Skip to content
/ redis Public

Pure-Crystal Redis client, supporting clustering, RedisJSON, RediSearch, RedisGraph, and RedisTimeSeries

License

Notifications You must be signed in to change notification settings

jgaskins/redis

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

redis

A pure-Crystal implementation of the Redis protocol

Installation

  1. Add the dependency to your shard.yml:

    dependencies:
      redis:
        github: jgaskins/redis
  2. Run shards

Usage

require "redis"

redis = Redis::Client.new # Defaults to `localhost` port 6379

redis.set "foo", "bar"
redis.get "foo" # => "bar"

redis.incr "counter" # => 1
redis.incr "counter" # => 2
redis.decr "counter" # => 1

redis.del "foo", "counter" # => 2

Pipelined queries

To mitigate latency with multiple queries whose inputs and outputs are completely independent of each other, you can "pipeline" your queries by sending them all at once before reading them. To do this, you can use the pipeline method:

redis.pipeline do |pipe|
  pipe.incr "foo"
  pipe.set "bar", "baz"
  pipe.lpush "my-list", "my value"
end

The return value of pipeline will be an array containing the values of each of those calls in the order they were sent. So in this case, it might be [1, nil, 2] to match the return values of incr, set, and lpush, respectively.

Transactions

The Redis MULTI command begins a transaction, so you can use the multi method to execute a transaction against the server:

redis.multi do |txn|
  txn.set "foo", "bar"
  txn.incr "baz"
  txn.lpush "my-list", "my value"
end

The transaction is automatically committed with EXEC at the end of the block. If an exception occurs within the block, the transaction will be rolled back with DISCARD before exiting the block.

You may also call txn.discard, which will effectively disable the transaction (all further methods called on the transaction do nothing), but will not exit the block. You will need to exit the block explicitly with break if there are operations within the block that cannot be rolled back, such as sending an email or sending a request to a third-party API.

The reason for this is that the only way to exit a containing block from an inner method in Crystal is to raise an exception, and this library chooses not to use exceptions for flow control.

Beyond localhost

To use a Redis server that isn't at localhost:6379, pass a URI to the client. For example, if you store it in your shell environment:

redis = Redis::Client.new(URI.parse(ENV["REDIS_URL"]))

# ... or ...

redis = Redis::Client.from_env("REDIS_URL")

To connect via SSL, make sure you use the rediss:// URL scheme. If your Redis server requires a password or uses a different database slot than 0, make sure you include them in the URL:

redis = Redis::Client.new(URI.parse("rediss://:[email protected]/3"))

Connection Pool

The Redis::Client maintains a connection pool, so there is no need to run your own within your application. When you execute a command on the Redis::Client, it is automatically executed against a connection. When you execute a pipeline or transaction with multi, all commands within that block will automatically be routed to the same connection.

Configuration

For this shard, we use the following default setting (outside of the Standard Lib defaults);

max_idle_pool_size = 25

You can override this manually using the URI parameters. All other settings follow the DB::Pool defaults.

The behaviour of the connection pool can be configured from a set of query string parameters in the connection URI.

Name Default value
initial_pool_size 1
max_pool_size 0 (unlimited)
max_idle_pool_size 1
checkout_timeout 5.0 (seconds)
retry_attempts 1
retry_delay 1.0 (seconds)

See Crystal guides to learn more.

Example

pool_params = "?initial_pool_size=1&max_pool_size=10&checkout_timeout=10&retry_attempts=2&retry_delay=0.5&max_idle_pool_size=50"
redis = Redis::Client.new(URI.parse("redis://localhost:6379/0#{pool_params}"))

Recommendations

If you encounter any issues, keep these setting the same;

  • initial_pool_size
  • max_pool_size
  • max_idle_pool_size

Example:

initial_pool_size = 50
max_pool_size = 50
max_idle_pool_size = 50

TCP Keep-Alive

The Redis::Client uses a pool of Redis::Connection under the hood. Within Redis::Connection we create a TCPSocket, which can accept keepalive params. The TCP keepalive settings can help you mitigate Redis connection stability issues.

NOTE: This behaviour is disabled by default. See Configuration below on how to enable it.

Configuration

For this shard, we use the following override setting;

Name Default value
keepalive false
keepalive_count 3
keepalive_idle 60
keepalive_interval 30

You can override this manually using the URI parameters. The settings above have proven to have good results in production environments. However, every environment is different, so tweaking these settings may be necessary.

See Crystal API to learn more.

Example

params = "?keepalive=true&keepalive_count=5&keepalive_idle=10&keepalive_interval=15"

redis = Redis::Client.new(URI.parse("redis://localhost:6379/0#{params}"))
# or direct connections
redis = Redis::Connection.new(URI.parse("redis://localhost:6379/0#{params}"))

Recommendations

Enable this setting with the defaults if you are encountering connection issues.

Example:

params = "?keepalive=true"

redis = Redis::Client.new(URI.parse("redis://localhost:6379/0#{params}"))
# or direct connections
redis = Redis::Connection.new(URI.parse("redis://localhost:6379/0#{params}"))

Development

Make sure you have a Redis or KeyDB server running locally on port 6379.

Redis must be installed with a stack server for the full text search (ft) and time series modules (ts) in order for all specs to run.

With Docker

You can use this for your docker-compose file

redis:
  image: redis/redis-stack-server
  ports:
    - "6379:6379"

With Homebrew

Install the redis-stack from homebrew

Contributing

  1. Fork it (https://github.com/jgaskins/redis/fork)
  2. Create your feature branch (git checkout -b my-new-feature)
  3. Commit your changes (git commit -am 'Add some feature')
  4. Push to the branch (git push origin my-new-feature)
  5. Create a new Pull Request

Contributors