rueidislock

rueidislock

A Redis Distributed Lock Pattern enhanced by Client Side Caching.

package main

import (
	"context"
	"github.com/redis/rueidis"
	"github.com/redis/rueidis/rueidislock"
)

func main() {
	locker, err := rueidislock.NewLocker(rueidislock.LockerOption{
		ClientOption: rueidis.ClientOption{InitAddress: []string{"node1:6379", "node2:6380", "node3:6379"}},
		KeyMajority:  2, // please make sure that all your `Locker`s share the same KeyMajority
	})
	if err != nil {
		panic(err)
	}
	defer locker.Close()

	// acquire the lock "my_lock"
	ctx, cancel, err := locker.WithContext(context.Background(), "my_lock")
	if err != nil {
		panic(err)
	}

	// "my_lock" is acquired. use the ctx as normal.
	doSomething(ctx)

	// invoke cancel() to release the lock.
	cancel()
}

Features backed by the Redis Client Side Caching

• The returned ctx will be canceled automatically and immediately once the KeyMajority is not held anymore, for example:
  • Redis down.
  • Related keys has been deleted by other program or administrator.
• The waiting Locker.WithContext will try acquiring the lock again automatically and immediately once it has been released by someone even from other program.

How it works

When the locker.WithContext is invoked, it will:

1. Try acquiring 3 keys (given that KeyMajority is 2), which are rueidislock:0:my_lock, rueidislock:1:my_lock and rueidislock:2:my_lock, by sending redis command SET NX PXAT or SET NX PX if FallbackSETPX is set.
2. If the KeyMajority is satisfied within the KeyValidity duration, the invocation is successful and a ctx is returned as the lock.
3. If the invocation is not successful, it will wait for client-side caching notification to retry again.
4. If the invocation is successful, the Locker will extend the ctx validity periodically and also watch client-side caching notification for canceling the ctx if the KeyMajority is not held anymore.