of arguments to IPython
+  Diana Carroll <[email protected]>
+  2014-04-03 15:48:42 -0700
+  Commit: d9c7a80, github.com/apache/spark/pull/294
+
+  Spark 1162 Implemented takeOrdered in pyspark.
+  Prashant Sharma <[email protected]>
+  2014-04-03 15:42:17 -0700
+  Commit: 28e7643, github.com/apache/spark/pull/97
+
+  fix path for jar, make sed actually work on OSX
+  Nick Lanham <[email protected]>
+  2014-03-28 13:33:35 -0700
+  Commit: a6c955a, github.com/apache/spark/pull/264
+
+  Make sed do -i '' on OSX
+  Nick Lanham <[email protected]>
+  2014-03-27 22:45:00 -0700
+  Commit: 4afbd19, github.com/apache/spark/pull/258
+
+
 Release 0.9.1
 
   Revert "[maven-release-plugin] prepare release v0.9.1-rc2"

assembly/pom.xml

@@ -21,7 +21,7 @@
   <parent>
     <groupId>org.apache.spark</groupId>
     <artifactId>spark-parent</artifactId>
-    <version>0.9.2-candidate-csd-8-SNAPSHOT</version>
+    <version>0.9.2-csd-1-SNAPSHOT</version>
     <relativePath>../pom.xml</relativePath>
   </parent>
 

bagel/pom.xml

@@ -21,7 +21,7 @@
   <parent>
     <groupId>org.apache.spark</groupId>
     <artifactId>spark-parent</artifactId>
-    <version>0.9.2-candidate-csd-8-SNAPSHOT</version>
+    <version>0.9.2-csd-1-SNAPSHOT</version>
     <relativePath>../pom.xml</relativePath>
   </parent>
 

core/pom.xml

@@ -21,7 +21,7 @@
     <parent>
         <groupId>org.apache.spark</groupId>
         <artifactId>spark-parent</artifactId>
-        <version>0.9.2-candidate-csd-8-SNAPSHOT</version>
+        <version>0.9.2-csd-1-SNAPSHOT</version>
         <relativePath>../pom.xml</relativePath>
     </parent>
 

core/src/main/scala/org/apache/spark/api/python/PythonRDD.scala

@@ -307,6 +307,8 @@ private class PythonAccumulatorParam(@transient serverHost: String, serverPort:
     } else {
       // This happens on the master, where we pass the updates to Python through a socket
       val socket = new Socket(serverHost, serverPort)
+      // SPARK-2282: Immediately reuse closed sockets because we create one per task.
+      socket.setReuseAddress(true)
       val in = socket.getInputStream
       val out = new DataOutputStream(new BufferedOutputStream(socket.getOutputStream, bufferSize))
       out.writeInt(val2.size)

core/src/main/scala/org/apache/spark/deploy/master/ApplicationInfo.scala

@@ -90,11 +90,13 @@ private[spark] class ApplicationInfo(
 
   def retryCount = _retryCount
 
-  def incrementRetryCount = {
+  def incrementRetryCount() = {
     _retryCount += 1
     _retryCount
   }
 
+  def resetRetryCount() = _retryCount = 0
+
   def markFinished(endState: ApplicationState.Value) {
     state = endState
     endTime = System.currentTimeMillis()

core/src/main/scala/org/apache/spark/deploy/master/Master.scala

@@ -264,27 +264,34 @@ private[spark] class Master(host: String, port: Int, webUiPort: Int) extends Act
       val execOption = idToApp.get(appId).flatMap(app => app.executors.get(execId))
       execOption match {
         case Some(exec) => {
+          val appInfo = idToApp(appId)
           exec.state = state
+          if (state == ExecutorState.RUNNING) { appInfo.resetRetryCount() }
           exec.application.driver ! ExecutorUpdated(execId, state, message, exitStatus)
           if (ExecutorState.isFinished(state)) {
-            val appInfo = idToApp(appId)
             // Remove this executor from the worker and app
-            logInfo("Removing executor " + exec.fullId + " because it is " + state)
+            logInfo(s"Removing executor ${exec.fullId} because it is $state")
             appInfo.removeExecutor(exec)
             exec.worker.removeExecutor(exec)
 
