This library is an implementation of Redlock (Redis destributed lock)
If available in Hex, the package can be installed
by adding redlock
to your list of dependencies in mix.exs
:
def deps do
[
{:redlock, "~> 1.0.21"}
]
end
Documentation can be generated with ExDoc and published on HexDocs. Once published, the docs can be found at https://hexdocs.pm/redlock.
resource = "example_key:#{user_id}"
lock_exp_sec = 10
case Redlock.lock(resource, lock_exp_sec) do
{:ok, mutex} ->
# some other code which write and read on RDBMS, KVS or other storage
# call unlock finally
Redlock.unlock(resource, mutex)
:error ->
Logger.error "failed to lock resource. maybe redis connection got trouble."
{:error, :system_error}
end
Or you can use transaction
function
def my_function() do
# do something, and return {:ok, :my_result} or {:error, :my_error}
end
def execute_with_lock() do
resource = "example_key:#{user_id}"
lock_exp_sec = 10
case Redlock.transaction(resource, lock_exp_sec, &my_function/0) do
{:ok, :my_result} ->
Logger.info "this is the return-value of my_function/0"
:ok
{:error, :my_error} ->
Logger.info "this is the return-value of my_function/0"
:error
{:error, :lock_failure} ->
Logger.info "if locking has failed, Redlock returns this error"
:error
end
end
children = [
# other workers/supervisors
{Redlock, [pool_size: 2, ...]}
]
Supervisor.start_link(children, strategy: :one_for_one)
readlock_opts = [
pool_size: 2,
drift_factor: 0.01,
max_retry: 3,
retry_interval_base: 300,
retry_interval_max: 3_000,
reconnection_interval_base: 500,
reconnection_interval_max: 5_000,
# you must set odd number of server
servers: [
[host: "redis1.example.com", port: 6379],
[host: "redis2.example.com", port: 6379],
[host: "redis3.example.com", port: 6379]
]
]
pool_size
: pool_size of number of connection pool for each Redis master node, default is 2drift_factor
: number used for calculating validity for results, see https://redis.io/topics/distlock for more detail.max_retry
: how many times you want to retry if you failed to lock resource.retry_interval_max
: (milliseconds) used to decide how long you want to wait untill your next try after a lock-failure.retry_interval_base
: (milliseconds) used to decide how long you want to wait untill your next try after a lock-failure.reconnection_interval_base
: (milliseconds) used to decide how long you want to wait until your next try after a redis-disconnectionreconnection_interval_max
: (milliseconds) used to decide how long you want to wait until your next try after a redis-disconnectionservers
: host, port and auth settings for each redis-server. this amount must be odd. Auth can be omitted if no authentication is reaquired
the interval(milliseconds) is decided by following calculation.
min(XXX_interval_max, (XXX_interval_base * (attempts_count ** 2)))
readlock_opts = [
pool_size: 2,
drift_factor: 0.01,
max_retry: 3,
retry_interval_base: 300,
retry_interval_max: 3_000,
reconnection_interval_base: 500,
reconnection_interval_max: 5_000,
cluster: [
# first node
[
# you must set odd number of server
[host: "redis1.example.com", port: 6379, auth: password],
[host: "redis2.example.com", port: 6379, auth: password],
[host: "redis3.example.com", port: 6379, auth: password]
],
# second node
[
# you must set odd number of server
[host: "redis4.example.com", port: 6379],
[host: "redis5.example.com", port: 6379],
[host: "redis6.example.com", port: 6379]
],
# third node
[
# you must set odd number of server
[host: "redis7.example.com", port: 6379],
[host: "redis8.example.com", port: 6379],
[host: "redis9.example.com", port: 6379]
]
]
]
Set cluster
option instead of servers
, then Redlock works as cluster mode.
When you want to lock some resource, Redlock chooses a node depends on a resource key with consistent-hashing way (ketama algorithm using md5).