Shulou(Shulou.com)05/31 Report--

In this issue, the editor will bring you the code implementation of Zookeeper distributed locks. The article is rich in content and analyzes and describes for you from a professional point of view. I hope you can get something after reading this article.

Preface Code implementation of Zookeeper distributed Lock

In the last blog post, we have analyzed how Zookeeper can help us implement distributed locks. Let's look directly at the code:

[distributed client]

[method of acquiring distributed locks lock: initializing ZK]

[method of obtaining distributed locks lock: creating temporary nodes and determining the minimum path]

[main Test]

[running result]

It should be noted that even if you monitor the deletion Watcher of a node with a smaller serial number than yourself, you also need to confirm it again!

As a result, it is clear that each thread acquires the lock in an orderly manner.

Zkclient

Zkclient is based on zookeeper native API to do a little encapsulation, simplifying the complexity of ZK.

Take a look at the code:

Let's take a look at the difference between the use of zkclient and the previous zookeeper-based native API.

First, native API requires us to use CountDownLatch to ensure initialization of ZK, and now zkclient helps us block out this detail

Second, native API cannot create nodes recursively, while zkclient can help us create non-existent parent nodes and delete them recursively.

Third, support serialization operations, the above code you can probably see some clues, that is, we have gone from operating byte [] to operating String. (in fact, you only need to implement the ZkSerializer interface in zkclient to complete the conversion from Object to byte [], but in actual development, it's good to use JSON! )

Fourth, and the most important point is that zkclient separates the operation of the node from the monitoring of the node, which are coupled together in the native API! Ideologically, it is easy to understand; from the code point of view, it is also concise (if written together, the head is big); more conveniently, zkclient completes the function of repeating watch for us!

[watch subscription mechanism]

See, isn't it a bit like MQ's subscription mechanism? it's very useful! But it is also a little imperfect, why is the data change of the child node not monitored? it is a bit imhuman! Luckily there is Curator...]

However, we know that ZK has many application scenarios, such as implementing distributed locks. Zkclient does not encapsulate it for us, but the Curator framework can help us do it!

Curator

In order to better implement Java to operate Zookeeper server, Curator framework appeared later, which is very powerful. It is now a top-level project of Apache, with many rich features, such as session overtime reconnection, master-slave election, distributed counters, distributed locks, etc., which is very conducive to the development of Zookeeper in complex scenarios.

POM file:

Add, delete, change and check:

The Curator framework uses chain programming style and is easy to read!

Note that both native API and zkclient-based API provide connectTimeout, while Curator provides sessionTimeout, which is very powerful.

Both native API and zkclient support asynchronous callbacks, but the Curator framework supports asynchronous callbacks while adding thread pools for us to optimize!

[NodeCacheListener]

[PathChildrenCacheListener]

For Curator, in order to solve the problem of repetitive Watch, it introduces a new idea: the mechanism of comparing Cache with ZK SERVER. Whether it is native API or ZKCLIENT-based, their solution is to re-register!

Thinking determines the way out! Curator compares the Cache of the client with the data of ZK SERVER through event-driven, which naturally solves the function of repeating WATCH! Why Curator can become the top project of Apache, I think it is probably because of its different design ideas!

In Curator, there are two kinds of Listener, one is the NodeCacheListener of the monitoring node, and the other is the PathChildrenCacheListener of monitoring the child node. PathChildernCacheListener can monitor the addition, modification and deletion of child nodes, which is very useful!

The above is the code implementation of how to carry out Zookeeper distributed locks shared by Xiaobian. If you happen to have similar doubts, you might as well refer to the above analysis to understand. If you want to know more about it, you are welcome to follow the industry information channel.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.