You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@flink.apache.org by se...@apache.org on 2016/12/14 14:10:38 UTC
[4/4] flink git commit: [FLINK-5258] [docs] Reorganize the docs to
improve navigation and reduce duplication
[FLINK-5258] [docs] Reorganize the docs to improve navigation and reduce duplication
This closes #2940
Project: http://git-wip-us.apache.org/repos/asf/flink/repo
Commit: http://git-wip-us.apache.org/repos/asf/flink/commit/79d7e301
Tree: http://git-wip-us.apache.org/repos/asf/flink/tree/79d7e301
Diff: http://git-wip-us.apache.org/repos/asf/flink/diff/79d7e301
Branch: refs/heads/master
Commit: 79d7e3017efe7c96e449e6f339fd7184ef3d1ba2
Parents: e4c767a
Author: David Anderson <da...@alpinegizmo.com>
Authored: Fri Nov 25 15:44:21 2016 +0100
Committer: Stephan Ewen <se...@apache.org>
Committed: Wed Dec 14 14:56:33 2016 +0100
----------------------------------------------------------------------
docs/Gemfile | 10 +-
docs/Gemfile.lock | 72 ++--
docs/_includes/sidenav.html | 4 +-
docs/build_docs.sh | 6 +-
docs/concepts/index.md | 230 +-----------
docs/concepts/programming-model.md | 173 +++++++++
docs/concepts/runtime.md | 127 +++++++
docs/dev/api_concepts.md | 471 +------------------------
docs/dev/apis.md | 24 --
docs/dev/batch/examples.md | 12 +-
docs/dev/batch/index.md | 28 +-
docs/dev/batch/iterations.md | 8 +-
docs/dev/cluster_execution.md | 74 +---
docs/dev/connectors/cassandra.md | 2 +-
docs/dev/connectors/elasticsearch.md | 2 +-
docs/dev/connectors/elasticsearch2.md | 4 +-
docs/dev/connectors/filesystem_sink.md | 2 +-
docs/dev/connectors/guarantees.md | 143 ++++++++
docs/dev/connectors/index.md | 4 +-
docs/dev/connectors/kafka.md | 2 +-
docs/dev/connectors/kinesis.md | 2 +-
docs/dev/connectors/nifi.md | 2 +-
docs/dev/connectors/rabbitmq.md | 2 +-
docs/dev/connectors/redis.md | 2 +-
docs/dev/connectors/twitter.md | 2 +-
docs/dev/custom_serializers.md | 112 ++++++
docs/dev/datastream_api.md | 16 +-
docs/dev/event_time.md | 4 +-
docs/dev/execution.md | 24 ++
docs/dev/execution_configuration.md | 86 +++++
docs/dev/execution_plans.md | 80 +++++
docs/dev/index.md | 4 +-
docs/dev/java8.md | 4 +-
docs/dev/libraries.md | 2 +-
docs/dev/libs/cep.md | 4 +-
docs/dev/libs/gelly/index.md | 2 +-
docs/dev/libs/ml/index.md | 4 +-
docs/dev/libs/ml/quickstart.md | 2 +-
docs/dev/linking.md | 94 +++++
docs/dev/linking_with_flink.md | 146 ++++++++
docs/dev/local_execution.md | 4 +-
docs/dev/packaging.md | 77 ++++
docs/dev/parallel.md | 175 +++++++++
docs/dev/quickstarts.md | 24 --
docs/dev/scala_api_extensions.md | 4 +-
docs/dev/scala_shell.md | 6 +-
docs/dev/state.md | 76 +++-
docs/dev/state_backends.md | 4 +-
docs/dev/table_api.md | 6 +-
docs/dev/types_serialization.md | 5 +-
docs/dev/windows.md | 4 +-
docs/examples/index.md | 39 ++
docs/index.md | 28 +-
docs/internals/add_operator.md | 253 -------------
docs/internals/components.md | 59 ++++
docs/internals/general_arch.md | 101 ------
docs/internals/ide_setup.md | 55 +--
docs/internals/index.md | 5 +-
docs/internals/state_backends.md | 27 +-
docs/monitoring/best_practices.md | 94 +----
docs/monitoring/index.md | 6 +-
docs/page/css/flink.css | 21 ++
docs/quickstart/java_api_quickstart.md | 11 +-
docs/quickstart/run_example_quickstart.md | 8 +-
docs/quickstart/scala_api_quickstart.md | 8 +-
docs/quickstart/setup_quickstart.md | 225 ++++++++----
docs/redirects/concepts.md | 2 +-
docs/setup/building.md | 6 +-
docs/setup/checkpoints.md | 57 +++
docs/setup/cluster_setup.md | 60 ++--
docs/setup/config.md | 11 +-
docs/setup/fault_tolerance.md | 250 ++-----------
docs/setup/flink_on_windows.md | 86 +++++
docs/setup/index.md | 6 +-
docs/setup/local_setup.md | 153 --------
docs/setup/savepoints.md | 13 +-
docs/setup/security-ssl.md | 2 +-
docs/start/index.md | 26 ++
78 files changed, 2030 insertions(+), 1959 deletions(-)
----------------------------------------------------------------------
http://git-wip-us.apache.org/repos/asf/flink/blob/79d7e301/docs/Gemfile
----------------------------------------------------------------------
diff --git a/docs/Gemfile b/docs/Gemfile
index b7a974c..c1a1555 100644
--- a/docs/Gemfile
+++ b/docs/Gemfile
@@ -17,11 +17,13 @@
################################################################################
source 'https://rubygems.org'
-
-ruby '>=1.9.0'
+ruby '~> 2.3.0'
# Dependencies required to build the Flink docs
-gem 'jekyll', '2.5.3'
-gem 'kramdown', '1.10.0'
+gem 'jekyll', '~> 3.3.0'
+gem 'kramdown', '~> 1.13.0'
gem 'pygments.rb', '0.6.3'
gem 'therubyracer', '0.12.2'
+group :jekyll_plugins do
+ gem 'hawkins'
+end
http://git-wip-us.apache.org/repos/asf/flink/blob/79d7e301/docs/Gemfile.lock
----------------------------------------------------------------------
diff --git a/docs/Gemfile.lock b/docs/Gemfile.lock
index bf86631..da7e04a 100644
--- a/docs/Gemfile.lock
+++ b/docs/Gemfile.lock
@@ -1,83 +1,67 @@
GEM
remote: https://rubygems.org/
specs:
- addressable (2.4.0)
- blankslate (2.1.2.4)
- classifier-reborn (2.0.4)
- fast-stemmer (~> 1.0)
- coffee-script (2.4.1)
- coffee-script-source
- execjs
- coffee-script-source (1.11.1)
- colorator (0.1)
- execjs (2.7.0)
- faraday (0.9.2)
- multipart-post (>= 1.2, < 3)
- fast-stemmer (1.0.2)
+ addressable (2.5.0)
+ public_suffix (~> 2.0, >= 2.0.2)
+ colorator (1.1.0)
+ em-websocket (0.5.1)
+ eventmachine (>= 0.12.9)
+ http_parser.rb (~> 0.6.0)
+ eventmachine (1.2.1)
ffi (1.9.14)
- jekyll (2.5.3)
- classifier-reborn (~> 2.0)
- colorator (~> 0.1)
- jekyll-coffeescript (~> 1.0)
- jekyll-gist (~> 1.0)
- jekyll-paginate (~> 1.0)
+ forwardable-extended (2.6.0)
+ hawkins (2.0.4)
+ em-websocket (~> 0.5)
+ jekyll (~> 3.1)
+ http_parser.rb (0.6.0)
+ jekyll (3.3.1)
+ addressable (~> 2.4)
+ colorator (~> 1.0)
jekyll-sass-converter (~> 1.0)
jekyll-watch (~> 1.1)
kramdown (~> 1.3)
- liquid (~> 2.6.1)
+ liquid (~> 3.0)
mercenary (~> 0.3.3)
- pygments.rb (~> 0.6.0)
- redcarpet (~> 3.1)
+ pathutil (~> 0.9)
+ rouge (~> 1.7)
safe_yaml (~> 1.0)
- toml (~> 0.1.0)
- jekyll-coffeescript (1.0.1)
- coffee-script (~> 2.2)
- jekyll-gist (1.4.0)
- octokit (~> 4.3.0)
- jekyll-paginate (1.1.0)
jekyll-sass-converter (1.5.0)
sass (~> 3.4)
jekyll-watch (1.5.0)
listen (~> 3.0, < 3.1)
- kramdown (1.10.0)
+ kramdown (1.13.1)
libv8 (3.16.14.17)
- liquid (2.6.3)
+ liquid (3.0.6)
listen (3.0.8)
rb-fsevent (~> 0.9, >= 0.9.4)
rb-inotify (~> 0.9, >= 0.9.7)
mercenary (0.3.6)
- multipart-post (2.0.0)
- octokit (4.3.0)
- sawyer (~> 0.7.0, >= 0.5.3)
- parslet (1.5.0)
- blankslate (~> 2.0)
+ pathutil (0.14.0)
+ forwardable-extended (~> 2.6)
posix-spawn (0.3.12)
+ public_suffix (2.0.4)
pygments.rb (0.6.3)
posix-spawn (~> 0.3.6)
yajl-ruby (~> 1.2.0)
rb-fsevent (0.9.8)
rb-inotify (0.9.7)
ffi (>= 0.5.0)
- redcarpet (3.3.4)
ref (2.0.0)
+ rouge (1.11.1)
safe_yaml (1.0.4)
sass (3.4.22)
- sawyer (0.7.0)
- addressable (>= 2.3.5, < 2.5)
- faraday (~> 0.8, < 0.10)
therubyracer (0.12.2)
libv8 (~> 3.16.14.0)
ref
- toml (0.1.2)
- parslet (~> 1.5.0)
yajl-ruby (1.2.1)
PLATFORMS
ruby
DEPENDENCIES
- jekyll (= 2.5.3)
- kramdown (= 1.10.0)
+ hawkins
+ jekyll (~> 3.3.0)
+ kramdown (~> 1.13.0)
pygments.rb (= 0.6.3)
therubyracer (= 0.12.2)
@@ -85,4 +69,4 @@ RUBY VERSION
ruby 2.3.1p112
BUNDLED WITH
- 1.13.2
+ 1.13.6
http://git-wip-us.apache.org/repos/asf/flink/blob/79d7e301/docs/_includes/sidenav.html
----------------------------------------------------------------------
diff --git a/docs/_includes/sidenav.html b/docs/_includes/sidenav.html
index b56bcf2..e3bb2d7 100644
--- a/docs/_includes/sidenav.html
+++ b/docs/_includes/sidenav.html
@@ -101,6 +101,8 @@ level is determined by 'nav-pos'.
