Cassandra Migrations

Cassandra Migrations is a Cassandra database schema migration library for Rails applications.

This gem provides:

• Multi-environment database configuration
• Versioned CQL schema migration management
• A schema modification DSL for simplified migration code
• Rake tasks for database schema management
• Support for forked processes

Use Cassandra in an organized and familiar way, without changing how you work with your ActiveRecord relational database schema.

• Cassandra >= 1.2, using the native transport protocol
• Ruby >= 1.9
• Rails >= 3.2

gem install cassandra_migrations

or, with bundler, add the following to your gemfile:

gem 'cassandra_migrations'

Similar to how config/database.yml stores your relational databse configuration, config/cassandra.yml stores your cassandra database configuration.

rails g cassandra_configuration

This will create the config/cassandra.yml with default settings. Configure the database names for each of your environments.

development:
  hosts:
    - 127.0.0.1
  port: 9042
  keyspace: my_keyspace_dev
  replication:
    class: SimpleStrategy
    replication_factor: 1

These are the minimum options to get started, for advanced configuration, read about [Database Configuration Options](wiki/Database Configuration Options).

Assuming your Cassandra database is running, you may now create your keyspace:

rake cassandra:create

Similar to how db/migrate/ stores your relational databse schema migration files, db/cassandra_migrate/ stores your Cassandra schema migration files.

rails g cassandra_migration create_posts

This will create a versioned migration in db/cassandra_migrate/, with familiar up and down methods that will be executed when applying or rolling back a migration, respectively.

class CreatePosts < CassandraMigrations::Migration
  def up
  end

  def down
  end
end

You may call execute from these methods to execute CQL statements to your configured keyspace.

For increased readability and ease of use, there is also a familiar DSL available from the migration methods.

class CreatePosts < CassandraMigrations::Migration
  def up
    create_table :posts,
                 partition_keys: [:id, :created_month],
                 primary_keys:   [:created_at] do |t|
      t.integer   :id
      t.string    :created_month
      t.timestamp :created_at
      t.string    :title
      t.string    :category
      t.set       :tags, type: :float
      t.map       :my_map, key_type: :uuid, value_type: :float
      t.text      :content
    end

    create_index :posts, :category, name: 'posts_by_category'
  end

  def down
    drop_index 'posts_by_category'
    drop_table :posts
  end
end

Use the following type methods for fields:

• text
• integer
• decimal
• float
• double
• boolean
• uuid
• timeuuid
• inet
• timestamp
• datetime
• binary
• list (with type option)
• set (with type option)
• map (with key_type and value_type options)

For more details on the DSL's types, advanced table options, keyspace manipulation, or other schema statment methods, read about the [Migration DSL](wiki/Migration DSL).

There are a collection of familiar rake tasks to help you manage your cassandra databases.

• rake cassandra:create Creates the configured keyspace in config/cassandra.yml.
• rake cassandra:drop Drops the configured keyspace in config/cassandra.yml.
• rake cassandra:migrate Runs migrations that have not run yet.
• rake cassandra:rollback Rolls back the latest migration that has been applied.
• rake cassandra:migrate:reset Runs cassandra:drop, cassandra:create and cassandra:migrate.

Each rake task will be run against the database that is configured for the current environment (via RAILS_ENV).

rake cassandra:migrate

== CreatePosts: migrating =====================================================
  create_table(posts)
  -> CREATE TABLE posts (id int, created_month varchar, created_at timestamp, title varchar, category varchar, content text, PRIMARY KEY((id, created_month), created_at))
  create_index(posts)
  -> CREATE INDEX posts_by_category ON posts (category)
== CreatePosts: migrated (0.3448s) ============================================

Migrated 1 version(s) up.

For more details on available rake tasks and options, read about [Rake Tasks](wiki/Rake Tasks).

When a migration requires working with data, not just schema, one option is using the Cassandra Migration query classes.

CassandraMigrations::Cassandra.select(:posts)

new posts = CassandraMigrations::Cassandra.select(:posts,
  projection: 'title, created_at',
  selection: 'id > 1234',
  order_by: 'created_at DESC',
  limit: 10
)

new_posts.each do |post|
  CassandraMigrations::Cassandra.update!(:posts,
    "id = #{post['id']}",
    {tags: ['new']},
    {operations: {tags: :+}})
end

For more details, options, and examples, read about the [Query Helpers](wiki/Query Helpers).

Cassandra Migrations has built-in support for Passenger's forked process model (e.g. smart spawning).

Each fork should contain its own session to avoid deadlock and other issues. Restart the connection after the process has forked.

# config/puma.rb
on_worker_boot do
  # re-establish the connection in this process fork
  CassandraMigrations::Cassandra.restart
end

To add cassandra database creation and migrations steps to your Capistrano recipe, add the following line to you deploy.rb: require 'cassandra_migrations/capistrano'

To run the test suite:

bundle install

rake

Additionally, this project aims for compatiblity with Rails 3.2 and above. For that, we use the appraisal gem. If you make changes, run our test suite against all supported version of Rails like so:

appraisal install

appraisal rake

• Henrique Gubert - @hsgubert
• Brian Sam-Bodden - @bsbodden
• Sid Tantia - @sstgithub
• and more...

This gem is built upon the official Ruby Driver for Apache Cassandra by DataStax. Which supersedes the cql-rb gem (thank you Theo for doing an awesome job).