Github user rmetzger commented on a diff in the pull request:
https://github.com/apache/flink/pull/189#discussion_r26108160
--- Diff: docs/flink_on_tez_guide.md ---
@@ -0,0 +1,293 @@
+---
+title: "Running Flink on YARN leveraging Tez"
+---
+<!--
+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.
+-->
+
+* This will be replaced by the TOC
+{:toc}
+
+
+<a href="#top"></a>
+
+## Introduction
+
+You can run Flink using Tez as an execution environment. Flink on Tez
+is currently included in *flink-staging* in alpha. All classes are
+localted in the *org.apache.flink.tez* package.
+
+## Why Flink on Tez
+
+[Apache Tez](tez.apache.org) is a scalable data processing
+platform. Tez provides an API for specifying a directed acyclic
+graph (DAG), and functionality for placing the DAG vertices in YARN
+containers, as well as data shuffling. In Flink's architecture,
+Tez is at about the same level as Flink's network stack. While Flink's
+network stack focuses heavily on low latency in order to support
+pipelining, data streaming, and iterative algorithms, Tez
+focuses on scalability and elastic resource usage.
+
+Thus, by replacing Flink's network stack with Tez, users can get
scalability
+and elastic resource usage in shared clusters while retaining Flink's
+APIs, optimizer, and runtime algorithms (local sorts, hash tables, etc).
+
+Flink programs can run almost unmodified using Tez as an execution
+environment. Tez supports local execution (e.g., for debugging), and
+remote execution on YARN.
+
+
+## Local execution
+
+The `LocalTezEnvironment` can be used run programs using the local
+mode provided by Tez. This is for example WordCount using Tez local mode.
+It is identical to a normal Flink WordCount, except that the
`LocalTezEnvironment` is used.
+To run in local Tez mode, you can simply run a Flink on Tez program
+from your IDE (e.g., right click and run).
+
+{% highlight java %}
+public class WordCountExample {
+ public static void main(String[] args) throws Exception {
+ final LocalTezEnvironment env = LocalTezEnvironment.create();
+
+ DataSet<String> text = env.fromElements(
+ "Who's there?",
+ "I think I hear them. Stand, ho! Who's there?");
+
+ DataSet<Tuple2<String, Integer>> wordCounts = text
+ .flatMap(new LineSplitter())
+ .groupBy(0)
+ .sum(1);
+
+ wordCounts.print();
+
+ env.execute("Word Count Example");
+ }
+
+ public static class LineSplitter implements FlatMapFunction<String,
Tuple2<String, Integer>> {
+ @Override
+ public void flatMap(String line, Collector<Tuple2<String,
Integer>> out) {
+ for (String word : line.split(" ")) {
+ out.collect(new Tuple2<String, Integer>(word, 1));
+ }
+ }
+ }
+}
+{% endhighlight %}
+
+## YARN execution
+
+### Setup
+
+- Install Tez on your Hadoop 2 cluster following the instructions from the
+ [Apache Tez website](http://tez.apache.org/install.html). If you are
able to run
+ the examples that ship with Tez, then Tez has been successfully
installed.
+
+- Currently, you need to build Flink yourself to obtain Flink on Tez
+ (the reason is a Hadoop version compatibility: Tez releases artifacts
+ on Maven central with a Hadoop 2.6.0 dependency). Build Flink
+ using `mvn -DskipTests clean package -Dhadoop.version=X.X.X
-Pinclude-tez`.
+ Make sure that the Hadoop version matches the version that Tez uses.
+ Obtain the jar file contained in the Flink distribution under
+ `flink-staging/flink-tez/target/flink-tez-x.y.z-flink-fat-jar.jar`
+ and upload it to some directory in HDFS. E.g., to upload the file
+ to the directory `/apps`, execute
+ {% highlight bash %}
+ $ hadoop fs -put /path/to/flink-tez-x.y.z-flink-fat-jar.jar /apps
+ {% endhighlight %}
+
+- Edit the tez-site.xml configuration file, adding an entry that points to
the
+ location of the file. E.g., assuming that the file is in the directory
`/apps/`,
+ add the following entry to tez-site.xml
+ ~~~<property>
+ <name>tez.aux.uris</name>
+
<value>${fs.default.name}/apps/flink-tez-x.y.z-flink-fat-jar.jar</value>
+ </property>
+ ~~~
+
+- At this point, you should be able to run the pre-packaged examples,
e.g., run WordCount as:
+ {% highlight bash %}
+ $ hadoop jar /path/to/flink-tez-x.y.z-flink-fat-jar.jar
org.apache.flink.tez.ExampleDriver \
--- End diff --
The classname `org.apache.flink.tez.ExampleDriver` is wrong.
---
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.
---