{% capture target %}"{{ site.baseurl }}{{ this.url }}"{% if active %} class="active"{% endif %}{% endcapture %}
{% capture overview_target %}"{{ site.baseurl }}{{ this.url }}"{% if this.url == page.url %} class="active"{% endif %}{% endcapture %}
+ {% if this.section-break %}<hr class="section-break"></hr>{% endif %}
+
{% assign pos = pos | plus: 1 %}
{% if this.nav-id %}
{% assign children = (site.pages | where: "nav-parent_id" , this.nav-id | sort: "nav-pos") %}
@@ -124,7 +126,7 @@ level is determined by 'nav-pos'.
{% endif %}
{% endfor %}
<li class="divider"></li>
- <li><a href="http://flink.apache.org"><i class="fa fa-external-link" aria-hidden="true"></i> Project Page</a></li>
+ <li><a href="http://flink.apache.org"><i class="fa fa-external-link title" aria-hidden="true"></i> Project Page</a></li>
</ul>
<div class="sidenav-search-box">
http://git-wip-us.apache.org/repos/asf/flink/blob/79d7e301/docs/build_docs.sh
----------------------------------------------------------------------
diff --git a/docs/build_docs.sh b/docs/build_docs.sh
index 2981c4a..df83ac4 100755
--- a/docs/build_docs.sh
+++ b/docs/build_docs.sh
@@ -46,11 +46,15 @@ DOCS_DST=${DOCS_SRC}/content
JEKYLL_CMD="build"
# if -p flag is provided, serve site on localhost
-while getopts ":p" opt; do
+# -i is like -p, but incremental (which has some issues, but is very fast)
+while getopts ":p:i" opt; do
case $opt in
p)
JEKYLL_CMD="serve --baseurl= --watch"
;;
+ i)
+ JEKYLL_CMD="liveserve --baseurl= --watch --incremental"
+ ;;
esac
done
http://git-wip-us.apache.org/repos/asf/flink/blob/79d7e301/docs/concepts/index.md
----------------------------------------------------------------------
diff --git a/docs/concepts/index.md b/docs/concepts/index.md
index a9638451..25565e8 100644
--- a/docs/concepts/index.md
+++ b/docs/concepts/index.md
@@ -2,8 +2,12 @@
title: Concepts
nav-id: concepts
nav-pos: 1
-nav-title: '<i class="fa fa-map-o" aria-hidden="true"></i> Concepts'
+nav-title: '<i class="fa fa-map-o title appetizer" aria-hidden="true"></i> Concepts'
nav-parent_id: root
+section-break: true
+layout: redirect
+redirect: /concepts/programming-model.html
+permalink: /concepts/index.html
---
<!--
Licensed to the Apache Software Foundation (ASF) under one
@@ -23,227 +27,3 @@ 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}
-
-## Programs and Dataflows
-
-The basic building blocks of Flink programs are **streams** and **transformations** (note that a DataSet is internally
-also a stream). A *stream* is an intermediate result, and a *transformation* is an operation that takes one or more streams
-as input, and computes one or more result streams from them.
-
-When executed, Flink programs are mapped to **streaming dataflows**, consisting of **streams** and transformation **operators**.
-Each dataflow starts with one or more **sources** and ends in one or more **sinks**. The dataflows may resemble
-arbitrary **directed acyclic graphs** *(DAGs)*. (Special forms of cycles are permitted via *iteration* constructs, we
-omit this here for simplicity).
-
-In most cases, there is a one-to-one correspondence between the transformations in the programs and the operators
-in the dataflow. Sometimes, however, one transformation may consist of multiple transformation operators.
-
-<img src="{{ site.baseurl }}/fig/program_dataflow.svg" alt="A DataStream program, and its dataflow." class="offset" width="80%" />
-
-{% top %}
-
-### Parallel Dataflows
-
-Programs in Flink are inherently parallel and distributed. *Streams* are split into **stream partitions** and
-*operators* are split into **operator subtasks**. The operator subtasks execute independently from each other,
-in different threads and on different machines or containers.
-
-The number of operator subtasks is the **parallelism** of that particular operator. The parallelism of a stream
-is always that of its producing operator. Different operators of the program may have a different parallelism.
-
-<img src="{{ site.baseurl }}/fig/parallel_dataflow.svg" alt="A parallel dataflow" class="offset" width="80%" />
-
-Streams can transport data between two operators in a *one-to-one* (or *forwarding*) pattern, or in a *redistributing* pattern:
-
- - **One-to-one** streams (for example between the *source* and the *map()* operators) preserves partitioning and order of
- elements. That means that subtask[1] of the *map()* operator will see the same elements in the same order, as they
- were produced by subtask[1] of the *source* operator.
-
- - **Redistributing** streams (between *map()* and *keyBy/window*, as well as between *keyBy/window* and *sink*) change
- the partitioning of streams. Each *operator subtask* sends data to different target subtasks,
- depending on the selected transformation. Examples are *keyBy()* (re-partitions by hash code), *broadcast()*, or
- *rebalance()* (random redistribution).
- In a *redistributing* exchange, order among elements is only preserved for each pair of sending- and receiving
- task (for example subtask[1] of *map()* and subtask[2] of *keyBy/window*).
-
-{% top %}
-
-### Tasks & Operator Chains
-
-For distributed execution, Flink *chains* operator subtasks together into *tasks*. Each task is executed by one thread.
-Chaining operators together into tasks is a useful optimization: it reduces the overhead of thread-to-thread
-handover and buffering, and increases overall throughput while decreasing latency.
-The chaining behavior can be configured in the APIs.
-
-The sample dataflow in the figure below is executed with five subtasks, and hence with five parallel threads.
-
-<img src="{{ site.baseurl }}/fig/tasks_chains.svg" alt="Operator chaining into Tasks" class="offset" width="80%" />
-
-{% top %}
-
-## Distributed Execution
-
-**Master, Worker, Client**
-
-The Flink runtime consists of two types of processes:
-
- - The **master** processes (also called *JobManagers*) coordinate the distributed execution. They schedule tasks, coordinate
- checkpoints, coordinate recovery on failures, etc.
-
- There is always at least one master process. A high-availability setup will have multiple master processes, out of
- which one is always the *leader*, and the others are *standby*.
-
- - The **worker** processes (also called *TaskManagers*) execute the *tasks* (or more specifically, the subtasks) of a dataflow,
- and buffer and exchange the data *streams*.
-
- There must always be at least one worker process.
-
-The master and worker processes can be started in an arbitrary fashion: Directly on the machines, via containers, or via
-resource frameworks like YARN. Workers connect to masters, announcing themselves as available, and get work assigned.
-
-The **client** is not part of the runtime and program execution, but is used to prepare and send a dataflow to the master.
-After that, the client can disconnect, or stay connected to receive progress reports. The client runs either as part of the
-Java/Scala program that triggers the execution, or in the command line process `./bin/flink run ...`.
-
-<img src="{{ site.baseurl }}/fig/processes.svg" alt="The processes involved in executing a Flink dataflow" class="offset" width="80%" />
-
-{% top %}
-
-### Workers, Slots, Resources
-
-Each worker (TaskManager) is a *JVM process*, and may execute one or more subtasks in separate threads.
-To control how many tasks a worker accepts, a worker has so called **task slots** (at least one).
-
-Each *task slot* represents a fixed subset of resources of the TaskManager. A TaskManager with three slots, for example,
-will dedicate 1/3 of its managed memory to each slot. Slotting the resources means that a subtask will not
-compete with subtasks from other jobs for managed memory, but instead has a certain amount of reserved
-managed memory. Note that no CPU isolation happens here, slots currently only separate managed memory of tasks.
-
-Adjusting the number of task slots thus allows users to define how subtasks are isolated against each other.
-Having one slot per TaskManager means each task group runs in a separate JVM (which can be started in a
-separate container, for example). Having multiple slots
-means more subtasks share the same JVM. Tasks in the same JVM share TCP connections (via multiplexing) and
-heartbeats messages. They may also share data sets and data structures, thus reducing the per-task overhead.
-
-<img src="{{ site.baseurl }}/fig/tasks_slots.svg" alt="A TaskManager with Task Slots and Tasks" class="offset" width="80%" />
-
-By default, Flink allows subtasks to share slots, if they are subtasks of different tasks, but from the same
-job. The result is that one slot may hold an entire pipeline of the job. Allowing this *slot sharing*
-has two main benefits:
-
- - A Flink cluster needs exactly as many tasks slots, as the highest parallelism used in the job.
- No need to calculate how many tasks (with varying parallelism) a program contains in total.
-
- - It is easier to get better resource utilization. Without slot sharing, the non-intensive
- *source/map()* subtasks would block as many resources as the resource intensive *window* subtasks.
- With slot sharing, increasing the base parallelism from two to six yields full utilization of the
- slotted resources, while still making sure that each TaskManager gets only a fair share of the
- heavy subtasks.
-
-The slot sharing behavior can be controlled in the APIs, to prevent sharing where it is undesirable.
-The mechanism for that are the *resource groups*, which define what (sub)tasks may share slots.
-
-As a rule-of-thumb, a good default number of task slots would be the number of CPU cores.
-With hyper threading, each slot then takes 2 or more hardware thread contexts.
-
-<img src="{{ site.baseurl }}/fig/slot_sharing.svg" alt="TaskManagers with shared Task Slots" class="offset" width="80%" />
-
-{% top %}
-
-## Time and Windows
-
-Aggregating events (e.g., counts, sums) works slightly differently on streams than in batch processing.
