Github user mateiz commented on a diff in the pull request:
https://github.com/apache/spark/pull/86#discussion_r10332714
--- Diff: core/src/main/scala/org/apache/spark/deploy/SparkApp.scala ---
@@ -0,0 +1,178 @@
+/*
+ * 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.deploy
+
+import java.io.BufferedReader
+import java.io.InputStream
+import java.io.InputStreamReader
+import java.io.PrintStream
+
+import scala.collection.mutable.HashMap
+import scala.collection.mutable.Map
+import scala.collection.mutable.ArrayBuffer
+import scala.collection.JavaConverters._
+
+object SparkApp {
+ val CLIENT = 1
+ val CLUSTER = 2
+ val YARN = 1
+ val STANDALONE = 2
+ val MESOS = 4
+ val LOCAL = 8
+ val ALL_CLUSTER_MGRS = YARN | STANDALONE | MESOS | LOCAL
+
+ var clusterManager: Int = LOCAL
+
+ def main(args: Array[String]) {
+ println("args: " + args.toList)
+ val appArgs = new SparkAppArguments(args)
+
+ if (appArgs.master != null) {
+ if (appArgs.master.startsWith("yarn")) {
+ clusterManager = YARN
+ } else if (appArgs.master.startsWith("spark")) {
+ clusterManager = STANDALONE
+ } else if (appArgs.master.startsWith("mesos")) {
+ clusterManager = MESOS
+ } else if (appArgs.master == "local") {
+ clusterManager = LOCAL
+ } else {
+ System.err.println("master must start with yarn, mesos, spark, or
be local")
+ System.exit(1)
+ }
+ }
+
+ val deployMode = if (appArgs.deployMode == "client") CLIENT else
CLUSTER
+ val childEnv = new HashMap[String, String]()
+ val childClasspath = new ArrayBuffer[String]()
+ val childArgs = new ArrayBuffer[String]()
+ childArgs += System.getenv("SPARK_HOME") + "/bin/spark-class"
+
+ if (clusterManager == MESOS && deployMode == CLUSTER) {
+ System.err.println("Mesos does not support running the driver on the
cluster")
+ System.exit(1)
+ }
+
+ if (deployMode == CLUSTER && clusterManager == STANDALONE) {
+ childArgs += "org.apache.spark.deploy.Client"
+ childArgs += "launch"
+ childArgs += appArgs.master
+ childArgs += appArgs.primaryResource
+ childArgs += appArgs.mainClass
+ } else if (deployMode == CLUSTER && clusterManager == YARN) {
+ childArgs += "org.apache.spark.deploy.yarn.Client"
+ childArgs += "--jar"
+ childArgs += appArgs.primaryResource
+ childArgs += "--class"
+ childArgs += appArgs.mainClass
+ } else {
+ childClasspath += appArgs.primaryResource
+ childArgs += appArgs.mainClass
+ }
+
+ // TODO: num-executors when not using YARN
+ val options = List[Opt](
+ new Opt(appArgs.driverMemory, YARN, CLUSTER, null,
"--master-memory", null),
+ new Opt(appArgs.name, YARN, CLUSTER, null, "--name", null),
+ new Opt(appArgs.queue, YARN, CLUSTER, null, "--queue", null),
+ new Opt(appArgs.queue, YARN, CLIENT, "SPARK_YARN_QUEUE", null, null),
+ new Opt(appArgs.numExecutors, YARN, CLUSTER, null, "--num-workers",
null),
+ new Opt(appArgs.executorMemory, YARN, CLIENT, "SPARK_WORKER_MEMORY",
null, null),
+ new Opt(appArgs.executorMemory, YARN, CLUSTER, null,
"--worker-memory", null),
+ new Opt(appArgs.executorMemory, STANDALONE, CLUSTER, null,
"--memory", null),
+ new Opt(appArgs.executorMemory, STANDALONE | MESOS, CLIENT, null,
null, "spark.executor.memory"),
+ new Opt(appArgs.executorCores, YARN, CLIENT, "SPARK_WORKER_CORES",
null, null),
+ new Opt(appArgs.executorCores, YARN, CLUSTER, null,
"--worker-cores", null),
+ new Opt(appArgs.executorCores, STANDALONE, CLUSTER, null, "--cores",
null),
+ new Opt(appArgs.executorCores, STANDALONE | MESOS, CLIENT, null,
null, "spark.cores.max"),
+ new Opt(appArgs.files, YARN, CLIENT, "SPARK_YARN_DIST_FILES", null,
null),
+ new Opt(appArgs.files, YARN, CLUSTER, null, "--files", null),
+ new Opt(appArgs.archives, YARN, CLIENT, "SPARK_YARN_DIST_ARCHIVES",
null, null),
+ new Opt(appArgs.archives, YARN, CLUSTER, null, "--archives", null),
+ new Opt(appArgs.moreJars, YARN, CLUSTER, null, "--addJars", null)
+ )
+
+ // args
+ if (deployMode == CLIENT || clusterManager == STANDALONE) {
+ childArgs ++= appArgs.childArgs
+ } else if (clusterManager == YARN) {
+ for (arg <- appArgs.childArgs) {
+ childArgs += "--args"
+ childArgs += arg
+ }
+ }
+
+ // client memory
+ if (appArgs.driverMemory != null && deployMode == CLIENT) {
+ childArgs += "-Xmx" + appArgs.driverMemory
+ }
+
+ for (opt <- options) {
+ if (opt.value != null && deployMode == opt.deployMode &&
+ (clusterManager & opt.clusterManager) != 0) {
+ if (opt.clOption != null) {
+ childArgs += opt.clOption
+ childArgs += opt.value
+ } else if (opt.envVar != null) {
+ childEnv.put(opt.envVar, opt.value)
+ } else if (opt.jvmSystemProperty != null) {
+ childArgs += "-D" + opt.jvmSystemProperty + "=" + opt.value
+ }
+ }
+ }
+
+ childEnv.put("SPARK_CLASSPATH", childClasspath.mkString(","))
+ launch(childEnv, childArgs)
+ }
+
+ def launch(childEnv: Map[String, String], childArgs:
ArrayBuffer[String]) {
+ val pb = new ProcessBuilder()
+ pb.environment().putAll(childEnv.asJava)
+ pb.command(childArgs.asJava)
+ val proc = pb.start()
+ redirectInputStream(proc.getInputStream, System.out)
+ redirectInputStream(proc.getErrorStream, System.err)
+ proc.waitFor()
+ }
+
+ def redirectInputStream(is: InputStream, ps: PrintStream) {
+ new Thread(new Runnable {
+ def run() {
+ val br = new BufferedReader(new InputStreamReader(is))
+ var line = br.readLine()
+ while (line != null) {
+ ps.println(line)
+ line = br.readLine()
+ }
+ }
+ }).start()
+ }
+}
+
+class LaunchContext(val env: Map[String, String],
+ val args: ArrayBuffer[String],
+ val classpath: ArrayBuffer[String]
+) {}
+
+class Opt(val value: String,
--- End diff --
I'm not sure Opt is the best name for this, it's almost something like an
OptionAssigner. Also as a tip you can use default parameters here to avoid
passing lots of nulls above (which is hard to read). Something like this:
```
class Opt(val value: String, val deployModes: Int, val clusterManagers:
Int, val envVar: String = null, ...)
```
Then you can call:
```
new OptionAssigner(foo, YARN, CLIENT, envVar="SPARK_FOO")
```
You can also make this a case class to avoid the `new`.
---
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.
---
