You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@flink.apache.org by uc...@apache.org on 2015/05/18 11:57:04 UTC
flink-web git commit: Update FAQ
Repository: flink-web
Updated Branches:
refs/heads/asf-site 601788cba -> cc3d1f332
Update FAQ
Project: http://git-wip-us.apache.org/repos/asf/flink-web/repo
Commit: http://git-wip-us.apache.org/repos/asf/flink-web/commit/cc3d1f33
Tree: http://git-wip-us.apache.org/repos/asf/flink-web/tree/cc3d1f33
Diff: http://git-wip-us.apache.org/repos/asf/flink-web/diff/cc3d1f33
Branch: refs/heads/asf-site
Commit: cc3d1f332250725c791d4a3d7d598c6fe5e2e6a1
Parents: 601788c
Author: Ufuk Celebi <uc...@apache.org>
Authored: Mon May 18 11:56:37 2015 +0200
Committer: Ufuk Celebi <uc...@apache.org>
Committed: Mon May 18 11:56:47 2015 +0200
----------------------------------------------------------------------
content/faq.html | 379 +++++++++++++++++++++++++++++++++++++++++++++++++-
faq.md | 327 ++++++++++++++++++++++++++++++++++++++++++-
2 files changed, 699 insertions(+), 7 deletions(-)
----------------------------------------------------------------------
http://git-wip-us.apache.org/repos/asf/flink-web/blob/cc3d1f33/content/faq.html
----------------------------------------------------------------------
diff --git a/content/faq.html b/content/faq.html
index e2e8d80..af34a37 100644
--- a/content/faq.html
+++ b/content/faq.html
@@ -135,13 +135,48 @@
<div class="page-toc">
<ul id="markdown-toc">
- <li><a href="#is-flink-a-hadoop-project" id="markdown-toc-is-flink-a-hadoop-project">Is Flink a Hadoop Project?</a></li>
- <li><a href="#do-i-have-to-install-apache-hadoop-to-use-flink" id="markdown-toc-do-i-have-to-install-apache-hadoop-to-use-flink">Do I have to install Apache Hadoop to use Flink?</a></li>
+ <li><a href="#general" id="markdown-toc-general">General</a> <ul>
+ <li><a href="#is-flink-a-hadoop-project" id="markdown-toc-is-flink-a-hadoop-project">Is Flink a Hadoop Project?</a></li>
+ <li><a href="#do-i-have-to-install-apache-hadoop-to-use-flink" id="markdown-toc-do-i-have-to-install-apache-hadoop-to-use-flink">Do I have to install Apache Hadoop to use Flink?</a></li>
+ </ul>
+ </li>
+ <li><a href="#usage" id="markdown-toc-usage">Usage</a> <ul>
+ <li><a href="#how-do-i-assess-the-progress-of-a-flink-program" id="markdown-toc-how-do-i-assess-the-progress-of-a-flink-program">How do I assess the progress of a Flink program?</a></li>
+ <li><a href="#how-can-i-figure-out-why-a-program-failed" id="markdown-toc-how-can-i-figure-out-why-a-program-failed">How can I figure out why a program failed?</a></li>
+ <li><a href="#how-do-i-debug-flink-programs" id="markdown-toc-how-do-i-debug-flink-programs">How do I debug Flink programs?</a></li>
+ <li><a href="#what-is-the-parallelism-how-do-i-set-it" id="markdown-toc-what-is-the-parallelism-how-do-i-set-it">What is the parallelism? How do I set it?</a></li>
+ </ul>
+ </li>
+ <li><a href="#errors" id="markdown-toc-errors">Errors</a> <ul>
+ <li><a href="#why-am-i-getting-a-nonserializableexception-" id="markdown-toc-why-am-i-getting-a-nonserializableexception-">Why am I getting a “NonSerializableException” ?</a></li>
+ <li><a href="#in-scala-api-i-get-an-error-about-implicit-values-and-evidence-parameters" id="markdown-toc-in-scala-api-i-get-an-error-about-implicit-values-and-evidence-parameters">In Scala API, I get an error about implicit values and evidence parameters</a></li>
+ <li><a href="#i-get-an-error-message-saying-that-not-enough-buffers-are-available-how-do-i-fix-this" id="markdown-toc-i-get-an-error-message-saying-that-not-enough-buffers-are-available-how-do-i-fix-this">I get an error message saying that not enough buffers are available. How do I fix this?</a></li>
+ <li><a href="#my-job-fails-early-with-a-javaioeofexception-what-could-be-the-cause" id="markdown-toc-my-job-fails-early-with-a-javaioeofexception-what-could-be-the-cause">My job fails early with a java.io.EOFException. What could be the cause?</a></li>
+ <li><a href="#in-eclipse-i-get-compilation-errors-in-the-scala-projects" id="markdown-toc-in-eclipse-i-get-compilation-errors-in-the-scala-projects">In Eclipse, I get compilation errors in the Scala projects</a></li>
+ <li><a href="#my-program-does-not-compute-the-correct-result-why-are-my-custom-key-types" id="markdown-toc-my-program-does-not-compute-the-correct-result-why-are-my-custom-key-types">My program does not compute the correct result. Why are my custom key types</a></li>
+ <li><a href="#i-get-a-javalanginstantiationexception-for-my-data-type-what-is-wrong" id="markdown-toc-i-get-a-javalanginstantiationexception-for-my-data-type-what-is-wrong">I get a java.lang.InstantiationException for my data type, what is wrong?</a></li>
+ <li><a href="#i-cant-stop-flink-with-the-provided-stop-scripts-what-can-i-do" id="markdown-toc-i-cant-stop-flink-with-the-provided-stop-scripts-what-can-i-do">I can’t stop Flink with the provided stop-scripts. What can I do?</a></li>
+ <li><a href="#i-got-an-outofmemoryexception-what-can-i-do" id="markdown-toc-i-got-an-outofmemoryexception-what-can-i-do">I got an OutOfMemoryException. What can I do?</a></li>
+ <li><a href="#why-do-the-taskmanager-log-files-become-so-huge" id="markdown-toc-why-do-the-taskmanager-log-files-become-so-huge">Why do the TaskManager log files become so huge?</a></li>
+ </ul>
+ </li>
+ <li><a href="#yarn-deployment" id="markdown-toc-yarn-deployment">YARN Deployment</a> <ul>
+ <li><a href="#the-yarn-session-runs-only-for-a-few-seconds" id="markdown-toc-the-yarn-session-runs-only-for-a-few-seconds">The YARN session runs only for a few seconds</a></li>
+ <li><a href="#the-yarn-session-crashes-with-a-hdfs-permission-exception-during-startup" id="markdown-toc-the-yarn-session-crashes-with-a-hdfs-permission-exception-during-startup">The YARN session crashes with a HDFS permission exception during startup</a></li>
+ </ul>
+ </li>
+ <li><a href="#features" id="markdown-toc-features">Features</a> <ul>
+ <li><a href="#what-kind-of-fault-tolerance-does-flink-provide" id="markdown-toc-what-kind-of-fault-tolerance-does-flink-provide">What kind of fault-tolerance does Flink provide?</a></li>
+ <li><a href="#are-hadoop-like-utilities-such-as-counters-and-the-distributedcache-supported" id="markdown-toc-are-hadoop-like-utilities-such-as-counters-and-the-distributedcache-supported">Are Hadoop-like utilities, such as Counters and the DistributedCache supported?</a></li>