-For example, it is impossible to first count all elements in the stream and then return the count,
-because streams are in general infinite (unbounded). Instead, aggregates on streams (counts, sums, etc),
-are scoped by **windows**, such as *"count over the last 5 minutes"*, or *"sum of the last 100 elements"*.
-
-Windows can be *time driven* (example: every 30 seconds) or *data driven* (example: every 100 elements).
-One typically distinguishes different types of windows, such as *tumbling windows* (no overlap),
-*sliding windows* (with overlap), and *session windows* (gap of activity).
-
-<img src="{{ site.baseurl }}/fig/windows.svg" alt="Time- and Count Windows" class="offset" width="80%" />
-
-More window examples can be found in this [blog post](https://flink.apache.org/news/2015/12/04/Introducing-windows.html).
-
-{% top %}
-
-### Time
-
-When referring to time in a streaming program (for example to define windows), one can refer to different notions
-of time:
-
- - **Event Time** is the time when an event was created. It is usually described by a timestamp in the events,
- for example attached by the producing sensor, or the producing service. Flink accesses event timestamps
- via [timestamp assigners]({{ site.baseurl }}/dev/event_timestamps_watermarks.html).
-
- - **Ingestion time** is the time when an event enters the Flink dataflow at the source operator.
-
- - **Processing Time** is the local time at each operator that performs a time-based operation.
-
-<img src="{{ site.baseurl }}/fig/event_ingestion_processing_time.svg" alt="Event Time, Ingestion Time, and Processing Time" class="offset" width="80%" />
-
-More details on how to handle time are in the [event time docs]({{ site.baseurl }}/dev/event_time.html).
-
-{% top %}
-
-## State and Fault Tolerance
-
-While many operations in a dataflow simply look at one individual *event at a time* (for example an event parser),
-some operations remember information across individual events (for example window operators).
-These operations are called **stateful**.
-
-The state of stateful operations is maintained in what can be thought of as an embedded key/value store.
-The state is partitioned and distributed strictly together with the streams that are read by the
-stateful operators. Hence, access the key/value state is only possible on *keyed streams*, after a *keyBy()* function,
-and is restricted to the values of the current event's key. Aligning the keys of streams and state
-makes sure that all state updates are local operations, guaranteeing consistency without transaction overhead.
-This alignment also allows Flink to redistribute the state and adjust the stream partitioning transparently.
-
-<img src="{{ site.baseurl }}/fig/state_partitioning.svg" alt="State and Partitioning" class="offset" width="50%" />
-
-{% top %}
-
-### Checkpoints for Fault Tolerance
-
-Flink implements fault tolerance using a combination of **stream replay** and **checkpoints**. A checkpoint
-defines a consistent point in streams and state from which a streaming dataflow can resume, and maintain consistency
-*(exactly-once processing semantics)*. The events and state updates since the last checkpoint are replayed from the input streams.
-
-The checkpoint interval is a means of trading off the overhead of fault tolerance during execution, with the recovery time (the amount
-of events that need to be replayed).
-
-More details on checkpoints and fault tolerance are in the [fault tolerance docs]({{ site.baseurl }}/internals/stream_checkpointing.html).
-
-<img src="{{ site.baseurl }}/fig/checkpoints.svg" alt="checkpoints and snapshots" class="offset" width="60%" />
-
-{% top %}
-
-### State Backends
-
-The exact data structures in which the key/values indexes are stored depend on the chosen **state backend**. One state backend
-stores data in an in-memory hash map, another state backend uses [RocksDB](http://rocksdb.org) as the key/value index.
-In addition to defining the data structure that holds the state, the state backends also implements the logic to
-take a point-in-time snapshot of the key/value state and store that snapshot as part of a checkpoint.
-
-{% top %}
-
-## Batch on Streaming
-
-Flink executes batch programs as a special case of streaming programs, where the streams are bounded (finite number of elements).
-A *DataSet* is treated internally as a stream of data. The concepts above thus apply to batch programs in the
-same way as well as they apply to streaming programs, with minor exceptions:
-
- - Programs in the DataSet API do not use checkpoints. Recovery happens by fully replaying the streams.
- That is possible, because inputs are bounded. This pushes the cost more towards the recovery,
- but makes the regular processing cheaper, because it avoids checkpoints.
-
- - Stateful operation in the DataSet API use simplified in-memory/out-of-core data structures, rather than
- key/value indexes.
-
- - The DataSet API introduces special synchronized (superstep-based) iterations, which are only possible on
- bounded streams. For details, check out the [iteration docs]({{ site.baseurl }}/dev/batch/iterations.html).
-
-{% top %}
http://git-wip-us.apache.org/repos/asf/flink/blob/79d7e301/docs/concepts/programming-model.md
----------------------------------------------------------------------
diff --git a/docs/concepts/programming-model.md b/docs/concepts/programming-model.md
new file mode 100644
index 0000000..5ab6b8f
--- /dev/null
+++ b/docs/concepts/programming-model.md
@@ -0,0 +1,173 @@
+---
+title: Dataflow Programming Model
+nav-id: programming-model
+nav-pos: 1
+nav-title: Programming Model
+nav-parent_id: concepts
+---
+<!--
+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}
+
+## Programs and Dataflows
+
+The basic building blocks of Flink programs are **streams** and **transformations**. (Note that the
+DataSets used in Flink's batch API are also streams internally -- more about that
+later.) Conceptually a *stream* is a never-ending flow of data records, and a *transformation* is an
+operation that takes one or more streams as input, and produces one or more output streams as a
+result.
+
+When executed, Flink programs are mapped to **streaming dataflows**, consisting of **streams** and transformation **operators**.
+Each dataflow starts with one or more **sources** and ends in one or more **sinks**. The dataflows resemble
+arbitrary **directed acyclic graphs** *(DAGs)*. Although special forms of cycles are permitted via
+*iteration* constructs, for the most part we will gloss over this for simplicity.
+
+<img src="{{ site.baseurl }}/fig/program_dataflow.svg" alt="A DataStream program, and its dataflow." class="offset" width="80%" />
+
+Often there is a one-to-one correspondence between the transformations in the programs and the operators
+in the dataflow. Sometimes, however, one transformation may consist of multiple transformation operators.
+
+{% top %}
+
+## Parallel Dataflows
+
+Programs in Flink are inherently parallel and distributed. This parallelism is expressed in Flink's
+DataStream API with the *keyBy()* operator, which can be thought of as a declaration that the stream can
+be operated on in parallel for different values of the key.
+
+*Streams* are split into **stream partitions**, and *operators* are split into **operator
+subtasks**. The operator subtasks are independent of one another, and execute in different threads
+and possibly on different machines or containers.
+
+The number of operator subtasks is the **parallelism** of that particular operator. The parallelism of a stream
+is always that of its producing operator. Different operators of the same program may have different
+levels of parallelism.
+
+<img src="{{ site.baseurl }}/fig/parallel_dataflow.svg" alt="A parallel dataflow" class="offset" width="80%" />
+
+Streams can transport data between two operators in a *one-to-one* (or *forwarding*) pattern, or in a *redistributing* pattern:
+
+ - **One-to-one** streams (for example between the *Source* and the *map()* operators in the figure
+ above) preserve the partitioning and ordering of the
+ elements. That means that subtask[1] of the *map()* operator will see the same elements in the same order as they
+ were produced by subtask[1] of the *Source* operator.
+
+ - **Redistributing** streams (as between *map()* and *keyBy/window* above, as well as between
+ *keyBy/window* and *Sink*) change the partitioning of streams. Each *operator subtask* sends
+ data to different target subtasks, depending on the selected transformation. Examples are
+ *keyBy()* (which re-partitions by hashing the key), *broadcast()*, or *rebalance()* (which
+ re-partitions randomly). In a *redistributing* exchange the ordering among the elements is
+ only preserved within each pair of sending and receiving subtasks (for example, subtask[1]
+ of *map()* and subtask[2] of *keyBy/window*). So in this example, the ordering within each key
+ is preserved, but the parallelism does introduce non-determinism regarding the order in
+ which the aggregated results for different keys arrive at the sink.
+
+{% top %}
+
+## Windows
+
+Aggregating events (e.g., counts, sums) works differently on streams than in batch processing.
+For example, it is impossible to count all elements in a stream,
+because streams are in general infinite (unbounded). Instead, aggregates on streams (counts, sums, etc),
+are scoped by **windows**, such as *"count over the last 5 minutes"*, or *"sum of the last 100 elements"*.
+
+Windows can be *time driven* (example: every 30 seconds) or *data driven* (example: every 100 elements).
+One typically distinguishes different types of windows, such as *tumbling windows* (no overlap),
+*sliding windows* (with overlap), and *session windows* (punctuated by a gap of inactivity).
+
+<img src="{{ site.baseurl }}/fig/windows.svg" alt="Time- and Count Windows" class="offset" width="80%" />
+
+More window examples can be found in this [blog post](https://flink.apache.org/news/2015/12/04/Introducing-windows.html).
+
+{% top %}
+
+## Time
+
+When referring to time in a streaming program (for example to define windows), one can refer to different notions
+of time:
+
+ - **Event Time** is the time when an event was created. It is usually described by a timestamp in the events,
+ for example attached by the producing sensor, or the producing service. Flink accesses event timestamps
+ via [timestamp assigners]({{ site.baseurl }}/dev/event_timestamps_watermarks.html).
+
+ - **Ingestion time** is the time when an event enters the Flink dataflow at the source operator.
+
+ - **Processing Time** is the local time at each operator that performs a time-based operation.
+
+<img src="{{ site.baseurl }}/fig/event_ingestion_processing_time.svg" alt="Event Time, Ingestion Time, and Processing Time" class="offset" width="80%" />
+
+More details on how to handle time are in the [event time docs]({{ site.baseurl }}/dev/event_time.html).
+
+{% top %}
+
+## Stateful Operations
+
+While many operations in a dataflow simply look at one individual *event at a time* (for example an event parser),
+some operations remember information across multiple events (for example window operators).
+These operations are called **stateful**.
+
+The state of stateful operations is maintained in what can be thought of as an embedded key/value store.