+            val normalExit = exitStatus == Some(0)
             // Only retry certain number of times so we don't go into an infinite loop.
-            if (appInfo.incrementRetryCount < ApplicationState.MAX_NUM_RETRY) {
-              schedule()
-            } else {
-              logError("Application %s with ID %s failed %d times, removing it".format(
-                appInfo.desc.name, appInfo.id, appInfo.retryCount))
-              removeApplication(appInfo, ApplicationState.FAILED)
+            if (!normalExit) {
+              if (appInfo.incrementRetryCount() < ApplicationState.MAX_NUM_RETRY) {
+                schedule()
+              } else {
+                val execs = appInfo.executors.values
+                if (!execs.exists(_.state == ExecutorState.RUNNING)) {
+                  logError(s"Application ${appInfo.desc.name} with ID ${appInfo.id} failed " +
+                    s"${appInfo.retryCount} times; removing it")
+                  removeApplication(appInfo, ApplicationState.FAILED)
+                }
+              }
             }
           }
         }
         case None =>
-          logWarning("Got status update for unknown executor " + appId + "/" + execId)
+          logWarning(s"Got status update for unknown executor $appId/$execId")
       }
     }
 
@@ -450,7 +457,7 @@ private[spark] class Master(host: String, port: Int, webUiPort: Int) extends Act
     // First schedule drivers, they take strict precedence over applications
     val shuffledWorkers = Random.shuffle(workers) // Randomization helps balance drivers
     for (worker <- shuffledWorkers if worker.state == WorkerState.ALIVE) {
-      for (driver <- waitingDrivers) {
+      for (driver <- List(waitingDrivers: _*)) { // iterate over a copy of waitingDrivers
         if (worker.memoryFree >= driver.desc.mem && worker.coresFree >= driver.desc.cores) {
           launchDriver(worker, driver)
           waitingDrivers -= driver

core/src/main/scala/org/apache/spark/deploy/worker/ExecutorRunner.scala

@@ -139,9 +139,10 @@ private[spark] class ExecutorRunner(
       Files.write(header, stderr, Charsets.UTF_8)
       CommandUtils.redirectStream(process.getErrorStream, stderr)
 
-      // Wait for it to exit; this is actually a bad thing if it happens, because we expect to run
-      // long-lived processes only. However, in the future, we might restart the executor a few
-      // times on the same machine.
+      state = ExecutorState.RUNNING
+      worker ! ExecutorStateChanged(appId, execId, state, None, None)
+      // Wait for it to exit; executor may exit with code 0 (when driver instructs it to shutdown)
+      // or with nonzero exit code
       val exitCode = process.waitFor()
       state = ExecutorState.FAILED
       val message = "Command exited with code " + exitCode

core/src/main/scala/org/apache/spark/deploy/worker/Worker.scala

@@ -211,18 +211,29 @@ private[spark] class Worker(
       if (masterUrl != activeMasterUrl) {
         logWarning("Invalid Master (" + masterUrl + ") attempted to launch executor.")
       } else {
-        logInfo("Asked to launch executor %s/%d for %s".format(appId, execId, appDesc.name))
-        // TODO (pwendell): We shuld make sparkHome an Option[String] in
-        // ApplicationDescription to be more explicit about this.
-        val effectiveSparkHome = Option(execSparkHome_).getOrElse(sparkHome.getAbsolutePath)
-        val manager = new ExecutorRunner(appId, execId, appDesc, cores_, memory_,
-          self, workerId, host, new File(effectiveSparkHome), workDir, akkaUrl, ExecutorState.RUNNING)
-        executors(appId + "/" + execId) = manager
-        manager.start()
-        coresUsed += cores_
-        memoryUsed += memory_
-        masterLock.synchronized {
-          master ! ExecutorStateChanged(appId, execId, manager.state, None, None)
+        try {
+          logInfo("Asked to launch executor %s/%d for %s".format(appId, execId, appDesc.name))
+          val effectiveSparkHome = Option(execSparkHome_).getOrElse(sparkHome.getAbsolutePath)
+          val manager = new ExecutorRunner(appId, execId, appDesc, cores_, memory_,
+            self, workerId, host, new File(effectiveSparkHome), workDir, akkaUrl, ExecutorState.LOADING)
+          executors(appId + "/" + execId) = manager
+          manager.start()
+          coresUsed += cores_
+          memoryUsed += memory_
+          masterLock.synchronized {
+            master ! ExecutorStateChanged(appId, execId, manager.state, None, None)
+          }
+        } catch {
+          case e: Exception => {
+            logError("Failed to launch executor %s/%d for %s".format(appId, execId, appDesc.name))
+            if (executors.contains(appId + "/" + execId)) {
+              executors(appId + "/" + execId).kill()
+              executors -= appId + "/" + execId
+            }
+            masterLock.synchronized {
+              master ! ExecutorStateChanged(appId, execId, ExecutorState.FAILED, None, None)
+            }
+          }
         }
       }