+ </ul>
+ </li>
</ul>
</div>
-<h2 id="is-flink-a-hadoop-project">Is Flink a Hadoop Project?</h2>
+<h2 id="general">General</h2>
+
+<h3 id="is-flink-a-hadoop-project">Is Flink a Hadoop Project?</h3>
<p>Flink is a data processing system and an <strong>alternative to Hadoop’s
MapReduce component</strong>. It comes with its <em>own runtime</em>, rather than building on top
@@ -152,7 +187,7 @@ manager (YARN) to provision cluster resources. Since most Flink users are
using Hadoop HDFS to store their data, Flink already ships the required libraries to
access HDFS.</p>
-<h2 id="do-i-have-to-install-apache-hadoop-to-use-flink">Do I have to install Apache Hadoop to use Flink?</h2>
+<h3 id="do-i-have-to-install-apache-hadoop-to-use-flink">Do I have to install Apache Hadoop to use Flink?</h3>
<p><strong>No</strong>. Flink can run <strong>without</strong> a Hadoop installation. However, a <em>very common</em>
setup is to use Flink to analyze data stored in the Hadoop Distributed
@@ -165,6 +200,342 @@ YARN</a>
is Hadoop’s cluster resource manager that allows to use
different execution engines next to each other on a cluster.</p>
+<h2 id="usage">Usage</h2>
+
+<h3 id="how-do-i-assess-the-progress-of-a-flink-program">How do I assess the progress of a Flink program?</h3>
+
+<p>There are a multiple of ways to track the progress of a Flink program:</p>
+
+<ul>
+ <li>The JobManager (the master of the distributed system) starts a web interface
+to observe program execution. In runs on port 8081 by default (configured in
+<code>conf/flink-config.yml</code>).</li>
+ <li>When you start a program from the command line, it will print the status
+changes of all operators as the program progresses through the operations.</li>
+ <li>All status changes are also logged to the JobManager’s log file.</li>
+</ul>
+
+<h3 id="how-can-i-figure-out-why-a-program-failed">How can I figure out why a program failed?</h3>
+
+<ul>
+ <li>The JobManager web frontend (by default on port 8081) displays the exceptions
+of failed tasks.</li>
+ <li>If you run the program from the command-line, task exceptions are printed to
+the standard error stream and shown on the console.</li>
+ <li>Both the command line and the web interface allow you to figure out which
+parallel task first failed and caused the other tasks to cancel the execution.</li>
+ <li>Failing tasks and the corresponding exceptions are reported in the log files
+of the master and the worker where the exception occurred
+(<code>log/flink-<user>-jobmanager-<host>.log</code> and
+<code>log/flink-<user>-taskmanager-<host>.log</code>).</li>
+</ul>
+
+<h3 id="how-do-i-debug-flink-programs">How do I debug Flink programs?</h3>
+
+<ul>
+ <li>When you start a program locally with the <a href="http://ci.apache.org/projects/flink/flink-docs-master/apis/local_execution.html">LocalExecutor</a>,
+you can place breakpoints in your functions and debug them like normal
+Java/Scala programs.</li>
+ <li>The <a href="http://ci.apache.org/projects/flink/flink-docs-master/apis/programming_guide.html#accumulators--counters">Accumulators</a> are very helpful in
+tracking the behavior of the parallel execution. They allow you to gather
+information inside the program’s operations and show them after the program
+execution.</li>
+</ul>
+
+<h3 id="what-is-the-parallelism-how-do-i-set-it">What is the parallelism? How do I set it?</h3>
+
+<p>In Flink programs, the parallelism determines how operations are split into
+individual tasks which are assigned to task slots. Each node in a cluster has at
+least one task slot. The total number of task slots is the number of all task slots
+on all machines. If the parallelism is set to <code>N</code>, Flink tries to divide an
+operation into <code>N</code> parallel tasks which can be computed concurrently using the
+available task slots. The number of task slots should be equal to the
+parallelism to ensure that all tasks can be computed in a task slot concurrently.</p>
+
+<p><strong>Note</strong>: Not all operations can be divided into multiple tasks. For example, a
+<code>GroupReduce</code> operation without a grouping has to be performed with a
+parallelism of 1 because the entire group needs to be present at exactly one
+node to perform the reduce operation. Flink will determine whether the
+parallelism has to be 1 and set it accordingly.</p>
+
+<p>The parallelism can be set in numerous ways to ensure a fine-grained control
+over the execution of a Flink program. See
+the <a href="http://ci.apache.org/projects/flink/flink-docs-master/setup/config.html#common-options">Configuration guide</a> for detailed instructions on how to
+set the parallelism. Also check out <a href="http://ci.apache.org/projects/flink/flink-docs-master/setup/config.html#configuring-taskmanager-processing-slots">this figure</a> detailing
+how the processing slots and parallelism are related to each other.</p>
+
+<h2 id="errors">Errors</h2>
+
+<h3 id="why-am-i-getting-a-nonserializableexception-">Why am I getting a “NonSerializableException” ?</h3>
+
+<p>All functions in Flink must be serializable, as defined by <a href="http://docs.oracle.com/javase/7/docs/api/java/io/Serializable.html">java.io.Serializable</a>.
+Since all function interfaces are serializable, the exception means that one
+of the fields used in your function is not serializable.</p>
+
+<p>In particular, if your function is an inner class, or anonymous inner class,
+it contains a hidden reference to the enclosing class (usually called <code>this$0</code>, if you look
+at the function in the debugger). If the enclosing class is not serializable, this is probably
+the source of the error. Solutions are to</p>
+
+<ul>
+ <li>make the function a standalone class, or a static inner class (no more reference to the enclosing class)</li>
+ <li>make the enclosing class serializable</li>
+ <li>use a Java 8 lambda function.</li>
+</ul>
+
+<h3 id="in-scala-api-i-get-an-error-about-implicit-values-and-evidence-parameters">In Scala API, I get an error about implicit values and evidence parameters</h3>
+
+<p>It means that the implicit value for the type information could not be provided.