+The state is partitioned and distributed strictly together with the streams that are read by the
+stateful operators. Hence, access to the key/value state is only possible on *keyed streams*, after a *keyBy()* function,
+and is restricted to the values associated with the current event's key. Aligning the keys of streams and state
+makes sure that all state updates are local operations, guaranteeing consistency without transaction overhead.
+This alignment also allows Flink to redistribute the state and adjust the stream partitioning transparently.
+
+<img src="{{ site.baseurl }}/fig/state_partitioning.svg" alt="State and Partitioning" class="offset" width="50%" />
+
+{% top %}
+
+## Checkpoints for Fault Tolerance
+
+Flink implements fault tolerance using a combination of **stream replay** and **checkpointing**. A
+checkpoint is related to a specific point in each of the input streams along with the corresponding state for each
+of the operators. A streaming dataflow can be resumed from a checkpoint while maintaining consistency *(exactly-once
+processing semantics)* by restoring the state of the operators and replaying the events from the
+point of the checkpoint.
+
+The checkpoint interval is a means of trading off the overhead of fault tolerance during execution with the recovery time (the number
+of events that need to be replayed).
+
+More details on checkpoints and fault tolerance are in the [fault tolerance docs]({{ site.baseurl }}/internals/stream_checkpointing.html).
+
+{% top %}
+
+## Batch on Streaming
+
+Flink executes batch programs as a special case of streaming programs, where the streams are bounded (finite number of elements).
+A *DataSet* is treated internally as a stream of data. The concepts above thus apply to batch programs in the
+same way as well as they apply to streaming programs, with minor exceptions:
+
+ - Programs in the DataSet API do not use checkpoints. Recovery happens by fully replaying the streams.
+ That is possible, because inputs are bounded. This pushes the cost more towards the recovery,
+ but makes the regular processing cheaper, because it avoids checkpoints.
+
+ - Stateful operations in the DataSet API use simplified in-memory/out-of-core data structures, rather than
+ key/value indexes.
+
+ - The DataSet API introduces special synchronized (superstep-based) iterations, which are only possible on
+ bounded streams. For details, check out the [iteration docs]({{ site.baseurl }}/dev/batch/iterations.html).
+
+{% top %}
+
+## Next Steps
+
+Continue with the basic concepts in Flink's [Distributed Runtime]({{ site.baseurl }}/concepts/runtime).
http://git-wip-us.apache.org/repos/asf/flink/blob/79d7e301/docs/concepts/runtime.md
----------------------------------------------------------------------
diff --git a/docs/concepts/runtime.md b/docs/concepts/runtime.md
new file mode 100644
index 0000000..016861a
--- /dev/null
+++ b/docs/concepts/runtime.md
@@ -0,0 +1,127 @@
+---
+title: Distributed Runtime Environment
+nav-pos: 2
+nav-title: Distributed Runtime
+nav-parent_id: concepts
+---
+<!--
+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}
+
+## Tasks and Operator Chains
+
+For distributed execution, Flink *chains* operator subtasks together into *tasks*. Each task is executed by one thread.
+Chaining operators together into tasks is a useful optimization: it reduces the overhead of thread-to-thread
+handover and buffering, and increases overall throughput while decreasing latency.
+The chaining behavior can be configured in the APIs.
+
+The sample dataflow in the figure below is executed with five subtasks, and hence with five parallel threads.
+
+<img src="{{ site.baseurl }}/fig/tasks_chains.svg" alt="Operator chaining into Tasks" class="offset" width="80%" />
+
+{% top %}
+
+## Job Managers, Task Managers, Clients
+
+The Flink runtime consists of two types of processes:
+
+ - The **JobManagers** (also called *masters*) coordinate the distributed execution. They schedule tasks, coordinate
+ checkpoints, coordinate recovery on failures, etc.
+
+ There is always at least one Job Manager. A high-availability setup will have multiple JobManagers, one of
+ which one is always the *leader*, and the others are *standby*.
+
+ - The **TaskManagers** (also called *workers*) execute the *tasks* (or more specifically, the subtasks) of a dataflow,
+ and buffer and exchange the data *streams*.
+
+ There must always be at least one TaskManager.
+
+The JobManagers and TaskManagers can be started in various ways: directly on the machines, in
+containers, or managed by resource frameworks like YARN. TaskManagers connect to JobManagers, announcing
+themselves as available, and are assigned work.
+
+The **client** is not part of the runtime and program execution, but is used to prepare and send a dataflow to the JobManager.
+After that, the client can disconnect, or stay connected to receive progress reports. The client runs either as part of the
+Java/Scala program that triggers the execution, or in the command line process `./bin/flink run ...`.
+
+<img src="{{ site.baseurl }}/fig/processes.svg" alt="The processes involved in executing a Flink dataflow" class="offset" width="80%" />
+
+{% top %}
+
+## Task Slots and Resources
+
+Each worker (TaskManager) is a *JVM process*, and may execute one or more subtasks in separate threads.
+To control how many tasks a worker accepts, a worker has so called **task slots** (at least one).
+
+Each *task slot* represents a fixed subset of resources of the TaskManager. A TaskManager with three slots, for example,
+will dedicate 1/3 of its managed memory to each slot. Slotting the resources means that a subtask will not
+compete with subtasks from other jobs for managed memory, but instead has a certain amount of reserved
+managed memory. Note that no CPU isolation happens here; currently slots only separate the managed memory of tasks.
+
+By adjusting the number of task slots, users can define how subtasks are isolated from each other.
+Having one slot per TaskManager means each task group runs in a separate JVM (which can be started in a
+separate container, for example). Having multiple slots
+means more subtasks share the same JVM. Tasks in the same JVM share TCP connections (via multiplexing) and
+heartbeat messages. They may also share data sets and data structures, thus reducing the per-task overhead.
+
+<img src="{{ site.baseurl }}/fig/tasks_slots.svg" alt="A TaskManager with Task Slots and Tasks" class="offset" width="80%" />
+
+By default, Flink allows subtasks to share slots even if they are subtasks of different tasks, so long as
+they are from the same job. The result is that one slot may hold an entire pipeline of the
+job. Allowing this *slot sharing* has two main benefits:
+
+ - A Flink cluster needs exactly as many task slots as the highest parallelism used in the job.
+ No need to calculate how many tasks (with varying parallelism) a program contains in total.
+
+ - It is easier to get better resource utilization. Without slot sharing, the non-intensive
+ *source/map()* subtasks would block as many resources as the resource intensive *window* subtasks.
+ With slot sharing, increasing the base parallelism in our example from two to six yields full utilization of the
+ slotted resources, while making sure that the heavy subtasks are fairly distributed among the TaskManagers.
+
+<img src="{{ site.baseurl }}/fig/slot_sharing.svg" alt="TaskManagers with shared Task Slots" class="offset" width="80%" />
+
+The APIs also include a *resource group* mechanism which can be used to prevent undesirable slot sharing.
+
+As a rule-of-thumb, a good default number of task slots would be the number of CPU cores.
+With hyper-threading, each slot then takes 2 or more hardware thread contexts.
+
+{% top %}
+
+## State Backends
+
+The exact data structures in which the key/values indexes are stored depends on the chosen **state backend**. One state backend
+stores data in an in-memory hash map, another state backend uses [RocksDB](http://rocksdb.org) as the key/value store.
+In addition to defining the data structure that holds the state, the state backends also implement the logic to
+take a point-in-time snapshot of the key/value state and store that snapshot as part of a checkpoint.
+
+<img src="{{ site.baseurl }}/fig/checkpoints.svg" alt="checkpoints and snapshots" class="offset" width="60%" />
+
+{% top %}
+
+## Savepoints
+
+Programs written in the Data Stream API can resume execution from a **savepoint**. Savepoints allow both updating your programs and your Flink cluster without losing any state.
+
+Savepoints are **manually triggered checkpoints**, which take a snapshot of the program and write it out to a state backend. They rely on the regular checkpointing mechanism for this. During execution programs are periodically snapshotted on the worker nodes and produce checkpoints. For recovery only the last completed checkpoint is needed and older checkpoints can be safely discarded as soon as a new one is completed.
+
+Savepoints are similar to these periodic checkpoints except that they are **triggered by the user** and **don't automatically expire** when newer checkpoints are completed.
+
+{% top %}
http://git-wip-us.apache.org/repos/asf/flink/blob/79d7e301/docs/dev/api_concepts.md
----------------------------------------------------------------------
diff --git a/docs/dev/api_concepts.md b/docs/dev/api_concepts.md
index 35e4d3a..aa064d0 100644
--- a/docs/dev/api_concepts.md
+++ b/docs/dev/api_concepts.md
@@ -1,7 +1,9 @@
---
title: "Basic API Concepts"
-nav-parent_id: apis
+nav-parent_id: dev
nav-pos: 1
+nav-show_overview: true
+nav-id: api-concepts
---
<!--
Licensed to the Apache Software Foundation (ASF) under one
@@ -24,15 +26,15 @@ under the License.
Flink programs are regular programs that implement transformations on distributed collections
(e.g., filtering, mapping, updating state, joining, grouping, defining windows, aggregating).
-Collections are initially created from sources (e.g., by reading files, kafka, or from local
+Collections are initially created from sources (e.g., by reading from files, kafka topics, or from local, in-memory
collections). Results are returned via sinks, which may for example write the data to
-(distributed) files, or to standard output (for example the command line terminal).
+(distributed) files, or to standard output (for example, the command line terminal).
Flink programs run in a variety of contexts, standalone, or embedded in other programs.
The execution can happen in a local JVM, or on clusters of many machines.
-Depending on the type of data sources, i.e. bounded or unbounded sources you would either
-write a batch program or a streaming program where the DataSet API is used for the former
-and the DataStream API is used for the latter. This guide will introduce the basic concepts
+Depending on the type of data sources, i.e. bounded or unbounded sources, you would either
+write a batch program or a streaming program where the DataSet API is used for batch
+and the DataStream API is used for streaming. This guide will introduce the basic concepts
that are common to both APIs but please see our
[Streaming Guide]({{ site.baseurl }}/dev/datastream_api.html) and
[Batch Guide]({{ site.baseurl }}/dev/batch/index.html) for concrete information about
@@ -45,129 +47,6 @@ in the `DataSet` API, just replace by `ExecutionEnvironment` and `DataSet`.
