Riak Elixir Client

A Riak client written in Elixir. Now includes connection pooling with pooler and a variety of other improvements from riex.

Setup

Prerequisites

• Riak 2.0+
• Elixir 1.0+

In an Elixir application

Add the following to mix.exs

...
def application do
  [ applications: [ :riak ]]
end
...
defp deps do
  [ {:riak, "~> 1.0"} ]
end
...

Usage

Establishing a Riak connection

{:ok, pid} = Riak.Connection.start_link('127.0.0.1', 8087) # Default values

Connection Pooling

Most functions in this module can be called by passing the pid of the established connection or using a pool of connections (provided by pooler). Define pools by using the group riak. Following is an example config/config.exs:

[pooler: [pools: [
  [ name: :riaklocal1,
    group: :riak,
    max_count: 10,
    init_count: 5,
    start_mfa: {Riak.Connection, :start_link, []}
  ],
   [ name: :riaklocal2,
    group: :riak,
    max_count: 15,
    init_count: 2,
    start_mfa: {Riak.Connection, :start_link, ['127.0.0.1', 9090]}
  ] ]
]]

For an example using this functionality with a local Riak instance, check config/config.exs. More information about Elixir configuration can be found on [http://elixir-lang.org(http://elixir-lang.org)]: Application environment and configuration.

Once a pool configuration is properly defined in a project, calls to Riak can omit the pid. For example:

This call uses a pid from the pool of connections provided by pooler:

Riak.delete("user", key)

This call requires a pid obtained by first calling Riak.Connection.start_link:

Riak.delete(pid, "user", key)

Save a value

o = Riak.Object.create(bucket: "user", key: "my_key", data: "Han Solo")
Riak.put(pid, o)

Find an object

o = Riak.find(pid, "user", "my_key")

Update an object

o = %{o | data: "Something Else"}
Riak.put(pid, o)

Delete an object

Using key

Riak.delete(pid, "user", key)

Using object

Riak.delete(pid, o)

Datatypes

Riak Datatypes (a.k.a. CRDTs) are avaiable in Riak versions 2.0 and greater. The types included are: maps, sets, counters, registers and flags.

Setup

Datatypes require the use of bucket-types. Maps, sets, and counters can be used as top-level bucket-type datatypes; Registers and flags may only be used within maps.

The following examples assume the presence of 3 datatype enabled bucket-types. You can create these bucket-types by running the following commands on a single Riak node in your cluster:

Bucket-Type: counters

riak-admin bucket-type create counters '{"props":{"datatype":"counter"}}'
riak-admin bucket-type activate counters

Bucket-Type: sets

riak-admin bucket-type create sets '{"props":{"datatype":"set"}}'
riak-admin bucket-type activate sets

Bucket-Type: maps

riak-admin bucket-type create maps '{"props":{"datatype":"map"}}'
riak-admin bucket-type activate maps

Counters

Create a counter (alias Riak.CRDT.Counter):

Counter.new
  |> Counter.increment
  |> Counter.increment(2)
  |> Riak.update("counters", "my_counter_bucket", "my_key")

Fetch a counter:

counter = Riak.find("counters", "my_counter_bucket", "my_key")
  |> Counter.value

counter will be 3.

NOTE: “Counter drift” is a possibility that needs to be accounted for with any distributed system such as Riak. The problem can manifest itself during failure states in either your applicaiton or Riak itself. If an increment operation fails from the client’s point of view, there is not sufficient information available to know whether or not that call made it to zero or all of the replicas for that counter object. As such, if the client attempts to retry the increment after recieving something like a error code 500 from Riak, that counter object is at risk of drifting positive. Similarly if the client decides not to retry, that counter object is at risk of drifting negative.

For these reasons, counters are only suggested for use-cases that can handle some (albeit small) amount of counter drift. Good examples of appropriate use-cases are: Facebook likes, Twitter retweet counts, Youtube view counts, etc. Some examples of poor use-cases for Riak counters are: bank account balances, anything related to money. It is possible to implement these types of solutions using Riak, but more client side logic is necessary. For an example of a client-side ledger with tunable retry options, check github.com/drewkerrigan/riak-ruby-ledger. Another approach could be the client-side implementation of a HAT (Highly Available Transaction) algorithm.

Sets

Create a set (alias Riak.CRDT.Set):

Set.new
  |> Set.put("foo")
  |> Set.put("bar")
  |> Riak.update("sets", "my_set_bucket", "my_key")

And fetch the set:

set = Riak.find("sets", "my_set_bucket", "my_key")
  |> Set.value

Where set is an orddict.

Maps

Maps handle binary keys with any other datatype (map, set, flag, register and counter).

Create a map (alias Riak.CRDT.Map):

register = Register.new("some string")
flag = Flag.new |> Flag.enable
Map.new
  |> Map.put("k1", register)
  |> Map.put("k2", flag)
  |> Riak.update("maps", "my_map_bucket", "map_key")

And fetch the map:

map = Riak.find("maps", "my_map_bucket", key) |> Map.value

Where map is an orddict.

Examples

Check the examples/ directory for a few example elixir applications using the riak client.

For more functionality, check test/ directory.

Tests

MIX_ENV=test mix do deps.get, test

License

Copyright 2015 Drew Kerrigan.
Copyright 2014 Eduardo Gurgel.

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.