+Make sure that you have an <code>import org.apache.flink.api.scala._</code> statement in your code.</p>
+
+<p>If you are using flink operations inside functions or classes that take
+generic parameters a TypeInformation must be available for that parameter.
+This can be achieved by using a context bound:</p>
+
+<div class="highlight"><pre><code class="language-scala"><span class="k">def</span> <span class="n">myFunction</span><span class="o">[</span><span class="kt">T:</span> <span class="kt">TypeInformation</span><span class="o">](</span><span class="n">input</span><span class="k">:</span> <span class="kt">DataSet</span><span class="o">[</span><span class="kt">T</span><span class="o">])</span><span class="k">:</span> <span class="kt">DataSet</span><span class="o">[</span><span class="kt">Seq</span><span class="o">[</span><span class="kt">T</span><span class="o">]]</span> <span class="k">=</span> <span class="o">{</span>
+ <span class="n">input</span><span class="o">.</span><span class="n">reduceGroup</span><span class="o">(</span> <span class="n">i</span> <span class="k">=></span> <span class="n">i</span><span class="o">.</span><span class="n">toSeq</span> <span class="o">)</span>
+<span class="o">}</span></code></pre></div>
+
+<p>See <a href="http://ci.apache.org/projects/flink/flink-docs-master/internals/types_serialization.html">Type Extraction and Serialization</a> for
+an in-depth discussion of how Flink handles types.</p>
+
+<h3 id="i-get-an-error-message-saying-that-not-enough-buffers-are-available-how-do-i-fix-this">I get an error message saying that not enough buffers are available. How do I fix this?</h3>
+
+<p>If you run Flink in a massively parallel setting (100+ parallel threads),
+you need to adapt the number of network buffers via the config parameter
+<code>taskmanager.network.numberOfBuffers</code>.
+As a rule-of-thumb, the number of buffers should be at least
+<code>4 * numberOfNodes * numberOfTasksPerNode^2</code>. See
+<a href="http://ci.apache.org/projects/flink/flink-docs-master/setup/config.html">Configuration Reference</a> for details.</p>
+
+<h3 id="my-job-fails-early-with-a-javaioeofexception-what-could-be-the-cause">My job fails early with a java.io.EOFException. What could be the cause?</h3>
+
+<p>The most common case for these exception is when Flink is set up with the
+wrong HDFS version. Because different HDFS versions are often not compatible
+with each other, the connection between the filesystem master and the client
+breaks.</p>
+
+<div class="highlight"><pre><code class="language-bash">Call to <host:port> failed on <span class="nb">local </span>exception: java.io.EOFException
+ at org.apache.hadoop.ipc.Client.wrapException<span class="o">(</span>Client.java:775<span class="o">)</span>
+ at org.apache.hadoop.ipc.Client.call<span class="o">(</span>Client.java:743<span class="o">)</span>
+ at org.apache.hadoop.ipc.RPC<span class="nv">$Invoker</span>.invoke<span class="o">(</span>RPC.java:220<span class="o">)</span>
+ at <span class="nv">$Proxy0</span>.getProtocolVersion<span class="o">(</span>Unknown Source<span class="o">)</span>
+ at org.apache.hadoop.ipc.RPC.getProxy<span class="o">(</span>RPC.java:359<span class="o">)</span>
+ at org.apache.hadoop.hdfs.DFSClient.createRPCNamenode<span class="o">(</span>DFSClient.java:106<span class="o">)</span>
+ at org.apache.hadoop.hdfs.DFSClient.<init><span class="o">(</span>DFSClient.java:207<span class="o">)</span>
+ at org.apache.hadoop.hdfs.DFSClient.<init><span class="o">(</span>DFSClient.java:170<span class="o">)</span>
+ at org.apache.hadoop.hdfs.DistributedFileSystem.initialize<span class="o">(</span>DistributedFileSystem.java:82<span class="o">)</span>
+ at org.apache.flinkruntime.fs.hdfs.DistributedFileSystem.initialize<span class="o">(</span>DistributedFileSystem.java:276</code></pre></div>
+
+<p>Please refer to the <a href="/downloads.html#maven">download page</a> and
+the https://github.com/apache/flink/tree/master/README.md
+for details on how to set up Flink for different Hadoop and HDFS versions.</p>
+
+<h3 id="in-eclipse-i-get-compilation-errors-in-the-scala-projects">In Eclipse, I get compilation errors in the Scala projects</h3>
+
+<p>Flink uses a new feature of the Scala compiler (called “quasiquotes”) that have not yet been properly
+integrated with the Eclipse Scala plugin. In order to make this feature available in Eclipse, you
+need to manually configure the <em>flink-scala</em> project to use a <em>compiler plugin</em>:</p>
+
+<ul>
+ <li>Right click on <em>flink-scala</em> and choose “Properties”</li>
+ <li>Select “Scala Compiler” and click on the “Advanced” tab. (If you do not have that, you probably have not set up Eclipse for Scala properly.)</li>
+ <li>Check the box “Use Project Settings”</li>
+ <li>In the field “Xplugin”, put the path “/home/<user-name>/.m2/repository/org/scalamacros/paradise_2.10.4/2.0.1/paradise_2.10.4-2.0.1.jar"</user-name></li>
+ <li>NOTE: You have to build Flink with Maven on the command line first, to make sure the plugin is downloaded.</li>
+</ul>
+
+<h3 id="my-program-does-not-compute-the-correct-result-why-are-my-custom-key-types">My program does not compute the correct result. Why are my custom key types</h3>
+<p>are not grouped/joined correctly?</p>
+
+<p>Keys must correctly implement the methods <code>java.lang.Object#hashCode()</code>,
+<code>java.lang.Object#equals(Object o)</code>, and <code>java.util.Comparable#compareTo(...)</code>.