* This will be replaced by the TOC
{:toc}
-Linking with Flink
-------------------
-
-To write programs with Flink, you need to include the Flink library corresponding to
-your programming language in your project.
-
-The simplest way to do this is to use one of the quickstart scripts: either for
-[Java]({{ site.baseurl }}/quickstart/java_api_quickstart.html) or for [Scala]({{ site.baseurl }}/quickstart/scala_api_quickstart.html). They
-create a blank project from a template (a Maven Archetype), which sets up everything for you. To
-manually create the project, you can use the archetype and create a project by calling:
-
-<div class="codetabs" markdown="1">
-<div data-lang="java" markdown="1">
-{% highlight bash %}
-mvn archetype:generate \
- -DarchetypeGroupId=org.apache.flink \
- -DarchetypeArtifactId=flink-quickstart-java \
- -DarchetypeVersion={{site.version }}
-{% endhighlight %}
-</div>
-<div data-lang="scala" markdown="1">
-{% highlight bash %}
-mvn archetype:generate \
- -DarchetypeGroupId=org.apache.flink \
- -DarchetypeArtifactId=flink-quickstart-scala \
- -DarchetypeVersion={{site.version }}
-{% endhighlight %}
-</div>
-</div>
-
-The archetypes are working for stable releases and preview versions (`-SNAPSHOT`).
-
-If you want to add Flink to an existing Maven project, add the following entry to your
-*dependencies* section in the *pom.xml* file of your project:
-
-<div class="codetabs" markdown="1">
-<div data-lang="java" markdown="1">
-{% highlight xml %}
-<!-- Use this dependency if you are using the DataStream API -->
-<dependency>
- <groupId>org.apache.flink</groupId>
- <artifactId>flink-streaming-java{{ site.scala_version_suffix }}</artifactId>
- <version>{{site.version }}</version>
-</dependency>
-<!-- Use this dependency if you are using the DataSet API -->
-<dependency>
- <groupId>org.apache.flink</groupId>
- <artifactId>flink-java</artifactId>
- <version>{{site.version }}</version>
-</dependency>
-<dependency>
- <groupId>org.apache.flink</groupId>
- <artifactId>flink-clients{{ site.scala_version_suffix }}</artifactId>
- <version>{{site.version }}</version>
-</dependency>
-{% endhighlight %}
-</div>
-<div data-lang="scala" markdown="1">
-{% highlight xml %}
-<!-- Use this dependency if you are using the DataStream API -->
-<dependency>
- <groupId>org.apache.flink</groupId>
- <artifactId>flink-streaming-scala{{ site.scala_version_suffix }}</artifactId>
- <version>{{site.version }}</version>
-</dependency>
-<!-- Use this dependency if you are using the DataSet API -->
-<dependency>
- <groupId>org.apache.flink</groupId>
- <artifactId>flink-scala{{ site.scala_version_suffix }}</artifactId>
- <version>{{site.version }}</version>
-</dependency>
-<dependency>
- <groupId>org.apache.flink</groupId>
- <artifactId>flink-clients{{ site.scala_version_suffix }}</artifactId>
- <version>{{site.version }}</version>
-</dependency>
-{% endhighlight %}
-
-**Important:** When working with the Scala API you must have one of these two imports:
-{% highlight scala %}
-import org.apache.flink.api.scala._
-{% endhighlight %}
-
-or
-
-{% highlight scala %}
-import org.apache.flink.api.scala.createTypeInformation
-{% endhighlight %}
-
-The reason is that Flink analyzes the types that are used in a program and generates serializers
-and comparaters for them. By having either of those imports you enable an implicit conversion
-that creates the type information for Flink operations.
-</div>
-</div>
-
-#### Scala Dependency Versions
-
-Because Scala 2.10 binary is not compatible with Scala 2.11 binary, we provide multiple artifacts
-to support both Scala versions.
-
-Starting from the 0.10 line, we cross-build all Flink modules for both 2.10 and 2.11. If you want
-to run your program on Flink with Scala 2.11, you need to add a `_2.11` suffix to the `artifactId`
-values of the Flink modules in your dependencies section.
-
-If you are looking for building Flink with Scala 2.11, please check
-[build guide]({{ site.baseurl }}/setup/building.html#scala-versions).
-
-#### Hadoop Dependency Versions
-
-If you are using Flink together with Hadoop, the version of the dependency may vary depending on the
-version of Hadoop (or more specifically, HDFS) that you want to use Flink with. Please refer to the
-[downloads page](http://flink.apache.org/downloads.html) for a list of available versions, and instructions
-on how to link with custom versions of Hadoop.
-
-In order to link against the latest SNAPSHOT versions of the code, please follow
-[this guide](http://flink.apache.org/how-to-contribute.html#snapshots-nightly-builds).
-
-The *flink-clients* dependency is only necessary to invoke the Flink program locally (for example to
-run it standalone for testing and debugging). If you intend to only export the program as a JAR
-file and [run it on a cluster]({{ site.baseurl }}/dev/cluster_execution.html), you can skip that dependency.
-
-{% top %}
-
DataSet and DataStream
----------------------
@@ -933,124 +812,6 @@ interface can be implemented by input formats and functions to tell the API
explicitly about their return type. The *input types* that the functions are invoked with can
usually be inferred by the result types of the previous operations.
-Execution Configuration
------------------------
-
-The `StreamExecutionEnvironment` also contains the `ExecutionConfig` which allows to set job specific configuration values for the runtime.
-
-<div class="codetabs" markdown="1">
-<div data-lang="java" markdown="1">
-{% highlight java %}
-StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
-ExecutionConfig executionConfig = env.getConfig();
-{% endhighlight %}
-</div>
-<div data-lang="scala" markdown="1">
-{% highlight scala %}
-val env = StreamExecutionEnvironment.getExecutionEnvironment
-var executionConfig = env.getConfig
-{% endhighlight %}
-</div>
-</div>
-
-The following configuration options are available: (the default is bold)
-
-- **`enableClosureCleaner()`** / `disableClosureCleaner()`. The closure cleaner is enabled by default. The closure cleaner removes unneeded references to the surrounding class of anonymous functions inside Flink programs.
-With the closure cleaner disabled, it might happen that an anonymous user function is referencing the surrounding class, which is usually not Serializable. This will lead to exceptions by the serializer.
-
-- `getParallelism()` / `setParallelism(int parallelism)` Set the default parallelism for the job.
-
-- `getMaxParallelism()` / `setMaxParallelism(int parallelism)` Set the default maximum parallelism for the job. This setting determines the maximum degree of parallelism and specifies the upper limit for dynamic scaling.
-
-- `getNumberOfExecutionRetries()` / `setNumberOfExecutionRetries(int numberOfExecutionRetries)` Sets the number of times that failed tasks are re-executed. A value of zero effectively disables fault tolerance. A value of `-1` indicates that the system default value (as defined in the configuration) should be used. This is deprecated, use [restart strategies]({{ site.baseurl }}/setup/fault_tolerance.html#restart-strategies) instead.
-
-- `getExecutionRetryDelay()` / `setExecutionRetryDelay(long executionRetryDelay)` Sets the delay in milliseconds that the system waits after a job has failed, before re-executing it. The delay starts after all tasks have been successfully been stopped on the TaskManagers, and once the delay is past, the tasks are re-started. This parameter is useful to delay re-execution in order to let certain time-out related failures surface fully (like broken connections that have not fully timed out), before attempting a re-execution and immediately failing again due to the same problem. This parameter only has an effect if the number of execution re-tries is one or more. This is deprecated, use [restart strategies]({{ site.baseurl }}/setup/fault_tolerance.html#restart-strategies) instead.
-
-- `getExecutionMode()` / `setExecutionMode()`. The default execution mode is PIPELINED. Sets the execution mode to execute the program. The execution mode defines whether data exchanges are performed in a batch or on a pipelined manner.
-
-- `enableForceKryo()` / **`disableForceKryo`**. Kryo is not forced by default. Forces the GenericTypeInformation to use the Kryo serializer for POJOS even though we could analyze them as a POJO. In some cases this might be preferable. For example, when Flink's internal serializers fail to handle a POJO properly.
-
-- `enableForceAvro()` / **`disableForceAvro()`**. Avro is not forced by default. Forces the Flink AvroTypeInformation to use the Avro serializer instead of Kryo for serializing Avro POJOs.
-
-- `enableObjectReuse()` / **`disableObjectReuse()`** By default, objects are not reused in Flink. Enabling the object reuse mode will instruct the runtime to reuse user objects for better performance. Keep in mind that this can lead to bugs when the user-code function of an operation is not aware of this behavior.
-
-- **`enableSysoutLogging()`** / `disableSysoutLogging()` JobManager status updates are printed to `System.out` by default. This setting allows to disable this behavior.
-
-- `getGlobalJobParameters()` / `setGlobalJobParameters()` This method allows users to set custom objects as a global configuration for the job. Since the `ExecutionConfig` is accessible in all user defined functions, this is an easy method for making configuration globally available in a job.
-
-- `addDefaultKryoSerializer(Class<?> type, Serializer<?> serializer)` Register a Kryo serializer instance for the given `type`.
-
-- `addDefaultKryoSerializer(Class<?> type, Class<? extends Serializer<?>> serializerClass)` Register a Kryo serializer class for the given `type`.
-
-- `registerTypeWithKryoSerializer(Class<?> type, Serializer<?> serializer)` Register the given type with Kryo and specify a serializer for it. By registering a type with Kryo, the serialization of the type will be much more efficient.
-
-- `registerKryoType(Class<?> type)` If the type ends up being serialized with Kryo, then it will be registered at Kryo to make sure that only tags (integer IDs) are written. If a type is not registered with Kryo, its entire class-name will be serialized with every instance, leading to much higher I/O costs.
-
-- `registerPojoType(Class<?> type)` Registers the given type with the serialization stack. If the type is eventually serialized as a POJO, then the type is registered with the POJO serializer. If the type ends up being serialized with Kryo, then it will be registered at Kryo to make sure that only tags are written. If a type is not registered with Kryo, its entire class-name will be serialized with every instance, leading to much higher I/O costs.
