Spark Platform

Description

Apache Spark™ is a fast and general engine for large-scale data processing.

This cookbook is designed to install and configure Spark.

Requirements

Cookbooks and gems

Declared in [metadata.rb](metadata.rb) and in [Gemfile](Gemfile).

Platforms

A systemd managed distribution: - RHEL Family 7, tested on Centos

Usage

First Setup

The recommended way to use this cookbook is through the creation of a role per spark cluster. This enables the search by role feature, allowing a simple service discovery.

In fact, there are two ways to configure the search: 1. with a static configuration through a list of hostnames (attributes hosts that is ['spark-platform']['hosts']) 2. with a real search, performed on a role (attributes role and size like in ['spark-platform']['role']). The role should be in the run-list of all nodes of the cluster. The size is a safety and should be the number of nodes in the cluster.

If hosts is configured, role and size are ignored

See [roles](test/integration/roles) for some examples and [Cluster Search][cluster-search] documentation for more information.

Master High Availability

To install properly a HA spark cluster, set ['spark-platform']['master_ha_with_zk'] to true. Then set the number of masters you want with ['spark-platform']['n_of_masters'].

You also need a Zookeeper cluster. This is not in the scope of this cookbook but if you need one, you should consider using [Zookeeper Platform][zookeeper-platform].

The configuration of Zookeeper hosts use search and is done similarly as for spark hosts, ie with a static list of hostnames or by using a search on a role. The attribute to configure is ['spark-platform']['zookeeper'].

Java

By default, this cookbook installs openjdk from the official repositories (openjdk-headless 8 on centos 7) just before starting the service. You can change this behavior by setting node['spark-platform']['java'] to "", or choose your package by setting the package name in node['spark-platform']['java'][node[:platform]].

Test

This cookbook is fully tested through the installation of the full platform in docker hosts. This uses kitchen, docker and some monkey-patching.

If you run kitchen list, you will see 4 suites:

• zookeeper-centos-7
• spark-platform-1-centos-7
• spark-platform-2-centos-7
• spark-platform-3-centos-7

Each corresponds to a different node in the cluster. They are connected through a bridge network named kitchen, which is created if necessary.

For more information, see [.kitchen.yml](.kitchen.yml) and [test](test) directory.

Local cluster

The cluster installed with kitchen converge is fully working and can thus be used as a local cluster for developments tests.

You can access it by using internal DNS of the docker network named kitchen or by declaring each node in your hosts file. You can get each IP by running:

docker inspect --format \
  '{{.NetworkSettings.Networks.kitchen.IPAddress}}' container_name

Spark group

In standalone mode, it is not possible to execute spark jobs with different users. This means that the result files will always be owned by spark user. To be able to use different user accounts, a solution is to add all users to spark group and make the files group modifiable. In this cookbook, the spark group is created with "append true" option so other cookbooks can modify it to add users.

Attributes

Configuration is done by overriding default attributes. All configuration keys have a default defined in [attributes/default.rb](attributes/default.rb). Please read it to have a comprehensive view of what and how you can configure this cookbook behavior.

Recipes

default

Include other recipes, that is:

1. user
2. install
3. config
4. systemd
5. service

user

Create user and group for Apache Spark.

install

Install Apache Spark using a tarball package.

config

Configure Apache Spark, searching for other cluster members if available.

systemd

Create systemd unit files

service

Enable and start services

Resources/Providers

None.

Changelog

Available in [CHANGELOG.md](CHANGELOG.md).

Contributing

Please read carefully [CONTRIBUTING.md](CONTRIBUTING.md) before making a merge request.

License and Author

• Author:: Samuel Bernard (samuel.bernard@s4m.io)
• Author:: Florian Philippon (florian.philippon@s4m.io)

Copyright (c) 2016 Sam4Mobile

Licensed 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.