+These methods are always backed with default implementations which are usually
+inadequate. Therefore, all keys must override <code>hashCode()</code> and <code>equals(Object o)</code>.</p>
+
+<h3 id="i-get-a-javalanginstantiationexception-for-my-data-type-what-is-wrong">I get a java.lang.InstantiationException for my data type, what is wrong?</h3>
+
+<p>All data type classes must be public and have a public nullary constructor
+(constructor with no arguments). Further more, the classes must not be abstract
+or interfaces. If the classes are internal classes, they must be public and
+static.</p>
+
+<h3 id="i-cant-stop-flink-with-the-provided-stop-scripts-what-can-i-do">I can’t stop Flink with the provided stop-scripts. What can I do?</h3>
+
+<p>Stopping the processes sometimes takes a few seconds, because the shutdown may
+do some cleanup work.</p>
+
+<p>In some error cases it happens that the JobManager or TaskManager cannot be
+stopped with the provided stop-scripts (<code>bin/stop-local.sh</code> or <code>bin/stop-
+cluster.sh</code>). You can kill their processes on Linux/Mac as follows:</p>
+
+<ul>
+ <li>Determine the process id (pid) of the JobManager / TaskManager process. You
+can use the <code>jps</code> command on Linux(if you have OpenJDK installed) or command
+<code>ps -ef | grep java</code> to find all Java processes.</li>
+ <li>Kill the process with <code>kill -9 <pid></code>, where <code>pid</code> is the process id of the
+affected JobManager or TaskManager process.</li>
+</ul>
+
+<p>On Windows, the TaskManager shows a table of all processes and allows you to
+destroy a process by right its entry.</p>
+
+<h3 id="i-got-an-outofmemoryexception-what-can-i-do">I got an OutOfMemoryException. What can I do?</h3>
+
+<p>These exceptions occur usually when the functions in the program consume a lot
+of memory by collection large numbers of objects, for example in lists or maps.
+The OutOfMemoryExceptions in Java are kind of tricky. The exception is not
+necessarily thrown by the component that allocated most of the memory but by the
+component that tried to requested the latest bit of memory that could not be
+provided.</p>
+
+<p>There are two ways to go about this:</p>
+
+<ol>
+ <li>
+ <p>See whether you can use less memory inside the functions. For example, use
+arrays of primitive types instead of object types.</p>
+ </li>
+ <li>
+ <p>Reduce the memory that Flink reserves for its own processing. The
+TaskManager reserves a certain portion of the available memory for sorting,
+hashing, caching, network buffering, etc. That part of the memory is unavailable
+to the user-defined functions. By reserving it, the system can guarantee to not
+run out of memory on large inputs, but to plan with the available memory and
+destage operations to disk, if necessary. By default, the system reserves around
+70% of the memory. If you frequently run applications that need more memory in
+the user-defined functions, you can reduce that value using the configuration
+entries <code>taskmanager.memory.fraction</code> or <code>taskmanager.memory.size</code>. See the
+<a href="http://ci.apache.org/projects/flink/flink-docs-master/setup/config.html">Configuration Reference</a> for details. This will leave more memory to JVM heap,
+but may cause data processing tasks to go to disk more often.</p>
+ </li>
+</ol>
+
+<h3 id="why-do-the-taskmanager-log-files-become-so-huge">Why do the TaskManager log files become so huge?</h3>
+
+<p>Check the logging behavior of your jobs. Emitting logging per or tuple may be
+helpful to debug jobs in small setups with tiny data sets, it becomes very
+inefficient and disk space consuming if used for large input data.</p>
+
+<h2 id="yarn-deployment">YARN Deployment</h2>
+
+<h3 id="the-yarn-session-runs-only-for-a-few-seconds">The YARN session runs only for a few seconds</h3>
+
+<p>The <code>./bin/yarn-session.sh</code> script is intended to run while the YARN-session is
+open. In some error cases however, the script immediately stops running. The
+output looks like this:</p>
+
+<div class="highlight"><pre><code>07:34:27,004 INFO org.apache.hadoop.yarn.client.api.impl.YarnClientImpl - Submitted application application_1395604279745_273123 to ResourceManager at jobtracker-host
+Flink JobManager is now running on worker1:6123
+JobManager Web Interface: http://jobtracker-host:54311/proxy/application_1295604279745_273123/
+07:34:51,528 INFO org.apache.flinkyarn.Client - Application application_1295604279745_273123 finished with state FINISHED at 1398152089553
+07:34:51,529 INFO org.apache.flinkyarn.Client - Killing the Flink-YARN application.
+07:34:51,529 INFO org.apache.hadoop.yarn.client.api.impl.YarnClientImpl - Killing application application_1295604279745_273123
+07:34:51,534 INFO org.apache.flinkyarn.Client - Deleting files in hdfs://user/marcus/.flink/application_1295604279745_273123
+07:34:51,559 INFO org.apache.flinkyarn.Client - YARN Client is shutting down
+</code></pre></div>
+
+<p>The problem here is that the Application Master (AM) is stopping and the YARN client assumes that the application has finished.</p>
+
+<p>There are three possible reasons for that behavior:</p>
+
+<ul>
+ <li>
+ <p>The ApplicationMaster exited with an exception. To debug that error, have a
+look in the logfiles of the container. The <code>yarn-site.xml</code> file contains the
+configured path. The key for the path is <code>yarn.nodemanager.log-dirs</code>, the
+default value is <code>${yarn.log.dir}/userlogs</code>.</p>
+ </li>
+ <li>
+ <p>YARN has killed the container that runs the ApplicationMaster. This case
+happens when the AM used too much memory or other resources beyond YARN’s
+limits. In this case, you’ll find error messages in the nodemanager logs on
+the host.</p>
+ </li>
+ <li>
+ <p>The operating system has shut down the JVM of the AM. This can happen if the
+YARN configuration is wrong and more memory than physically available is
+configured. Execute <code>dmesg</code> on the machine where the AM was running to see if
+this happened. You see messages from Linux’ <a href="http://linux-mm.org/OOM_Killer">OOM killer</a>.</p>
+ </li>
+</ul>
+
+<h3 id="the-yarn-session-crashes-with-a-hdfs-permission-exception-during-startup">The YARN session crashes with a HDFS permission exception during startup</h3>
