Redlock
This is a node.js implementation of the redlock algorithm for distributed redis locks. It provides strong guarantees in both single-redis and multi-redis environments, and provides fault tolerance through use of multiple independent redis instances or clusters.
High-Availability Recommendations
- Use at least 3 independent servers or clusters
- Use an odd number of independent redis servers for most installations
- Use an odd number of independent redis clusters for massive installations
- When possible, distribute redis nodes across different physical machines
Using Cluster/Sentinel
It is completely possible to use a single redis cluster or sentinal configuration by passing one preconfigured client to redlock. While you do gain high availability and vastly increased throughput under this scheme, the failure modes are a bit different, and it becomes theoretically possible that a lock is acquired twice:
Assume you are using eventually-consistent redis replication, and you acquire a lock for a resource. Immediately after acquiring your lock, the redis master for that shard crashes. Redis does its thing and fails over to the slave which hasn't yet synced your lock. If another process attempts to acquire a lock for the same resource, it will succeed!
This is why redlock allows you to specify multiple independent nodes/clusters: by requiring consensus between them, we can safely take out or fail-over a minority of nodes without invalidating active locks.
To learn more about the the algorithm, check out the redis distlock page.
How do I check if something is locked?
Redlock cannot tell you with certainty if a resource is currently locked. For example, if you are on the smaller side of a network partition you will fail to acquire a lock, but you don't know if the lock exists on the other side; all you know is that you can't guarantee exclusivity on yours.
That said, for many tasks it's sufficient to attempt a lock with retryCount=0
, and treat a failure as the resource being "locked" or (more correctly) "unavailable",
Installation
npm install --save redlock
Configuration
Redlock can use node redis, ioredis or any other compatible redis library to keep its client connections.
A redlock object is instantiated with an array of at least one redis client and an optional options
object. Properties of the Redlock object should NOT be changed after it is firstused, as doing so could have unintended consequences for live locks.
var client1 = ;var client2 = ;var client3 = ;var Redlock = ; var redlock = // you should have one client for each redis node // in your cluster client1 client2 client3 // the expected clock drift; for more details // see http://redis.io/topics/distlock driftFactor: 001 // time in ms // the max number of times Redlock will attempt // to lock a resource before erroring retryCount: 10 // the time in ms between attempts retryDelay: 400 // time in ms // the max time in ms randomly added to retries // to improve performance under high contention // see https://www.awsarchitectureblog.com/2015/03/backoff.html retryJitter: 400 // time in ms ;
Error Handling
Because redlock is designed for high availability, it does not care if a minority of redis instances/clusters fail at an operation. If you want to write logs or take another action when a redis client fails, you can listen for the clientError
event:
// ... redlock; // ...
Usage (promise style)
Locking & Unlocking
// the string identifier for the resource you want to lockvar resource = 'locks:account:322456'; // the maximum amount of time you want the resource locked,// keeping in mind that you can extend the lock up until// the point when it expiresvar ttl = 1000; redlock;
Locking and Extending
redlock;
Usage (disposer style)
Locking & Unlocking
var using = using; // the string identifier for the resource you want to lockvar resource = 'locks:account:322456'; // the maximum amount of time you want the resource locked,// keeping in mind that you can extend the lock up until// the point when it expiresvar ttl = 1000; // if we weren't able to reach redis, your lock will eventually// expire, but you probably want to do something like log that// an error occurred; if you don't pass a handler, this error// will be ignored { console;} ; // <-- unlock is automatically handled by bluebird
Locking and Extending
; // <-- unlock is automatically handled by bluebird
Usage (callback style)
Locking & Unlocking
// the string identifier for the resource you want to lockvar resource = 'locks:account:322456'; // the maximum amount of time you want the resource locked,// keeping in mind that you can extend the lock up until// the point when it expiresvar ttl = 1000; redlock;
Locking and Extending
redlock
API Docs
Redlock.lock(resource, ttl, ?callback)
resource (string)
resource to be lockedttl (number)
time in ms until the lock expirescallback (function)
callback returning:err (Error)
lock (Lock)
Redlock.unlock(lock, ?callback)
lock (Lock)
lock to be releasedcallback (function)
callback returning:err (Error)
Redlock.extend(lock, ttl, ?callback)
lock (Lock)
lock to be extendedttl (number)
time in ms to extend the lock's expirationcallback (function)
callback returning:err (Error)
lock (Lock)
Redlock.disposer(resource, ttl, ?unlockErrorHandler)
resource (string)
resource to be lockedttl (number)
time in ms to extend the lock's expirationcallback (function)
error handler called with:err (Error)
Lock.unlock(?callback)
callback (function)
callback returning:err (Error)
Lock.extend(ttl, ?callback)
ttl (number)
time in ms to extend the lock's expirationcallback (function)
callback returning:err (Error)
lock (Lock)