-
-Note that types registered with `registerKryoType()` are not available to Flink's Kryo serializer instance.
-
-- `disableAutoTypeRegistration()` Automatic type registration is enabled by default. The automatic type registration is registering all types (including sub-types) used by usercode with Kryo and the POJO serializer.
-
-- `setTaskCancellationInterval(long interval)` Sets the the interval (in milliseconds) to wait between consecutive attempts to cancel a running task. When a task is canceled a new thread is created which periodically calls `interrupt()` on the task thread, if the task thread does not terminate within a certain time. This parameter refers to the time between consecutive calls to `interrupt()` and is set by default to **30000** milliseconds, or **30 seconds**.
-
-The `RuntimeContext` which is accessible in `Rich*` functions through the `getRuntimeContext()` method also allows to access the `ExecutionConfig` in all user defined functions.
-
-{% top %}
-
-Program Packaging and Distributed Execution
------------------------------------------
-
-As described earlier, Flink programs can be executed on
-clusters by using a `remote environment`. Alternatively, programs can be packaged into JAR Files
-(Java Archives) for execution. Packaging the program is a prerequisite to executing them through the
-[command line interface]({{ site.baseurl }}/setup/cli.html).
-
-#### Packaging Programs
-
-To support execution from a packaged JAR file via the command line or web interface, a program must
-use the environment obtained by `StreamExecutionEnvironment.getExecutionEnvironment()`. This environment
-will act as the cluster's environment when the JAR is submitted to the command line or web
-interface. If the Flink program is invoked differently than through these interfaces, the
-environment will act like a local environment.
-
-To package the program, simply export all involved classes as a JAR file. The JAR file's manifest
-must point to the class that contains the program's *entry point* (the class with the public
-`main` method). The simplest way to do this is by putting the *main-class* entry into the
-manifest (such as `main-class: org.apache.flinkexample.MyProgram`). The *main-class* attribute is
-the same one that is used by the Java Virtual Machine to find the main method when executing a JAR
-files through the command `java -jar pathToTheJarFile`. Most IDEs offer to include that attribute
-automatically when exporting JAR files.
-
-
-#### Packaging Programs through Plans
-
-Additionally, we support packaging programs as *Plans*. Instead of defining a progam in the main
-method and calling
-`execute()` on the environment, plan packaging returns the *Program Plan*, which is a description of
-the program's data flow. To do that, the program must implement the
-`org.apache.flink.api.common.Program` interface, defining the `getPlan(String...)` method. The
-strings passed to that method are the command line arguments. The program's plan can be created from
-the environment via the `ExecutionEnvironment#createProgramPlan()` method. When packaging the
-program's plan, the JAR manifest must point to the class implementing the
-`org.apache.flinkapi.common.Program` interface, instead of the class with the main method.
-
-
-#### Summary
-
-The overall procedure to invoke a packaged program is as follows:
-
-1. The JAR's manifest is searched for a *main-class* or *program-class* attribute. If both
-attributes are found, the *program-class* attribute takes precedence over the *main-class*
-attribute. Both the command line and the web interface support a parameter to pass the entry point
-class name manually for cases where the JAR manifest contains neither attribute.
-
-2. If the entry point class implements the `org.apache.flinkapi.common.Program`, then the system
-calls the `getPlan(String...)` method to obtain the program plan to execute.
-
-3. If the entry point class does not implement the `org.apache.flinkapi.common.Program` interface,
-the system will invoke the main method of the class.
-
{% top %}
Accumulators & Counters
@@ -1114,7 +875,7 @@ different operator functions of your job. Flink will internally merge all accumu
name.
A note on accumulators and iterations: Currently the result of accumulators is only available after
-the overall job ended. We plan to also make the result of the previous iteration available in the
+the overall job has ended. We plan to also make the result of the previous iteration available in the
next iteration. You can use
{% gh_link /flink-java/src/main/java/org/apache/flink/api/java/operators/IterativeDataSet.java#L98 "Aggregators" %}
to compute per-iteration statistics and base the termination of iterations on such statistics.
@@ -1135,217 +896,3 @@ result type ```R``` for the final result. E.g. for a histogram, ```V``` is a num
{% top %}
-Parallel Execution
-------------------
-
-This section describes how the parallel execution of programs can be configured in Flink. A Flink
-program consists of multiple tasks (transformations/operators, data sources, and sinks). A task is split into
-several parallel instances for execution and each parallel instance processes a subset of the task's
-input data. The number of parallel instances of a task is called its *parallelism*.
-
-
-The parallelism of a task can be specified in Flink on different levels.
-
-### Operator Level
-
-The parallelism of an individual operator, data source, or data sink can be defined by calling its
-`setParallelism()` method. For example, like this:
-
-<div class="codetabs" markdown="1">
-<div data-lang="java" markdown="1">
-{% highlight java %}
-final StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
-
-DataStream<String> text = [...]
-DataStream<Tuple2<String, Integer>> wordCounts = text
- .flatMap(new LineSplitter())
- .keyBy(0)
- .timeWindow(Time.seconds(5))
- .sum(1).setParallelism(5);
-
-wordCounts.print();
-
-env.execute("Word Count Example");
-{% endhighlight %}
-</div>
-<div data-lang="scala" markdown="1">
-{% highlight scala %}
-val env = StreamExecutionEnvironment.getExecutionEnvironment
-
-val text = [...]
-val wordCounts = text
- .flatMap{ _.split(" ") map { (_, 1) } }
- .keyBy(0)
- .timeWindow(Time.seconds(5))
- .sum(1).setParallelism(5)
-wordCounts.print()
-
-env.execute("Word Count Example")
-{% endhighlight %}
-</div>
-</div>
-
-### Execution Environment Level
-
-As mentioned [here](#anatomy-of-a-flink-program) Flink programs are executed in the context
-of an execution environment. An
-execution environment defines a default parallelism for all operators, data sources, and data sinks
-it executes. Execution environment parallelism can be overwritten by explicitly configuring the
-parallelism of an operator.
-
-The default parallelism of an execution environment can be specified by calling the
-`setParallelism()` method. To execute all operators, data sources, and data sinks with a parallelism
-of `3`, set the default parallelism of the execution environment as follows:
-
-<div class="codetabs" markdown="1">
-<div data-lang="java" markdown="1">
-{% highlight java %}
-final StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
-env.setParallelism(3);
-
-DataStream<String> text = [...]
-DataStream<Tuple2<String, Integer>> wordCounts = [...]
-wordCounts.print();
-
-env.execute("Word Count Example");
-{% endhighlight %}
-</div>
-<div data-lang="scala" markdown="1">
-{% highlight scala %}
-val env = StreamExecutionEnvironment.getExecutionEnvironment
-env.setParallelism(3)
-
-val text = [...]
-val wordCounts = text
- .flatMap{ _.split(" ") map { (_, 1) } }
- .keyBy(0)
- .timeWindow(Time.seconds(5))
- .sum(1)
-wordCounts.print()
-
-env.execute("Word Count Example")
-{% endhighlight %}
-</div>
-</div>
-
-### Client Level
-
-The parallelism can be set at the Client when submitting jobs to Flink. The
-Client can either be a Java or a Scala program. One example of such a Client is
-Flink's Command-line Interface (CLI).
-
-For the CLI client, the parallelism parameter can be specified with `-p`. For
-example:
-
- ./bin/flink run -p 10 ../examples/*WordCount-java*.jar
-
-
-In a Java/Scala program, the parallelism is set as follows:
-
-<div class="codetabs" markdown="1">
-<div data-lang="java" markdown="1">
-{% highlight java %}
-
-try {
- PackagedProgram program = new PackagedProgram(file, args);
- InetSocketAddress jobManagerAddress = RemoteExecutor.getInetFromHostport("localhost:6123");
- Configuration config = new Configuration();
-
- Client client = new Client(jobManagerAddress, config, program.getUserCodeClassLoader());
-
- // set the parallelism to 10 here
- client.run(program, 10, true);
-
-} catch (ProgramInvocationException e) {
- e.printStackTrace();
-}
-
-{% endhighlight %}
-</div>
-<div data-lang="scala" markdown="1">
-{% highlight scala %}
-try {
- PackagedProgram program = new PackagedProgram(file, args)
- InetSocketAddress jobManagerAddress = RemoteExecutor.getInetFromHostport("localhost:6123")
- Configuration config = new Configuration()
-
- Client client = new Client(jobManagerAddress, new Configuration(), program.getUserCodeClassLoader())
-
- // set the parallelism to 10 here
- client.run(program, 10, true)
-
-} catch {
- case e: Exception => e.printStackTrace
-}
-{% endhighlight %}
-</div>
-</div>
-
-
-### System Level
-
-A system-wide default parallelism for all execution environments can be defined by setting the
-`parallelism.default` property in `./conf/flink-conf.yaml`. See the
-[Configuration]({{ site.baseurl }}/setup/config.html) documentation for details.
-
-{% top %}
-
-Execution Plans
----------------
-
-Depending on various parameters such as data size or number of machines in the cluster, Flink's
-optimizer automatically chooses an execution strategy for your program. In many cases, it can be
-useful to know how exactly Flink will execute your program.
-
-__Plan Visualization Tool__
-
-Flink comes packaged with a visualization tool for execution plans. The HTML document containing
-the visualizer is located under ```tools/planVisualizer.html```. It takes a JSON representation of
-the job execution plan and visualizes it as a graph with complete annotations of execution
-strategies.
-
-The following code shows how to print the execution plan JSON from your program:
-
-<div class="codetabs" markdown="1">
-<div data-lang="java" markdown="1">
-{% highlight java %}
-final ExecutionEnvironment env = ExecutionEnvironment.getExecutionEnvironment();
-
-...
-
-System.out.println(env.getExecutionPlan());
-{% endhighlight %}
-</div>
-<div data-lang="scala" markdown="1">
-{% highlight scala %}
-val env = ExecutionEnvironment.getExecutionEnvironment
-
-...
