Skip to content

Latest commit

 

History

History
76 lines (49 loc) · 3.67 KB

README.md

File metadata and controls

76 lines (49 loc) · 3.67 KB
Raw

zookeeper-node-shared-lock

This library implements a distributed global lock in Node.js using Zookeeper. The lock protocol is a variant of the Zookeeper recipe.

The library is only suitable for educational references, and should not be used directly in production.

API

Creation

zookeeper-node-shared-lock returns a factory function that creates a lock service. The function expects a set of configuration options. Once the service is created, you may subscribe to the ready event. Once the ready event is fired, you may begin locking/unlocking resources.

Parameters

The configuration parameters includes the following:

  • host - The hostname or IP of the zookeeper service.
  • port - The port number of zookeeper service.
  • resourceName - The name of the resource. Internally, this will be used as the parent ephemeral znode that holds all the child nodes.
  • maxRetryCount - The maximum number of times the lock will retry after request timeout. By default, this value is 0, which implies we will wait indefinitely for this resource.
  • initialRetryWaitMs - The amount of time to wait after initial failure. As failures accumulates, the retry wait will increase using the binary exponential backoff algorithm. This value has no effect if maxRetryCount is 0.

Sample

const lockService = require('zookeeper-node-shared-lock')({
  resourceName: 'my_global_resource', // the name of the resource
  host: 'localhost', // default: localhost
  port: 2181 // default: 2181
});

lockService.events.on('ready', () => {

  lockService.lock('resource1', 0, (err, acquiredResource) => {
    console.log('resource locked');

    lockService.unlock(acquiredResource, (err) => {
      console.log('resource unlocked');
    });
  });
});

lock(resourceId, ttl, [cb])

lock acquires a resource asynchronously. The resourceId is the shared resource to lock. If the shared resource is acquired, or failed to be acquired, the cb(err, acquiredResource) will be invoked.

The return acquiredResource argument is used to unlock the acquired resource.

If ttl (in milliseconds) is provided, the lock service will automatically release the lock upon TTL expiration. User may subscribe to the ttlExpired event for this notification.

If ttl is 0 (default), the acquired lock resource will be held indefinitely.

unlock(acquiredResource, [cb])

unlock releases the acquired resource asynchronously. The acquiredResource is the acquired resource returned from lock.

Events

  • ready - fired when Lock Service is ready.
  • ttlExpired - fired with function (resourceId) if a resource is being forcefully released.
  • disconnected - fired when we lost connection to Zookeeper

ZooKeeper Lock Recipe

Clients wishing to obtain a lock do the following:

  1. Call create() with a pathname of "locknode/guid-lock-" and the sequence and ephemeral flags set. The guid is needed in case the create() result is missed.

  2. Call getChildren() on the lock node without setting the watch flag (this is important to avoid the herd effect).

  3. If the pathname created in step 1 has the lowest sequence number suffix, the client has the lock and the client exits the protocol.

  4. The client calls exists() with the watch flag set on the path in the lock directory with the next lowest sequence number.

  5. if exists() returns false, go to step 2. Otherwise, wait for a notification for the pathname from the previous step before going to step 2. The unlock protocol is very simple: clients wishing to release a lock simply delete the node they created in step 1.