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

In this issue, the editor will bring you an analysis of how to carry out Zookeeper distributed locks. The article is rich in content and analyzes and narrates it from a professional point of view. I hope you can get something after reading this article.

We use Zookeeper to implement a non-reentrant distributed lock!

Distributed locks are powerless for Java's JUC package. So we need to use the minimum version of Zookeeper, the set function of Redis, and database locks to implement distributed locks.

We know that in Zookeeper, the file directory format is used to store node content, where the node types are as follows:

Persistent node (PERSISTENT): after the node is created, it exists until the node is actively deleted.

Temporary node (EPHEMERAL): lifecycle and client session binding, which is automatically deleted once the client session fails.

Sequential node (SEQUENTIAL): when multiple threads create the same sequential node, each thread will get a numbered node, and the node number is incremented and not duplicated, as shown in the following figure:

As shown in the figure above, three threads create nodes with a path of / root/node respectively. You can see that there are three node nodes under root on the zk server side, and the thread number is the only increment.

Specifically in the process of node creation, it can be mixed, such as temporary sequential nodes (EPHEMERAL_SEQUENTIAL). In this paper, we use temporary sequential nodes to implement distributed locking.

Realization principle

Create a temporary sequential node, such as / root/node, assuming the return result is nodeId.

Get all the child nodes under / root, and compare the sequence number of the nodeId you created with all the child nodes to see if you have the lowest number. If it is the smallest, it is equivalent to acquiring the lock. If you are not the smallest, get a node smaller than you from all the child nodes, then set the event that listens for that node, and then suspend the current thread.

When the thread with the lowest number acquires the lock and deletes its corresponding nodeId after processing the business, the thread that activates the node one size larger than itself changes from blocking to running mode. The activated thread should be the one with the lowest node sequence number, and then the lock will be acquired.

Once you understand the principle, the code is very simple to write.

The constructor of ZookeeperDistributedLock creates the zkclient, registers the listening event, and then calls connectedSignal.await () to suspend the current thread. When zkclient is linked to the server, a SyncConnected event is sent to the listener, which determines that the current link has been established, then calls connectedSignal.countDown (); activates the current thread, and then creates the root node.

The method of lock is to create the sequential temporary nodes of / root/lockName internally, then get all the child nodes under / root, sort the child nodes, and then determine whether you are the lowest number. If you return true directly, you will be marked to acquire the lock successfully. Otherwise, it depends on whether a node one size smaller than you exists, registers the event of the node, then suspends the current thread and waits for the node one number smaller than you to release the lock to send the node deletion event, in which the current thread is activated.

The method of releasing the lock unlock is relatively simple, which is simply to delete the node created when the lock is acquired.

Zookeeper is very powerful, and when you really understand it, you will find that it can do a lot of things.

In terms of optical distribution locks, it can help us implement several kinds, such as the ones I listed below:

Reentrant lock Shared Reentrant Lock

Non-reentrant lock Shared Lock

Reentrant read-write lock Shared Reentrant Read Write Lock

Semaphore Shared Semaphore

Multi-lock object Multi Shared Lock

The above is the analysis of how to carry out Zookeeper distributed locks shared by the editor. 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.