+
+<p>While starting the YARN session, you are receiving an exception like this:</p>
+
+<div class="highlight"><pre><code>Exception in thread "main" org.apache.hadoop.security.AccessControlException: Permission denied: user=robert, access=WRITE, inode="/user/robert":hdfs:supergroup:drwxr-xr-x
+ at org.apache.hadoop.hdfs.server.namenode.FSPermissionChecker.check(FSPermissionChecker.java:234)
+ at org.apache.hadoop.hdfs.server.namenode.FSPermissionChecker.check(FSPermissionChecker.java:214)
+ at org.apache.hadoop.hdfs.server.namenode.FSPermissionChecker.checkPermission(FSPermissionChecker.java:158)
+ at org.apache.hadoop.hdfs.server.namenode.FSNamesystem.checkPermission(FSNamesystem.java:5193)
+ at org.apache.hadoop.hdfs.server.namenode.FSNamesystem.checkPermission(FSNamesystem.java:5175)
+ at org.apache.hadoop.hdfs.server.namenode.FSNamesystem.checkAncestorAccess(FSNamesystem.java:5149)
+ at org.apache.hadoop.hdfs.server.namenode.FSNamesystem.startFileInternal(FSNamesystem.java:2090)
+ at org.apache.hadoop.hdfs.server.namenode.FSNamesystem.startFileInt(FSNamesystem.java:2043)
+ at org.apache.hadoop.hdfs.server.namenode.FSNamesystem.startFile(FSNamesystem.java:1996)
+ at org.apache.hadoop.hdfs.server.namenode.NameNodeRpcServer.create(NameNodeRpcServer.java:491)
+ at org.apache.hadoop.hdfs.protocolPB.ClientNamenodeProtocolServerSideTranslatorPB.create(ClientNamenodeProtocolServerSideTranslatorPB.java:301)
+ at org.apache.hadoop.hdfs.protocol.proto.ClientNamenodeProtocolProtos$ClientNamenodeProtocol$2.callBlockingMethod(ClientNamenodeProtocolProtos.java:59570)
+ at org.apache.hadoop.ipc.ProtobufRpcEngine$Server$ProtoBufRpcInvoker.call(ProtobufRpcEngine.java:585)
+ at org.apache.hadoop.ipc.RPC$Server.call(RPC.java:928)
+ at org.apache.hadoop.ipc.Server$Handler$1.run(Server.java:2053)
+ at org.apache.hadoop.ipc.Server$Handler$1.run(Server.java:2049)
+ at java.security.AccessController.doPrivileged(Native Method)
+ at javax.security.auth.Subject.doAs(Subject.java:396)
+ at org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1491)
+ at org.apache.hadoop.ipc.Server$Handler.run(Server.java:2047)
+
+ at sun.reflect.NativeConstructorAccessorImpl.newInstance0(Native Method)
+ at sun.reflect.NativeConstructorAccessorImpl.newInstance(NativeConstructorAccessorImpl.java:39)
+ at sun.reflect.DelegatingConstructorAccessorImpl.newInstance(DelegatingConstructorAccessorImpl.java:27)
+ at java.lang.reflect.Constructor.newInstance(Constructor.java:513)
+ at org.apache.hadoop.ipc.RemoteException.instantiateException(RemoteException.java:106)
+ at org.apache.hadoop.ipc.RemoteException.unwrapRemoteException(RemoteException.java:73)
+ at org.apache.hadoop.hdfs.DFSOutputStream.newStreamForCreate(DFSOutputStream.java:1393)
+ at org.apache.hadoop.hdfs.DFSClient.create(DFSClient.java:1382)
+ at org.apache.hadoop.hdfs.DFSClient.create(DFSClient.java:1307)
+ at org.apache.hadoop.hdfs.DistributedFileSystem$6.doCall(DistributedFileSystem.java:384)
+ at org.apache.hadoop.hdfs.DistributedFileSystem$6.doCall(DistributedFileSystem.java:380)
+ at org.apache.hadoop.fs.FileSystemLinkResolver.resolve(FileSystemLinkResolver.java:81)
+ at org.apache.hadoop.hdfs.DistributedFileSystem.create(DistributedFileSystem.java:380)
+ at org.apache.hadoop.hdfs.DistributedFileSystem.create(DistributedFileSystem.java:324)
+ at org.apache.hadoop.fs.FileSystem.create(FileSystem.java:905)
+ at org.apache.hadoop.fs.FileSystem.create(FileSystem.java:886)
+ at org.apache.hadoop.fs.FileSystem.create(FileSystem.java:783)
+ at org.apache.hadoop.fs.FileUtil.copy(FileUtil.java:365)
+ at org.apache.hadoop.fs.FileUtil.copy(FileUtil.java:338)
+ at org.apache.hadoop.fs.FileSystem.copyFromLocalFile(FileSystem.java:2021)
+ at org.apache.hadoop.fs.FileSystem.copyFromLocalFile(FileSystem.java:1989)
+ at org.apache.hadoop.fs.FileSystem.copyFromLocalFile(FileSystem.java:1954)
+ at org.apache.flinkyarn.Utils.setupLocalResource(Utils.java:176)
+ at org.apache.flinkyarn.Client.run(Client.java:362)
+ at org.apache.flinkyarn.Client.main(Client.java:568)
+</code></pre></div>
+
+<p>The reason for this error is, that the home directory of the user <strong>in HDFS</strong>
+has the wrong permissions. The user (in this case <code>robert</code>) can not create
+directories in his own home directory.</p>
+
+<p>Flink creates a <code>.flink/</code> directory in the users home directory
+where it stores the Flink jar and configuration file.</p>
+
+<h2 id="features">Features</h2>
+
+<h3 id="what-kind-of-fault-tolerance-does-flink-provide">What kind of fault-tolerance does Flink provide?</h3>
+
+<p>Flink can restart failed jobs. Mid-query fault tolerance will go into the
+open source project in the next versions.</p>
+
+<h3 id="are-hadoop-like-utilities-such-as-counters-and-the-distributedcache-supported">Are Hadoop-like utilities, such as Counters and the DistributedCache supported?</h3>
+
+<p><a href="http://ci.apache.org/projects/flink/flink-docs-master/apis/programming_guide.html#accumulators--counters">Flink’s Accumulators</a> work very similar like
+[Hadoop’s counters, but are more powerful.</p>
+
+<p>Flink has a https://github.com/apache/flink/tree/master//flink-core/src/main/java/org/apache/flink/api/common/cache/DistributedCache.java that is deeply integrated with the APIs. Please refer to the https://github.com/apache/flink/tree/master//flink-java/src/main/java/org/apache/flink/api/java/ExecutionEnvironment.java#L561 for details on how to use it.</p>
+
+<p>In order to make data sets available on all tasks, we encourage you to use <a href="http://ci.apache.org/projects/flink/flink-docs-master/apis/programming_guide.html#broadcast-variables">Broadcast Variables</a> instead. They are more efficient and easier to use than the distributed cache.</p>
+
</div>
</div>
http://git-wip-us.apache.org/repos/asf/flink-web/blob/cc3d1f33/faq.md
----------------------------------------------------------------------
diff --git a/faq.md b/faq.md
index 3111814..ec45656 100644
--- a/faq.md
+++ b/faq.md
@@ -6,7 +6,9 @@ The following questions are frequently asked with regard to the Flink project **
{% toc %}
-## Is Flink a Hadoop Project?
+## General
+
+### Is Flink a Hadoop Project?
