2014-04-28 12:32
浏览 59

PHP pThreads - 你如何执行垃圾收集?

Given the following code, how can you ensure that the completed MyWorker objects are destroyed/their memory freed?

Due to what my script does I need ~50 threads constantly obtaining data from cURL, and processing it.

I've tried both having the threads never leave run(), or as shown in this sample code where they leave run and have the collect function spawn a new copy of them.

But not matter what I hit the memory limits after a minute or so. Could you tell me what I'm doing wrong?

class MyWorker extends Threaded
    public $complete;
    public function __construct() {$this->complete = false;}
    public function run() {$this->complete = true;}

$pool = new Pool(50);
for($i=0; $i<50; $i++)
    $pool->submit(new MyWorker());
    global $pool;
    if($worker->complete == true)
        $pool->submit(new MyWorker());
    return $worker->complete;

图片转代码服务由CSDN问答提供 功能建议



我已经尝试过让线程永远不会离开 run(),或者如这个示例代码所示,他们离开run并让collect函数产生一个 他们的新副本。

但不管我在一分钟左右后达到内存限制的程度。 你能告诉我我做错了什么吗?

 public $ complete; 
 public function __construct(){$ this-&gt; complete = false;} 
公共函数运行 (){$ this-&gt; complete = true;} 
 $ pool = new Pool(50); 
for($ i = 0; $ i&lt; 50; $ i ++)
 $ pool-  &gt;提交(新MyWorker()); 
 $ pool-&gt; collect(函数($ worker)
全局$ pool; 
 if($ worker-&gt; complete == true)
 $  pool-&gt; submit(new MyWorker()); 
返回$ worker-&gt; complete; 
 $ pool-&gt; shutdown(); 
  • 点赞
  • 写回答
  • 关注问题
  • 收藏
  • 邀请回答

1条回答 默认 最新

  • douxin0251 2014-04-30 09:08


    Why should I collect anyway ?

    The Worker threads provided by pthreads require that the programmer retain the correct references to Threaded objects that are being executed. This is difficult for the programmer to achieve in userland reliably, so pthreads provides the Pool abstraction of Workers which maintains references for you.

    In order to maintain those reference pthreads needs to know when an object is garbage, it provides the Pool::collect interface for this purpose. Pool::collect takes a Closure which should accept a Threaded object and return boolean true if the passed object is finished executing.


    The task at hand ...

    In order to keep submitting tasks for execution and not exhaust resources, you must create a queue of completed tasks for resubmission to the Pool

    The following code demonstrates a sane way of doing this:

    define("LOG", Mutex::create());
    /* thread safe log to stdout */
    function slog($message, $args = []) {
        $args = func_get_args();
        if (($message = array_shift($args))) {
            echo vsprintf(
    ", $args);
    class Request extends Threaded {
        public function __construct($url) {
            $this->url = $url;
        public function run() {
            $response = @file_get_contents($this->url);
            slog("%s returned %d bytes",
                $this->url, strlen($response));
        public function getURL()        { return $this->url; }
        public function isQueued()      { return $this->queued; }
        public function reQueue()       { $this->queued = true; }
        protected $url;
        protected $queued = false;
    /* create a pool of 50 threads */
    $pool = new Pool(50);
    /* submit 50 requests for execution */
    while (@$i++<50) {
        $pool->submit(new Request(sprintf(
            "", md5($i))));
    do {
        $queue = array();
        $pool->collect(function($request) use ($pool, &$queue) {
            /* check for items to requeue */
            if ($request->isQueued()) {
                /* get the url for the request, insert into queue */
                $queue[] = 
                /* allow this job to be collected */
                return true;
        /* resubmit completed tasks to pool */
        if (count($queue)) {
            foreach ($queue as $queued)
                $pool->submit(new Request($queued));
        /* sleep for a couple of seconds here ... because, be nice ! */
        usleep(2.5 * 1000000);
    } while (true);
    点赞 评论

相关推荐 更多相似问题