Github user andrewor14 commented on a diff in the pull request:
https://github.com/apache/spark/pull/3916#discussion_r22840665
--- Diff:
launcher/src/main/java/org/apache/spark/launcher/SparkLauncher.java ---
@@ -0,0 +1,371 @@
+/*
+ * 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.launcher;
+
+import java.io.File;
+import java.io.IOException;
+import java.lang.reflect.Method;
+import java.net.URL;
+import java.net.URLClassLoader;
+import java.util.ArrayList;
+import java.util.Collections;
+import java.util.List;
+import java.util.Map;
+import java.util.Properties;
+import java.util.concurrent.atomic.AtomicLong;
+
+/**
+ * Launcher for Spark applications.
+ * <p/>
+ * Use this class to start Spark applications programatically. The class
uses a builder pattern
+ * to allow clients to configure the Spark application and launch it as a
child process.
+ * <p/>
+ * There's also support for running the application on a separate thread,
although that is to
+ * be considered experimental and avoided in production environments.
+ * <p/>
+ * Note that launching Spark applications using this class will not
automatically load environment
+ * variables from the "spark-env.sh" or "spark-env.cmd" scripts in the
configuration directory.
+ */
+public class SparkLauncher extends AbstractLauncher<SparkLauncher> {
+
+ private static final AtomicLong THREAD_ID = new AtomicLong();
+
+ protected boolean verbose;
+ protected String appName;
+ protected String master;
+ protected String deployMode;
+ protected String userClass;
+ protected String userResource;
+ protected final List<String> sparkArgs;
+ protected final List<String> userArgs;
+ protected final List<String> jars;
+ protected final List<String> files;
+ protected final List<String> pyFiles;
+
+ public SparkLauncher() {
+ this.sparkArgs = new ArrayList<String>();
+ this.userArgs = new ArrayList<String>();
+ this.jars = new ArrayList<String>();
+ this.files = new ArrayList<String>();
+ this.pyFiles = new ArrayList<String>();
+ }
+
+ /** Set the application name. */
+ public SparkLauncher setAppName(String appName) {
+ checkNotNull(appName, "appName");
+ this.appName = appName;
+ return this;
+ }
+
+ /** Set the Spark master for the application. */
+ public SparkLauncher setMaster(String master) {
+ checkNotNull(master, "master");
+ this.master = master;
+ return this;
+ }
+
+ /** Set the deploy mode for the application. */
+ public SparkLauncher setDeployMode(String mode) {
+ checkNotNull(mode, "mode");
+ this.deployMode = mode;
+ return this;
+ }
+
+ /**
+ * Set the main application resource. This should be the location of a
jar file for Scala/Java
+ * applications, or a python script for PySpark applications.
+ */
+ public SparkLauncher setAppResource(String path) {
+ checkNotNull(path, "path");
+ this.userResource = path;
+ return this;
+ }
+
+ /** Sets the application class name for Java/Scala applications. */
+ public SparkLauncher setClass(String userClass) {
+ checkNotNull(userClass, "userClass");
+ this.userClass = userClass;
+ return this;
+ }
+
+ /** Adds command line arguments for the application. */
+ public SparkLauncher addArgs(String... args) {
+ for (String arg : args) {
+ checkNotNull(arg, "arg");
+ userArgs.add(arg);
+ }
+ return this;
+ }
+
+ /** Adds a jar file to be submitted with the application. */
+ public SparkLauncher addJar(String path) {
+ checkNotNull(path, "path");
+ jars.add(path);
+ return this;
+ }
+
+ /** Adds a file to be submitted with the application. */
+ public SparkLauncher addFile(String path) {
+ checkNotNull(path, "path");
+ files.add(path);
+ return this;
+ }
+
+ /** Adds a a python file / zip / egg to be submitted with the
application. */
+ public SparkLauncher addPyFile(String path) {
+ checkNotNull(path, "path");
+ pyFiles.add(path);
+ return this;
+ }
+
+ /** Enables verbose reporting for SparkSubmit. */
+ public SparkLauncher setVerbose(boolean verbose) {
+ this.verbose = verbose;
+ return this;
+ }
+
+ /**
+ * Starts a new thread that will run the Spark application.
+ * <p/>
+ * The application will run on a separate thread and use a separate,
isolated class loader.
+ * No classes or resources from the current thread's class loader will
be visible to the app.
+ * <p/>
+ * This mode does not support certain configuration parameters, like
configuring the amount of
+ * driver memory or custom driver command line options. If such
configuration is detected, an
+ * exception will be thrown.
+ * <p/>
+ * This is extremely experimental and should not be used in production
environments.
+ * <p/>
+ * NOTE: SparkSubmit uses system properties to propagate some
configuration value to the app
+ * are run concurrently, they may affect each other's configurations.
+ * <p/>
+ * NOTE: for users running JDK versions older than 8, this option can
add a lot of overhead
+ * to the VM's perm gen.
+ *
+ * @param exceptionHandler Optional handler for handling exceptions in
the app thread.
+ * @param daemon Whether to start a daemon thread.
+ * @return A non-daemon thread that will run the application using
SparkSubmit. The thread will
+ * already be started.
+ */
+ public Thread start(Thread.UncaughtExceptionHandler handler, boolean
daemon) throws IOException {
--- End diff --
Hm, can you motivate the use case of this? Is there a scenario that a user
would prefer to launch it in a thread instead? Also, if we do decide to support
this we should probably name this something more specific, since it's confusing
to have a `start` and a `launch` and have them do different things.
---
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 [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