Flink is a data processing system and an **alternative to Hadoop's
MapReduce component**. It comes with its *own runtime*, rather than building on top
@@ -17,7 +19,7 @@ manager (YARN) to provision cluster resources. Since most Flink users are
using Hadoop HDFS to store their data, Flink already ships the required libraries to
access HDFS.
-## Do I have to install Apache Hadoop to use Flink?
+### Do I have to install Apache Hadoop to use Flink?
**No**. Flink can run **without** a Hadoop installation. However, a *very common*
setup is to use Flink to analyze data stored in the Hadoop Distributed
@@ -28,4 +30,323 @@ Additionally, we provide a special YARN Enabled download of Flink for
users with an existing Hadoop YARN cluster. [Apache Hadoop
YARN](http://hadoop.apache.org/docs/r2.2.0/hadoop-yarn/hadoop-yarn-site/YARN.html)
is Hadoop's cluster resource manager that allows to use
-different execution engines next to each other on a cluster.
\ No newline at end of file
+different execution engines next to each other on a cluster.
+
+## Usage
+
+### How do I assess the progress of a Flink program?
+
+There are a multiple of ways to track the progress of a Flink program:
+
+- The JobManager (the master of the distributed system) starts a web interface
+to observe program execution. In runs on port 8081 by default (configured in
+`conf/flink-config.yml`).
+- When you start a program from the command line, it will print the status
+changes of all operators as the program progresses through the operations.
+- All status changes are also logged to the JobManager's log file.
+
+### How can I figure out why a program failed?
+
+- The JobManager web frontend (by default on port 8081) displays the exceptions
+of failed tasks.
+- If you run the program from the command-line, task exceptions are printed to
+the standard error stream and shown on the console.
+- Both the command line and the web interface allow you to figure out which
+parallel task first failed and caused the other tasks to cancel the execution.
+- Failing tasks and the corresponding exceptions are reported in the log files
+of the master and the worker where the exception occurred
+(`log/flink-<user>-jobmanager-<host>.log` and
+`log/flink-<user>-taskmanager-<host>.log`).
+
+### How do I debug Flink programs?
+
+- When you start a program locally with the [LocalExecutor](http://ci.apache.org/projects/flink/flink-docs-master/apis/local_execution.html),
+you can place breakpoints in your functions and debug them like normal
+Java/Scala programs.
+- The [Accumulators](http://ci.apache.org/projects/flink/flink-docs-master/apis/programming_guide.html#accumulators--counters) are very helpful in
+tracking the behavior of the parallel execution. They allow you to gather
+information inside the program's operations and show them after the program
+execution.
+
+### What is the parallelism? How do I set it?
+
+In Flink programs, the parallelism determines how operations are split into
+individual tasks which are assigned to task slots. Each node in a cluster has at
+least one task slot. The total number of task slots is the number of all task slots
+on all machines. If the parallelism is set to `N`, Flink tries to divide an
+operation into `N` parallel tasks which can be computed concurrently using the
+available task slots. The number of task slots should be equal to the
+parallelism to ensure that all tasks can be computed in a task slot concurrently.
+
+**Note**: Not all operations can be divided into multiple tasks. For example, a
+`GroupReduce` operation without a grouping has to be performed with a
+parallelism of 1 because the entire group needs to be present at exactly one
+node to perform the reduce operation. Flink will determine whether the
+parallelism has to be 1 and set it accordingly.
+
+The parallelism can be set in numerous ways to ensure a fine-grained control
+over the execution of a Flink program. See
+the [Configuration guide](http://ci.apache.org/projects/flink/flink-docs-master/setup/config.html#common-options) for detailed instructions on how to
+set the parallelism. Also check out [this figure](http://ci.apache.org/projects/flink/flink-docs-master/setup/config.html#configuring-taskmanager-processing-slots) detailing
+how the processing slots and parallelism are related to each other.
+
+## Errors
+
+### Why am I getting a "NonSerializableException" ?
+
+All functions in Flink must be serializable, as defined by [java.io.Serializable](http://docs.oracle.com/javase/7/docs/api/java/io/Serializable.html).
+Since all function interfaces are serializable, the exception means that one
+of the fields used in your function is not serializable.
+
+In particular, if your function is an inner class, or anonymous inner class,
+it contains a hidden reference to the enclosing class (usually called `this$0`, if you look
+at the function in the debugger). If the enclosing class is not serializable, this is probably
+the source of the error. Solutions are to
+
+- make the function a standalone class, or a static inner class (no more reference to the enclosing class)
+- make the enclosing class serializable
+- use a Java 8 lambda function.
+
+### In Scala API, I get an error about implicit values and evidence parameters
+
+It means that the implicit value for the type information could not be provided.
+Make sure that you have an `import org.apache.flink.api.scala._` statement in your code.
+
+If you are using flink operations inside functions or classes that take
+generic parameters a TypeInformation must be available for that parameter.
+This can be achieved by using a context bound:
+
+~~~scala
+def myFunction[T: TypeInformation](input: DataSet[T]): DataSet[Seq[T]] = {
+ input.reduceGroup( i => i.toSeq )
+}
+~~~
+
+See [Type Extraction and Serialization](http://ci.apache.org/projects/flink/flink-docs-master/internals/types_serialization.html) for
+an in-depth discussion of how Flink handles types.
+
+### I get an error message saying that not enough buffers are available. How do I fix this?
+
+If you run Flink in a massively parallel setting (100+ parallel threads),
+you need to adapt the number of network buffers via the config parameter
+`taskmanager.network.numberOfBuffers`.
+As a rule-of-thumb, the number of buffers should be at least
+`4 * numberOfNodes * numberOfTasksPerNode^2`. See
+[Configuration Reference](http://ci.apache.org/projects/flink/flink-docs-master/setup/config.html) for details.
+
+### My job fails early with a java.io.EOFException. What could be the cause?
+
+The most common case for these exception is when Flink is set up with the
+wrong HDFS version. Because different HDFS versions are often not compatible
+with each other, the connection between the filesystem master and the client
+breaks.
+
+~~~bash
+Call to <host:port> failed on local exception: java.io.EOFException
+ at org.apache.hadoop.ipc.Client.wrapException(Client.java:775)
+ at org.apache.hadoop.ipc.Client.call(Client.java:743)
+ at org.apache.hadoop.ipc.RPC$Invoker.invoke(RPC.java:220)
+ at $Proxy0.getProtocolVersion(Unknown Source)
+ at org.apache.hadoop.ipc.RPC.getProxy(RPC.java:359)
+ at org.apache.hadoop.hdfs.DFSClient.createRPCNamenode(DFSClient.java:106)
+ at org.apache.hadoop.hdfs.DFSClient.<init>(DFSClient.java:207)
+ at org.apache.hadoop.hdfs.DFSClient.<init>(DFSClient.java:170)
+ at org.apache.hadoop.hdfs.DistributedFileSystem.initialize(DistributedFileSystem.java:82)
+ at org.apache.flinkruntime.fs.hdfs.DistributedFileSystem.initialize(DistributedFileSystem.java:276
+~~~
+
+Please refer to the [download page]({{ site.baseurl }}/downloads.html#maven) and
+the {% github README.md master "build instructions" %}
+for details on how to set up Flink for different Hadoop and HDFS versions.