-
-println(env.getExecutionPlan())
-{% endhighlight %}
-</div>
-</div>
-
-
-To visualize the execution plan, do the following:
-
-1. **Open** ```planVisualizer.html``` with your web browser,
-2. **Paste** the JSON string into the text field, and
-3. **Press** the draw button.
-
-After these steps, a detailed execution plan will be visualized.
-
-<img alt="A flink job execution graph." src="{{ site.baseurl }}/fig/plan_visualizer.png" width="80%">
-
-
-__Web Interface__
-
-Flink offers a web interface for submitting and executing jobs. The interface is part of the JobManager's
-web interface for monitoring, per default running on port 8081. Job submission via this interfaces requires
-that you have set `jobmanager.web.submit.enable: true` in `flink-conf.yaml`.
-
-You may specify program arguments before the job is executed. The plan visualization enables you to show
-the execution plan before executing the Flink job.
-
-{% top %}
http://git-wip-us.apache.org/repos/asf/flink/blob/79d7e301/docs/dev/apis.md
----------------------------------------------------------------------
diff --git a/docs/dev/apis.md b/docs/dev/apis.md
deleted file mode 100644
index 5e06e14..0000000
--- a/docs/dev/apis.md
+++ /dev/null
@@ -1,24 +0,0 @@
----
-title: "APIs"
-nav-id: apis
-nav-parent_id: dev
-nav-pos: 2
----
-<!--
-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.
--->
http://git-wip-us.apache.org/repos/asf/flink/blob/79d7e301/docs/dev/batch/examples.md
----------------------------------------------------------------------
diff --git a/docs/dev/batch/examples.md b/docs/dev/batch/examples.md
index 63d6c7a..7b6132c 100644
--- a/docs/dev/batch/examples.md
+++ b/docs/dev/batch/examples.md
@@ -1,8 +1,8 @@
---
-title: "Bundled Examples"
-nav-title: Examples
-nav-parent_id: batch
-nav-pos: 5
+title: "Batch Examples"
+nav-title: Batch Examples
+nav-parent_id: examples
+nav-pos: 20
---
<!--
Licensed to the Apache Software Foundation (ASF) under one
@@ -25,7 +25,7 @@ under the License.
The following example programs showcase different applications of Flink
from simple word counting to graph algorithms. The code samples illustrate the
-use of [Flink's API](index.html).
+use of [Flink's DataSet API](/dev/batch/index.html).
The full source code of the following and more examples can be found in the __flink-examples-batch__
or __flink-examples-streaming__ module of the Flink source repository.
@@ -36,7 +36,7 @@ or __flink-examples-streaming__ module of the Flink source repository.
## Running an example
-In order to run a Flink example, we assume you have a running Flink instance available. The "Setup" tab in the navigation describes various ways of starting Flink.
+In order to run a Flink example, we assume you have a running Flink instance available. The "Quickstart" and "Setup" tabs in the navigation describe various ways of starting Flink.
The easiest way is running the `./bin/start-local.sh` script, which will start a JobManager locally.
http://git-wip-us.apache.org/repos/asf/flink/blob/79d7e301/docs/dev/batch/index.md
----------------------------------------------------------------------
diff --git a/docs/dev/batch/index.md b/docs/dev/batch/index.md
index 0b1c9f9..48d60e1 100644
--- a/docs/dev/batch/index.md
+++ b/docs/dev/batch/index.md
@@ -2,8 +2,8 @@
title: "Flink DataSet API Programming Guide"
nav-id: batch
nav-title: Batch (DataSet API)
-nav-parent_id: apis
-nav-pos: 3
+nav-parent_id: dev
+nav-pos: 30
nav-show_overview: true
---
<!--
@@ -49,7 +49,7 @@ Example Program
The following program is a complete, working example of WordCount. You can copy & paste the code
to run it locally. You only have to include the correct Flink's library into your project
-(see Section [Linking with Flink]({{ site.baseurl }}/dev/api_concepts.html#linking-with-flink)) and specify the imports. Then you are ready
+(see Section [Linking with Flink]({{ site.baseurl }}/dev/linking_with_flink)) and specify the imports. Then you are ready
to go!
<div class="codetabs" markdown="1">
@@ -272,7 +272,7 @@ DataSet<Tuple3<Integer, String, Double>> output = input.sum(0).andMin(2);
Joins two data sets by creating all pairs of elements that are equal on their keys.
Optionally uses a JoinFunction to turn the pair of elements into a single element, or a
FlatJoinFunction to turn the pair of elements into arbitrarily many (including none)
- elements. See the <a href="#specifying-keys">keys section</a> to learn how to define join keys.
+ elements. See the <a href="/dev/api_concepts#specifying-keys">keys section</a> to learn how to define join keys.
{% highlight java %}
result = input1.join(input2)
.where(0) // key of the first input (tuple field 0)
@@ -298,7 +298,7 @@ result = input1.join(input2, JoinHint.BROADCAST_HASH_FIRST)
<tr>
<td><strong>OuterJoin</strong></td>
<td>
- Performs a left, right, or full outer join on two data sets. Outer joins are similar to regular (inner) joins and create all pairs of elements that are equal on their keys. In addition, records of the "outer" side (left, right, or both in case of full) are preserved if no matching key is found in the other side. Matching pairs of elements (or one element and a <code>null</code> value for the other input) are given to a JoinFunction to turn the pair of elements into a single element, or to a FlatJoinFunction to turn the pair of elements into arbitrarily many (including none) elements. See the <a href="#specifying-keys">keys section</a> to learn how to define join keys.
+ Performs a left, right, or full outer join on two data sets. Outer joins are similar to regular (inner) joins and create all pairs of elements that are equal on their keys. In addition, records of the "outer" side (left, right, or both in case of full) are preserved if no matching key is found in the other side. Matching pairs of elements (or one element and a <code>null</code> value for the other input) are given to a JoinFunction to turn the pair of elements into a single element, or to a FlatJoinFunction to turn the pair of elements into arbitrarily many (including none) elements. See the <a href="/dev/api_concepts#specifying-keys">keys section</a> to learn how to define join keys.
{% highlight java %}
input1.leftOuterJoin(input2) // rightOuterJoin or fullOuterJoin for right or full outer joins
.where(0) // key of the first input (tuple field 0)
@@ -320,7 +320,7 @@ input1.leftOuterJoin(input2) // rightOuterJoin or fullOuterJoin for right or ful
<td>
<p>The two-dimensional variant of the reduce operation. Groups each input on one or more
fields and then joins the groups. The transformation function is called per pair of groups.
- See the <a href="#specifying-keys">keys section</a> to learn how to define coGroup keys.</p>
+ See the <a href="/dev/api_concepts#specifying-keys">keys section</a> to learn how to define coGroup keys.</p>
{% highlight java %}
data1.coGroup(data2)
.where(0)
@@ -592,7 +592,7 @@ val output: DataSet[(Int, String, Double)] = input.sum(0).min(2)
Joins two data sets by creating all pairs of elements that are equal on their keys.
Optionally uses a JoinFunction to turn the pair of elements into a single element, or a
FlatJoinFunction to turn the pair of elements into arbitrarily many (including none)
- elements. See the <a href="#specifying-keys">keys section</a> to learn how to define join keys.
+ elements. See the <a href="/dev/api_concepts#specifying-keys">keys section</a> to learn how to define join keys.
{% highlight scala %}
// In this case tuple fields are used as keys. "0" is the join field on the first tuple
// "1" is the join field on the second tuple.
@@ -618,7 +618,7 @@ val result = input1.join(input2, JoinHint.BROADCAST_HASH_FIRST)
<tr>
<td><strong>OuterJoin</strong></td>
<td>
- Performs a left, right, or full outer join on two data sets. Outer joins are similar to regular (inner) joins and create all pairs of elements that are equal on their keys. In addition, records of the "outer" side (left, right, or both in case of full) are preserved if no matching key is found in the other side. Matching pairs of elements (or one element and a `null` value for the other input) are given to a JoinFunction to turn the pair of elements into a single element, or to a FlatJoinFunction to turn the pair of elements into arbitrarily many (including none) elements. See the <a href="#specifying-keys">keys section</a> to learn how to define join keys.
+ Performs a left, right, or full outer join on two data sets. Outer joins are similar to regular (inner) joins and create all pairs of elements that are equal on their keys. In addition, records of the "outer" side (left, right, or both in case of full) are preserved if no matching key is found in the other side. Matching pairs of elements (or one element and a `null` value for the other input) are given to a JoinFunction to turn the pair of elements into a single element, or to a FlatJoinFunction to turn the pair of elements into arbitrarily many (including none) elements. See the <a href="/dev/api_concepts#specifying-keys">keys section</a> to learn how to define join keys.
{% highlight scala %}
val joined = left.leftOuterJoin(right).where(0).equalTo(1) {
(left, right) =>
@@ -634,7 +634,7 @@ val joined = left.leftOuterJoin(right).where(0).equalTo(1) {
<td>
<p>The two-dimensional variant of the reduce operation. Groups each input on one or more
fields and then joins the groups. The transformation function is called per pair of groups.
- See the <a href="#specifying-keys">keys section</a> to learn how to define coGroup keys.</p>
+ See the <a href="/dev/api_concepts#specifying-keys">keys section</a> to learn how to define coGroup keys.</p>
{% highlight scala %}
data1.coGroup(data2).where(0).equalTo(1)
{% endhighlight %}
@@ -780,7 +780,7 @@ is not supported by the API out-of-the-box. To use this feature, you should use
</div>
</div>
-The [parallelism]({{ site.baseurl }}/dev/api_concepts.html#parallel-execution) of a transformation can be defined by `setParallelism(int)` while
+The [parallelism]({{ site.baseurl }}/dev/parallel) of a transformation can be defined by `setParallelism(int)` while
`name(String)` assigns a custom name to a transformation which is helpful for debugging. The same is
possible for [Data Sources](#data-sources) and [Data Sinks](#data-sinks).
@@ -1990,7 +1990,7 @@ data.map(new RichMapFunction<String, String>() {
Make sure that the names (`broadcastSetName` in the previous example) match when registering and
accessing broadcasted data sets. For a complete example program, have a look at
-{% gh_link /flink-examples/flink-examples-batch/src/main/java/org/apache/flink/examples/java/clustering/KMeans.java#L96 "K-Means Algorithm" %}.
