PHP-RESQUE - 实现重试

由于PHP-Resque 的重试部分须要本身写，网上又没啥轮子，并且resque也已经好久不更新了，因此本身研究下resque的源码，而后也借鉴了Laravel的队列重试机制，实现了PHP-Resque的重试机制。

Resque地址

https://github.com/chrisboulton/php-resque

设计思路

1.这里须要阅读resque源码，resque的worker把失败的队列的数据都放在了"resque:failed"列表,数据以下。

2.个人项目是在yii2下统一处理resque的，打算开一个守护进程不断重试这些失败的队列。
3.我直接resque源码在新增了retry方法，在每个failJob的payload增长了attempts尝试次数，
一开始建立的时候attempts为0，重试以后attempts每次加1而后进行retry，直到到达尝试数次才中止重试。

核心代码

1. 调用重试的代码

/**
     * Retry Job
     */
    public function actionRetry()
    {
        $redis = new Redis('redis');
        while (true) {
            $json = $redis->rPop('resque:failed');
            $array = json_decode($json, true);
            if ($array) {
                $jobArray = $array['payload'];
                if ($jobArray['attempts'] < $this->retryTimes) {
                    //retry
                    \Resque::retry($jobArray, $array['queue']);
                    echo "Queued job " . $jobArray['id'] . ' has retry!' . "\n";
                } else {
                    //stop retry
                    $redis->lPush('resque:failed', [$json]);
                }
            }
            //take a sleep
            echo "*** Sleeping for ".$this->sleep. "\n";
            sleep($this->sleep);
        }

        return true;
    }

这里能够弄成守护进程一直处理失败的队列任务。

2. /php-resque/lib/Resque.php

/**
     * retry job and save it to the specified queue.
     *
     * @param array $jobArray The attempts of the the job .
     * array(
     *  'id'=> The id of the job
     *  'class' => The name of the class that contains the code to execute the job.
     *  'args' => Any optional arguments that should be passed when the job is executed.''
     *  'attempts'=> The retry attempts of the job
     * )
     * @param string $queue The name of the queue to place the job in.
     *
     * @return boolean
     */
    public static function retry($jobArray,$queue)
    {
        require_once dirname(__FILE__) . '/Resque/Job.php';
        $result = Resque_Job::retry($jobArray,$queue);
        if ($result) {
            Resque_Event::trigger('afterEnqueue', array(
                'class' => $jobArray['class'],
                'args'  => $jobArray['args'],
                'queue' => $queue,
            ));
        }

        return true;
    }

3. php-resque/lib/Resque/Job.php

/**
     * retry  job and save it to the specified queue.
     * *
     * @param array $jobArray The data of the job.
     * array(
     *  'id'=> The id of the job
     *  'class' => The name of the class that contains the code to execute the job.
     *  'args' => Any optional arguments that should be passed when the job is executed.''
     *  'attempts'=> The retry attempts of the job
     * )
     * @param string $queue The name of the queue to place the job in.
     *
     * @return string
     */
    public static function retry($jobArray,$queue)
    {
        $args = $jobArray['args'];
        if($args !== null && !is_array($args)) {
            throw new InvalidArgumentException(
                'Supplied $args must be an array.'
            );
        }
        $jobArray['attempts']++;
        Resque::push($queue, array(
            'class' => $jobArray['class'],
            'args'  => $args,
            'id'    => $jobArray['id'],
            'attempts'=>$jobArray['attempts']
        ));
        return true;
    }

结果：

我这里我设置了重试次数为3次，每隔5秒处理一个队列。

注意

1. 只有任务JOB实现类出现异常才会被从新扔回到"resque:failed"队列里面。
    2. 对于正常的业务错误不须要重试，目前只考虑到对curl会出现的异常进行重试。