+
+### In Eclipse, I get compilation errors in the Scala projects
+
+Flink uses a new feature of the Scala compiler (called "quasiquotes") that have not yet been properly
+integrated with the Eclipse Scala plugin. In order to make this feature available in Eclipse, you
+need to manually configure the *flink-scala* project to use a *compiler plugin*:
+
+- Right click on *flink-scala* and choose "Properties"
+- Select "Scala Compiler" and click on the "Advanced" tab. (If you do not have that, you probably have not set up Eclipse for Scala properly.)
+- Check the box "Use Project Settings"
+- In the field "Xplugin", put the path "/home/<user-name>/.m2/repository/org/scalamacros/paradise_2.10.4/2.0.1/paradise_2.10.4-2.0.1.jar"
+- NOTE: You have to build Flink with Maven on the command line first, to make sure the plugin is downloaded.
+
+### My program does not compute the correct result. Why are my custom key types
+are not grouped/joined correctly?
+
+Keys must correctly implement the methods `java.lang.Object#hashCode()`,
+`java.lang.Object#equals(Object o)`, and `java.util.Comparable#compareTo(...)`.
+These methods are always backed with default implementations which are usually
+inadequate. Therefore, all keys must override `hashCode()` and `equals(Object o)`.
+
+### I get a java.lang.InstantiationException for my data type, what is wrong?
+
+All data type classes must be public and have a public nullary constructor
+(constructor with no arguments). Further more, the classes must not be abstract
+or interfaces. If the classes are internal classes, they must be public and
+static.
+
+### I can't stop Flink with the provided stop-scripts. What can I do?
+
+Stopping the processes sometimes takes a few seconds, because the shutdown may
+do some cleanup work.
+
+In some error cases it happens that the JobManager or TaskManager cannot be
+stopped with the provided stop-scripts (`bin/stop-local.sh` or `bin/stop-
+cluster.sh`). You can kill their processes on Linux/Mac as follows:
+
+- Determine the process id (pid) of the JobManager / TaskManager process. You
+can use the `jps` command on Linux(if you have OpenJDK installed) or command
+`ps -ef | grep java` to find all Java processes.
+- Kill the process with `kill -9 <pid>`, where `pid` is the process id of the
+affected JobManager or TaskManager process.
+
+On Windows, the TaskManager shows a table of all processes and allows you to
+destroy a process by right its entry.
+
+### I got an OutOfMemoryException. What can I do?
+
+These exceptions occur usually when the functions in the program consume a lot
+of memory by collection large numbers of objects, for example in lists or maps.
+The OutOfMemoryExceptions in Java are kind of tricky. The exception is not
+necessarily thrown by the component that allocated most of the memory but by the
+component that tried to requested the latest bit of memory that could not be
+provided.
+
+There are two ways to go about this:
+
+1. See whether you can use less memory inside the functions. For example, use
+arrays of primitive types instead of object types.
+
+2. Reduce the memory that Flink reserves for its own processing. The
+TaskManager reserves a certain portion of the available memory for sorting,
+hashing, caching, network buffering, etc. That part of the memory is unavailable
+to the user-defined functions. By reserving it, the system can guarantee to not
+run out of memory on large inputs, but to plan with the available memory and
+destage operations to disk, if necessary. By default, the system reserves around
+70% of the memory. If you frequently run applications that need more memory in
+the user-defined functions, you can reduce that value using the configuration
+entries `taskmanager.memory.fraction` or `taskmanager.memory.size`. See the
+[Configuration Reference](http://ci.apache.org/projects/flink/flink-docs-master/setup/config.html) for details. This will leave more memory to JVM heap,
+but may cause data processing tasks to go to disk more often.
+
+### Why do the TaskManager log files become so huge?
+
+Check the logging behavior of your jobs. Emitting logging per or tuple may be
+helpful to debug jobs in small setups with tiny data sets, it becomes very
+inefficient and disk space consuming if used for large input data.
+
+## YARN Deployment
+
+### The YARN session runs only for a few seconds
+
+The `./bin/yarn-session.sh` script is intended to run while the YARN-session is
+open. In some error cases however, the script immediately stops running. The
+output looks like this:
+
+~~~
+07:34:27,004 INFO org.apache.hadoop.yarn.client.api.impl.YarnClientImpl - Submitted application application_1395604279745_273123 to ResourceManager at jobtracker-host
+Flink JobManager is now running on worker1:6123
+JobManager Web Interface: http://jobtracker-host:54311/proxy/application_1295604279745_273123/
+07:34:51,528 INFO org.apache.flinkyarn.Client - Application application_1295604279745_273123 finished with state FINISHED at 1398152089553
+07:34:51,529 INFO org.apache.flinkyarn.Client - Killing the Flink-YARN application.
+07:34:51,529 INFO org.apache.hadoop.yarn.client.api.impl.YarnClientImpl - Killing application application_1295604279745_273123
+07:34:51,534 INFO org.apache.flinkyarn.Client - Deleting files in hdfs://user/marcus/.flink/application_1295604279745_273123
+07:34:51,559 INFO org.apache.flinkyarn.Client - YARN Client is shutting down
+~~~
+
+The problem here is that the Application Master (AM) is stopping and the YARN client assumes that the application has finished.
+
+There are three possible reasons for that behavior:
+
+- The ApplicationMaster exited with an exception. To debug that error, have a
+look in the logfiles of the container. The `yarn-site.xml` file contains the
+configured path. The key for the path is `yarn.nodemanager.log-dirs`, the
+default value is `${yarn.log.dir}/userlogs`.
+
+- YARN has killed the container that runs the ApplicationMaster. This case
+happens when the AM used too much memory or other resources beyond YARN's
+limits. In this case, you'll find error messages in the nodemanager logs on
+the host.
+
+- The operating system has shut down the JVM of the AM. This can happen if the
+YARN configuration is wrong and more memory than physically available is
+configured. Execute `dmesg` on the machine where the AM was running to see if
+this happened. You see messages from Linux' [OOM killer](http://linux-mm.org/OOM_Killer).
