Experience using queue service (queue) in Laravel 4.2, laravelqueue_PHP tutorial

Using experience of queue service (queue) in Laravel 4.2, laravelqueue

In the past half month, I participated in rewriting a WeChat official account back-end system, using laravel 4.2 for the first time, and laravel’s proud queue service (queue).

Since the entire system involves multi-terminal interaction and a large number of voice transmission and processing services, we found that the response time was too long in some places. The previous system was based on node.js and mongoDB. Since node is inherently asynchronous and has a daemon process, this problem did not occur. However, this rewrite will inevitably introduce asynchronous processes. Queue comes into our sight.

According to this page of "Chinese Documentation" which is almost entirely in English, laravel has just introduced redis as a queue storage in version 4.2, which is very good news. OK, that’s it for the background introduction, let’s talk about the dry stuff below.

The queue service in laravel is no different from other queue services. They are the simplest and most common processes that are most in line with human thinking: there is a place to store queue information, a PHP process writes tasks at runtime, and another A PHP daemon polls the queue for information, executes and deletes tasks that meet the execution requirements. Since PHP is a URL-driven synchronous language and inherently blocking, it is not surprising that Laravel provides a daemon tool to query and execute queue information.

Laravel's queue configuration file is /app/config/queue.php. In the Default Queue Driver, you can choose from "sync", "beanstalkd", "sqs", "iron", "redis". drive.

1. sync is a synchronization driver for local debugging

2. beanstalkd is a professional queue service driver: http://kr.github.io/beanstalkd/

3. sqs and iron are foreign third-party queue services

4. The last item redis gives us a reason to use redis, so we can migrate all cache services and session services to redis.

0. By the way, don’t use mysql as the session driver. The processing time of 1S is not a dream. Hey, who cares? I’m talking about you, Brother 1S!

The queue service needs to create a new task class. As an independent class, they do not need to inherit the class, because the tasks in the queue are called independently by the PHP daemon when they are executed. Of course, if you want to use other classes If you call it again, there will be no error. Previously, I separated many additional services into a separate folder /app/services, such as input information verification validator, special security verification module, etc. This time the queue classes are located in it.

The use of queue is very simple. Here is a simple example:

This is a standard queue push process. Of course, here I put the CurlJsonQueue class in the services root directory. This directory has been registered by me in the "classmap" of "autoload" in composer.json. It is located in the top-level namespace and can be called directly. If needed, AppOOXX can be written in non-top-level namespaces. Our system requires a lot of interaction with the WeChat server, so we created this class independently.

The default method of this class is fire(), and the parameters are also fixed two $job and $data. Since I encapsulated the post curl module in BaseController, I called it. In addition, there is a small pit here. When I wrote base_post_curl(), I used protected, which caused use BaseController to be invalid and must be inherited.

By executing the above code, a new task is placed in the queue, and laravel starts the daemon process through the following command:

Then the daemon starts processing the queue. Please adjust the paths of the PHP commands and artisan files in this code yourself.

You may have noticed that the queue system we are going to use uses redis and PHP command line. If you are in a test environment, you can add a boot or even start it manually, but in a production environment, you need more stable tools. To guard these two programs, we use supervisor. For the installation and configuration of supervisor, you can refer to this article: http://blog.segmentfault.com/qianfeng/1190000000532561 Note that there are some pitfalls in the article, please check it yourself. . .

OK, after everything is configured, run the redis and PHP command lines, and the entire system will start to run happily~

Usage experience:

The queue service is super easy to use. The previous interaction process with the app took 6-7S. After asynchronous, it was reduced to less than 2S. This is basically the transmission time and PHP code running time. Time-consuming special operations have been asynchronous. However, the queue service defaults to 1S to open a process to check whether there are any services that can be run in redis. On the Alibaba Cloud server, it can account for about 10% of a single core. The consumption is slightly larger, and the queue processing time is relatively long, because there is no File loading benefits during previous synchronization. However, if there are multiple tasks, the PHP process will be executed continuously, not one every 1 second.

Let’s talk about the pitfalls:

1. Since the queue core class uses a special function, variables without clear types will be stored in json in the form of single-element arrays and then stored in redis. The solution is to add ''. in front of each data to be put in. Since the $url and $json above have already been type-declared in quotes, this step was not performed.

2. If you want to pass the URL to the queue, the system queue class will add two \ before each /. This may have a fatal impact on some special operations. (Just kidding, is it as deadly as the one above!)

Use Contains (the content you want to check) to make a judgment (return value Boolean type). For example: namespace TestBed
{ public class MainTest
{ public static void Main()
{
int[] numbers = { 1,1,2,3,4,2,5,6, 7};
Queue q = new Queue();
foreach (var item in numbers)
{
if (!q.Contains(item))
{
q.Enqueue(item);
}
}
} }
}

Dear, it’s different. The first expression controls the pointer of the queue to jump in a loop within the range of queuesize, while the second expression pointer will continuously jump backward by the memory distance of one node. It is very possible If it exceeds the effective memory range of the queuesize you defined and occupies unallocated memory, the compiler will report an error. Do you understand? ? ?