+{% gh_link /flink-examples/flink-examples-batch/src/main/java/org/apache/flink/examples/java/clustering/KMeans.java "K-Means Algorithm" %}.
</div>
<div data-lang="scala" markdown="1">
@@ -2016,7 +2016,7 @@ data.map(new RichMapFunction[String, String]() {
Make sure that the names (`broadcastSetName` in the previous example) match when registering and
accessing broadcasted data sets. For a complete example program, have a look at
-{% gh_link /flink-examples/flink-examples-batch/src/main/scala/org/apache/flink/examples/scala/clustering/KMeans.scala#L96 "KMeans Algorithm" %}.
+{% gh_link /flink-examples/flink-examples-batch/src/main/scala/org/apache/flink/examples/scala/clustering/KMeans.scala "KMeans Algorithm" %}.
</div>
</div>
@@ -2103,7 +2103,7 @@ val result: DataSet[Integer] = input.map(new MyMapper())
env.execute()
{% endhighlight %}
-Access the cached file in a user function (here a `MapFunction`). The function must extend a [RichFunction]({{ site.baseurl }}/apis/common/index.html#rich-functions) class because it needs access to the `RuntimeContext`.
+Access the cached file in a user function (here a `MapFunction`). The function must extend a [RichFunction]({{ site.baseurl }}/dev/api_concepts#rich-functions) class because it needs access to the `RuntimeContext`.
{% highlight scala %}
@@ -2258,7 +2258,7 @@ Please note that you can also pass a custom class extending the `ExecutionConfig
**Accessing values from the global configuration**
-Objects in the global job parameters are accessible in many places in the system. All user functions implementing a `Rich*Function` interface have access through the runtime context.
+Objects in the global job parameters are accessible in many places in the system. All user functions implementing a `RichFunction` interface have access through the runtime context.
{% highlight java %}
public static final class Tokenizer extends RichFlatMapFunction<String, Tuple2<String, Integer>> {
http://git-wip-us.apache.org/repos/asf/flink/blob/79d7e301/docs/dev/batch/iterations.md
----------------------------------------------------------------------
diff --git a/docs/dev/batch/iterations.md b/docs/dev/batch/iterations.md
index cfeffe6..73a1d57 100644
--- a/docs/dev/batch/iterations.md
+++ b/docs/dev/batch/iterations.md
@@ -1,9 +1,7 @@
---
-title: "Iterations"
-
-# Sub-level navigation
-sub-nav-group: batch
-sub-nav-pos: 3
+title: Iterations
+nav-parent_id: batch
+nav-pos: 2
---
<!--
Licensed to the Apache Software Foundation (ASF) under one
http://git-wip-us.apache.org/repos/asf/flink/blob/79d7e301/docs/dev/cluster_execution.md
----------------------------------------------------------------------
diff --git a/docs/dev/cluster_execution.md b/docs/dev/cluster_execution.md
index 31b4d4a..d614846 100644
--- a/docs/dev/cluster_execution.md
+++ b/docs/dev/cluster_execution.md
@@ -1,6 +1,6 @@
---
title: "Cluster Execution"
-nav-parent_id: dev
+nav-parent_id: batch
nav-pos: 12
---
<!--
@@ -81,75 +81,3 @@ public static void main(String[] args) throws Exception {
Note that the program contains custom user code and hence requires a JAR file with
the classes of the code attached. The constructor of the remote environment
takes the path(s) to the JAR file(s).
-
-## Linking with modules not contained in the binary distribution
-
-The binary distribution contains jar packages in the `lib` folder that are automatically
-provided to the classpath of your distributed programs. Almost all of Flink classes are
-located there with a few exceptions, for example the streaming connectors and some freshly
-added modules. To run code depending on these modules you need to make them accessible
-during runtime, for which we suggest two options:
-
-1. Either copy the required jar files to the `lib` folder onto all of your TaskManagers.
-Note that you have to restart your TaskManagers after this.
-2. Or package them with your code.
-
-The latter version is recommended as it respects the classloader management in Flink.
-
-### Packaging dependencies with your usercode with Maven
-
-To provide these dependencies not included by Flink we suggest two options with Maven.
-
-1. The maven assembly plugin builds a so-called uber-jar (executable jar) containing all your dependencies.
-The assembly configuration is straight-forward, but the resulting jar might become bulky.
-See [maven-assembly-plugin](http://maven.apache.org/plugins/maven-assembly-plugin/usage.html) for further information.
-2. The maven unpack plugin unpacks the relevant parts of the dependencies and
-then packages it with your code.
-
-Using the latter approach in order to bundle the Kafka connector, `flink-connector-kafka`
-you would need to add the classes from both the connector and the Kafka API itself. Add
-the following to your plugins section.
-
-~~~xml
-<plugin>
- <groupId>org.apache.maven.plugins</groupId>
- <artifactId>maven-dependency-plugin</artifactId>
- <version>2.9</version>
- <executions>
- <execution>
- <id>unpack</id>
- <!-- executed just before the package phase -->
- <phase>prepare-package</phase>
- <goals>
- <goal>unpack</goal>
- </goals>
- <configuration>
- <artifactItems>
- <!-- For Flink connector classes -->
- <artifactItem>
- <groupId>org.apache.flink</groupId>
- <artifactId>flink-connector-kafka</artifactId>
- <version>{{ site.version }}</version>
- <type>jar</type>
- <overWrite>false</overWrite>
- <outputDirectory>${project.build.directory}/classes</outputDirectory>
- <includes>org/apache/flink/**</includes>
- </artifactItem>
- <!-- For Kafka API classes -->
- <artifactItem>
- <groupId>org.apache.kafka</groupId>
- <artifactId>kafka_<YOUR_SCALA_VERSION></artifactId>
- <version><YOUR_KAFKA_VERSION></version>
- <type>jar</type>
- <overWrite>false</overWrite>
- <outputDirectory>${project.build.directory}/classes</outputDirectory>
- <includes>kafka/**</includes>
- </artifactItem>
- </artifactItems>
- </configuration>
- </execution>
- </executions>
-</plugin>
-~~~
-
-Now when running `mvn clean package` the produced jar includes the required dependencies.
http://git-wip-us.apache.org/repos/asf/flink/blob/79d7e301/docs/dev/connectors/cassandra.md
----------------------------------------------------------------------
diff --git a/docs/dev/connectors/cassandra.md b/docs/dev/connectors/cassandra.md
index 90be0e3..19d483b 100644
--- a/docs/dev/connectors/cassandra.md
+++ b/docs/dev/connectors/cassandra.md
@@ -35,7 +35,7 @@ To use this connector, add the following dependency to your project:
</dependency>
{% endhighlight %}
-Note that the streaming connectors are currently not part of the binary distribution. See how to link with them for cluster execution [here]({{ site.baseurl}}/dev/cluster_execution.html#linking-with-modules-not-contained-in-the-binary-distribution).
+Note that the streaming connectors are currently not part of the binary distribution. See how to link with them for cluster execution [here]({{ site.baseurl}}/dev/linking).
#### Installing Apache Cassandra
Follow the instructions from the [Cassandra Getting Started page](http://wiki.apache.org/cassandra/GettingStarted).
http://git-wip-us.apache.org/repos/asf/flink/blob/79d7e301/docs/dev/connectors/elasticsearch.md
----------------------------------------------------------------------
diff --git a/docs/dev/connectors/elasticsearch.md b/docs/dev/connectors/elasticsearch.md
index be45f98..1907740 100644
--- a/docs/dev/connectors/elasticsearch.md
+++ b/docs/dev/connectors/elasticsearch.md
@@ -37,7 +37,7 @@ following dependency to your project:
Note that the streaming connectors are currently not part of the binary
distribution. See
-[here]({{site.baseurl}}/dev/cluster_execution.html#linking-with-modules-not-contained-in-the-binary-distribution)
+[here]({{site.baseurl}}/dev/linking)
for information about how to package the program with the libraries for
cluster execution.
http://git-wip-us.apache.org/repos/asf/flink/blob/79d7e301/docs/dev/connectors/elasticsearch2.md
----------------------------------------------------------------------
diff --git a/docs/dev/connectors/elasticsearch2.md b/docs/dev/connectors/elasticsearch2.md
index 5f4267e..a796280 100644
--- a/docs/dev/connectors/elasticsearch2.md
+++ b/docs/dev/connectors/elasticsearch2.md
@@ -37,7 +37,7 @@ following dependency to your project:
Note that the streaming connectors are currently not part of the binary
distribution. See
-[here]({{site.baseurl}}/dev/cluster_execution.html#linking-with-modules-not-contained-in-the-binary-distribution)
+[here]({{site.baseurl}}/dev/linking)
for information about how to package the program with the libraries for
cluster execution.
@@ -145,7 +145,7 @@ More information about Elasticsearch can be found [here](https://elastic.co).
For the execution of your Flink program,
it is recommended to build a so-called uber-jar (executable jar) containing all your dependencies
-(see [here]({{site.baseurl}}/dev/cluster_execution.html#linking-with-modules-not-contained-in-the-binary-distribution) for further information).
+(see [here]({{site.baseurl}}/dev/linking) for further information).
However,
when an uber-jar containing an Elasticsearch sink is executed,
http://git-wip-us.apache.org/repos/asf/flink/blob/79d7e301/docs/dev/connectors/filesystem_sink.md
----------------------------------------------------------------------
diff --git a/docs/dev/connectors/filesystem_sink.md b/docs/dev/connectors/filesystem_sink.md
index 79cfe6a..030e9d9 100644
--- a/docs/dev/connectors/filesystem_sink.md
+++ b/docs/dev/connectors/filesystem_sink.md
@@ -37,7 +37,7 @@ following dependency to your project:
Note that the streaming connectors are currently not part of the binary
distribution. See
-[here]({{site.baseurl}}/dev/cluster_execution.html#linking-with-modules-not-contained-in-the-binary-distribution)
+[here]({{site.baseurl}}/dev/linking)
for information about how to package the program with the libraries for
cluster execution.