+
+### The YARN session crashes with a HDFS permission exception during startup
+
+While starting the YARN session, you are receiving an exception like this:
+
+~~~
+Exception in thread "main" org.apache.hadoop.security.AccessControlException: Permission denied: user=robert, access=WRITE, inode="/user/robert":hdfs:supergroup:drwxr-xr-x
+ at org.apache.hadoop.hdfs.server.namenode.FSPermissionChecker.check(FSPermissionChecker.java:234)
+ at org.apache.hadoop.hdfs.server.namenode.FSPermissionChecker.check(FSPermissionChecker.java:214)
+ at org.apache.hadoop.hdfs.server.namenode.FSPermissionChecker.checkPermission(FSPermissionChecker.java:158)
+ at org.apache.hadoop.hdfs.server.namenode.FSNamesystem.checkPermission(FSNamesystem.java:5193)
+ at org.apache.hadoop.hdfs.server.namenode.FSNamesystem.checkPermission(FSNamesystem.java:5175)
+ at org.apache.hadoop.hdfs.server.namenode.FSNamesystem.checkAncestorAccess(FSNamesystem.java:5149)
+ at org.apache.hadoop.hdfs.server.namenode.FSNamesystem.startFileInternal(FSNamesystem.java:2090)
+ at org.apache.hadoop.hdfs.server.namenode.FSNamesystem.startFileInt(FSNamesystem.java:2043)
+ at org.apache.hadoop.hdfs.server.namenode.FSNamesystem.startFile(FSNamesystem.java:1996)
+ at org.apache.hadoop.hdfs.server.namenode.NameNodeRpcServer.create(NameNodeRpcServer.java:491)
+ at org.apache.hadoop.hdfs.protocolPB.ClientNamenodeProtocolServerSideTranslatorPB.create(ClientNamenodeProtocolServerSideTranslatorPB.java:301)
+ at org.apache.hadoop.hdfs.protocol.proto.ClientNamenodeProtocolProtos$ClientNamenodeProtocol$2.callBlockingMethod(ClientNamenodeProtocolProtos.java:59570)
+ at org.apache.hadoop.ipc.ProtobufRpcEngine$Server$ProtoBufRpcInvoker.call(ProtobufRpcEngine.java:585)
+ at org.apache.hadoop.ipc.RPC$Server.call(RPC.java:928)
+ at org.apache.hadoop.ipc.Server$Handler$1.run(Server.java:2053)
+ at org.apache.hadoop.ipc.Server$Handler$1.run(Server.java:2049)
+ at java.security.AccessController.doPrivileged(Native Method)
+ at javax.security.auth.Subject.doAs(Subject.java:396)
+ at org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1491)
+ at org.apache.hadoop.ipc.Server$Handler.run(Server.java:2047)
+
+ at sun.reflect.NativeConstructorAccessorImpl.newInstance0(Native Method)
+ at sun.reflect.NativeConstructorAccessorImpl.newInstance(NativeConstructorAccessorImpl.java:39)
+ at sun.reflect.DelegatingConstructorAccessorImpl.newInstance(DelegatingConstructorAccessorImpl.java:27)
+ at java.lang.reflect.Constructor.newInstance(Constructor.java:513)
+ at org.apache.hadoop.ipc.RemoteException.instantiateException(RemoteException.java:106)
+ at org.apache.hadoop.ipc.RemoteException.unwrapRemoteException(RemoteException.java:73)
+ at org.apache.hadoop.hdfs.DFSOutputStream.newStreamForCreate(DFSOutputStream.java:1393)
+ at org.apache.hadoop.hdfs.DFSClient.create(DFSClient.java:1382)
+ at org.apache.hadoop.hdfs.DFSClient.create(DFSClient.java:1307)
+ at org.apache.hadoop.hdfs.DistributedFileSystem$6.doCall(DistributedFileSystem.java:384)
+ at org.apache.hadoop.hdfs.DistributedFileSystem$6.doCall(DistributedFileSystem.java:380)
+ at org.apache.hadoop.fs.FileSystemLinkResolver.resolve(FileSystemLinkResolver.java:81)
+ at org.apache.hadoop.hdfs.DistributedFileSystem.create(DistributedFileSystem.java:380)
+ at org.apache.hadoop.hdfs.DistributedFileSystem.create(DistributedFileSystem.java:324)
+ at org.apache.hadoop.fs.FileSystem.create(FileSystem.java:905)
+ at org.apache.hadoop.fs.FileSystem.create(FileSystem.java:886)
+ at org.apache.hadoop.fs.FileSystem.create(FileSystem.java:783)
+ at org.apache.hadoop.fs.FileUtil.copy(FileUtil.java:365)
+ at org.apache.hadoop.fs.FileUtil.copy(FileUtil.java:338)
+ at org.apache.hadoop.fs.FileSystem.copyFromLocalFile(FileSystem.java:2021)
+ at org.apache.hadoop.fs.FileSystem.copyFromLocalFile(FileSystem.java:1989)
+ at org.apache.hadoop.fs.FileSystem.copyFromLocalFile(FileSystem.java:1954)
+ at org.apache.flinkyarn.Utils.setupLocalResource(Utils.java:176)
+ at org.apache.flinkyarn.Client.run(Client.java:362)
+ at org.apache.flinkyarn.Client.main(Client.java:568)
+~~~
+
+The reason for this error is, that the home directory of the user **in HDFS**
+has the wrong permissions. The user (in this case `robert`) can not create
+directories in his own home directory.
+
+Flink creates a `.flink/` directory in the users home directory
+where it stores the Flink jar and configuration file.
+
+## Features
+
+### What kind of fault-tolerance does Flink provide?
+
+Flink can restart failed jobs. Mid-query fault tolerance will go into the
+open source project in the next versions.
+
+### Are Hadoop-like utilities, such as Counters and the DistributedCache supported?
+
+[Flink's Accumulators](http://ci.apache.org/projects/flink/flink-docs-master/apis/programming_guide.html#accumulators--counters) work very similar like
+[Hadoop's counters, but are more powerful.
+
+Flink has a {% github /flink-core/src/main/java/org/apache/flink/api/common/cache/DistributedCache.java "Distributed Cache" %} that is deeply integrated with the APIs. Please refer to the {% github /flink-java/src/main/java/org/apache/flink/api/java/ExecutionEnvironment.java#L561 "JavaDocs" %} for details on how to use it.
+
+In order to make data sets available on all tasks, we encourage you to use [Broadcast Variables](http://ci.apache.org/projects/flink/flink-docs-master/apis/programming_guide.html#broadcast-variables) instead. They are more efficient and easier to use than the distributed cache.