Octopus will enter into maintainance mode once Rails 6 is released - since Rails 6 will include all the features for Database Sharding / Replication into Rails Core (PR: rails/rails#34052). Once the first version of Rails 6 beta is released, there will be a migration guide to help users migrate from Octopus to Rails 6.
Octopus is a better way to do Database Sharding in ActiveRecord. Sharding allows multiple databases in the same rails application. While there are several projects that implement Sharding (e.g. DbCharmer, DataFabric, MultiDb), each project has its own limitations. The main goal of octopus project is to provide a better way of doing Database Sharding.
The api is designed to be simple as possible. Octopus focuses on the end user, giving the power of multiple databases but with reliable code and flexibility. Octopus is compatible with Rails 4 and Rails 5.
Octopus supports:
- Sharding (with multiple shards, and grouped shards).
- Replication (Master/slave support, with multiple slaves).
- Moving data between shards with migrations.
- Tools to manage database configurations. (soon)
When using replication, all writes queries will be sent to master, and read queries to slaves. More info could be found at: Wiki
When using sharding, you need to specify which shard to send the query. Octopus supports selecting the shard inside a controller, or manually in each object. More could be found at Wiki
When using replication and sharding concurrently, you must specify a shard, and can optionally specify a slave group. All write queries will be sent to each shard's master. If the slave group is specified read queries will be sent to slaves in it, or else to shard's master. More info could be found at Wiki
Add this line to Gemfile:
gem 'ar-octopus'
Currently, Octopus doesn't support Rails 2. If you need support for rails 2, please use the version 0.5.0.
Octopus < 0.5.0 stored schema version information in the master database defined in the database.yml file, and assumed that each shard's schema matched the others and the master database. Beginning with Octopus 0.5.0, the schema version information for each shard is stored within that shard's database.
If you are upgrading from < 0.5.0 run the copy_schema_versions
rake task to copy the schema version information in the
master database to each of the shards:
rake octopus:copy_schema_versions
Once the task completes migrations will operate normally and schema information will be stored in each shard database going forward.
First, you need to create a config file, shards.yml, inside your config/ directory. to see the syntax and how this file should look, please checkout this page on wiki.
Octopus adds a method to each AR Class and object: the using method is used to select the shard like this:
User.where(:name => "Boba").limit(3).using(:read_replica_one)
Octopus also supports queries within a block. When you pass a block to the using method, all queries inside the block will be sent to the specified shard.
Octopus.using(:read_replica_two) do
User.create(:name => "Thiago")
end
If you want to use the same code for all shards or all shards in a specific group (for example in db/seeds.rb
), you can use this syntax.
# This will return a list of the given block's results, per shard.
Octopus.using_all do
User.create_from_csv!
end
# This will return a list of the given block's results, per shard in history_shards group.
Octopus.using_group(:history_shards) do
HistoryCategory.create_from_csv!
end
Each model instance knows which shard it came from so this will work automatically:
# This will find the user in the shard1
@user = User.using(:shard1).find_by_name("Joao")
# This will find the user in the master database
@user2 = User.find_by_name("Jose")
#Sets the name
@user.name = "Mike"
# Save the user in the correct shard, shard1.
@user.save
In migrations, you also have access to the using method. The syntax is basically the same. This migration will run in the brazil and canada shards.
class CreateUsersOnBothShards < ActiveRecord::Migration
using(:brazil, :canada)
def self.up
User.create!(:name => "Both")
end
def self.down
User.delete_all
end
end
You also could send a migration to a group of shards. This migration will be sent to all shards that belongs to history_shards group, specified in shards.yml:
class CreateUsersOnMultiplesGroups < ActiveRecord::Migration
using_group(:history_shards)
def self.up
User.create!(:name => "MultipleGroup")
end
def self.down
User.delete_all
end
end
You can specify a default_migration_group
for migrations, so that modifications to each individual migration file are not needed:
octopus:
default_migration_group: europe_databases
There is no need for a corresponding default_migration_shard
- simply define that database to be your master. You might want this setting if all of your databases have identical schemas, but are not replicated.
You can configure a master shard for the rails application, to connect to when rails is going up. The safest would be to configure this to the shard specified in database.yml
(some things still use it).
octopus:
master_shard: <%= ENV['SHARD'] || 'shard1' %>
Then you can use the SHARD
environment variable to override the master_shard
specified in config/shards.yml
, useful for running rake tasks.
SHARD=shard1 rake db:setup && SHARD=shard2 rake db:setup
If you want to send a specified action, or all actions from a controller, to a specific shard, use this syntax:
class ApplicationController < ActionController::Base
around_filter :select_shard
def select_shard(&block)
Octopus.using(:brazil, &block)
end
end
To see the complete list of features and syntax, please check out our Wiki Want to see sample rails applications using octopus features? please check it out: Sharding Example and Replication Example. Also, we have an example that shows how to use Octopus without Rails: Octopus + Sinatra Example.
If you want to set a custom connection to a specific model, use the syntax octopus_establish_connection
syntax:
#This class sets its own connection
class CustomConnection < ActiveRecord::Base
octopus_establish_connection(:adapter => "mysql", :database => "octopus_shard2")
end
If you'd like to use specific shards with a model that has a Rails-managed connection, you can use allow_shard
:
class CustomConnectedModel
octopus_establish_connection(...)
allow_shard :my_shard
end
#This uses :my_shard
CustomConnectedModel.using(:my_shard).first
#This uses the Rails-managed connection pool (the call to 'using' is ignored)
CustomConnectedModel.using(:some_other_shard).first
This can be useful if you have a model that lives in a separate database and would like to add sharding or replication to it. For other use cases, you may be better off with slave groups.
Contributors are welcome! To run the test suite, you need mysql, postgresql and sqlite3 installed. This is what you need to setup your Octopus development environment:
git clone http://github.com/thiagopradi/octopus.git
cd octopus
bundle install
bundle exec rake db:prepare
bundle exec rake appraisal:install
bundle exec rake spec
This command will run the spec suite for all rails versions supported. To run our integrations tests inside sample_app, you need to following commands:
cd sample_app
bundle install
cucumber
For a more consistent build/test/development environment, use Docker Compose.
[me@host octopus]$ docker-compose -f docker-compose.yml up
Creating network "octopus_default" with the default driver
Creating octopus_postgres_1 ...
Creating octopus_mysql_1 ...
Creating octopus_postgres_1
. . .
octopus_1 | mysqld is alive
octopus_1 | MySQL is ready
octopus_1 | + bundle exec rake db:prepare
. . .
octopus_1 | + bundle exec rake spec
. . .
octopus_1 | Finished in 7 minutes 44 seconds (files took 0.71416 seconds to load)
octopus_1 | 359 examples, 0 failures, 2 pending
(Ctrl+C to shut down the stack)
Your workstation's octopus clone is mounted inside the docker container, so
you may still use your favorite development tools. The octopus_1
container
is available for re-running the test suite. To start a new shell in another
terminal, run:
[me@host octopus]$ docker-compose exec octopus /bin/bash
root@f6ae68df6fc8:/usr/src/app#
# run bundle exec rake spec again
To completely destory the stack and start over:
[me@host octopus]$ docker-compose -f docker-compose.yml down
[me@host octopus]$ docker-compose -f docker-compose.yml up
If you are having issues running the octopus spec suite, verify your database users and passwords match those inside the config files and your permissions are correct.
This project is sponsored by the Ruby Summer of Code, Rapid River Software, and my mentors Mike Perham and Amit Agarwal.
Copyright (c) Thiago Pradi, released under the MIT license.