NOTE: The documentation below refers to an unreleased version of kuby-redis. Please click here for the documentation for the currently released version.

Redis plugin for Kuby. If you're looking for an easy way to add Sidekiq to your app, check out kuby-sidekiq.

The redis plugin provides the ability to stand up arbitrary redis instances. Behind the scenes it uses the excellent Spotahome Redis operator.

Add the kuby-redis gem to your Gemfile and run the setup command, eg:

bundle exec kuby -e production setup

Add a Redis instance like this:

require 'kuby/redis'

Kuby.define(:production) do
  kubernetes do

    add_plugin(:redis) do
      instance(:my_rails_cache)
    end

  end
end

The kuby-redis plugin supports a number of additional configuration options:

• cpu_request, (default: '100m'). The number of CPU units to request from Kubernetes. 1 CPU core = 1000 units.
• memory_request, (default: '100Mi'). The amount of memory to request from Kubernetes.
• cpu_limit, (default: '400m'). A limit on the number of CPU units the Redis instance may consume. 1 CPU core = 1000 units.
• memory_limit, (default: '500Mi'). A limit on the amount of memory the Redis instance may consume.
• sentinel_replicas (default: 1). The number of Redis sentinels to run.
• redis_replicas (default: 1). The number of replicated Redis servers to run. For a highly-available configuration, set this option to a value >= 3.
• storage_access_modes, (default: ['ReadWriteOnce']). An array of Kubernetes storage access modes for the persistent volume that the Redis instance will use to persist data.
• storage, (default: '1Gi'). The amount of persistent storage to request. Note that this is not a Redis memory limit, but rather a request for an amount of persistent (i.e. disk-based) storage Redis will write to. Persistent storage is necessary to prevent data loss if/when an individual Redis instance fails.

NOTE: All memory amounts must be strings parseable by Go's go-units module. Valid examples include '100Mi' (100 megabytes) and '2Gi' (2 gigabytes).

Example:

Kuby.define(:production) do
  kubernetes do

    add_plugin(:redis) do
      instance(:my_rails_cache) do
        # Launch three redis replicas (this is the recommended configuration
        # for high availability).
        redis_replicas 3

        # The amount of persistent storage to request.
        storage '5Gi'
      end
    end

  end
end

Redis instances defined in your Kuby config respond to the #service_name, #service_port, and #connection_params methods to facilitate connecting to them from your Rails app. One caveat is that clients must be sentinel-aware, but for the most part users won't have to worry about this since the redis-rb gem supports sentinel configurations.

In your Rails config (eg. config/environments/production.rb), point your cache store at your redis instance like so:

# make sure Kuby config is loaded
Kuby.load!

redis_instance = Kuby.environment.kubernetes
  .plugin(:redis)
  .instance(:my_rails_cache)

config.cache_store = :redis_cache_store, redis_instance.connection_params

You can also use the redis-rb gem directly:

# make sure Kuby config is loaded
Kuby.load!

require 'redis'

redis_instance = Kuby.environment.kubernetes
  .plugin(:redis)
  .instance(:my_rails_cache)

redis = Redis.new(**redis_instance.connection_params)

Licensed under the MIT license. See LICENSE for details.

• Cameron C. Dutro: http://github.com/camertron