我正在使用 pcntl 来加速一个相当复杂的 CLI php 脚本，它主要由一个类组成，负责在我的应用程序上发送所有自动电子邮件。
我的目标如下: 我想在 foreach 循环中将每个进程分配给某个任务，我使用的实现是下面的代码示例中所示的实现。
问题是，一旦你fork了一个进程，它就会异步执行，并且还会得到父进程堆栈的副本。 在我的例子中，一个任务只是执行了几次， 我的问题是，我如何才能将此脚本设计得更智能以避免此类行为？
代码:
/**
@description this is the main procedure of this class, it iteratates over the relevant tasks and sends the emails using the SendGrid wrapper class
@see SendGridWrapper
@return void
*/
public function execute(){
if(!isset($this->tasks)){
throw new exception("Please call getRelevantTasks() prior to trying to execute anything");
}
$proccesses = array();
foreach($this->tasks as $myTask){
$pid = pcntl_fork();
if($pid){
$proccesses[] = $pid;
}
else if($pid == -1){
die('FORK FAILED, STATUS -1');
}
else{
if(isset($myTask['recipient_model'])){
$this->currentModel = $myTask['recipient_model'];
$lang = $myTask['lang'];
$classPath = self::$modelsDir . $myTask['recipient_model'] . '.php';
$className = $myTask['recipient_model'];
if(!class_exists($myTask['recipient_model'] )){
require_once(dirname(__FILE__) . '/../' . $classPath);
}
else if(isset($recipientFetcher)){
unset($this->model);
unset($this->mailingList);
unset($this->substitutionList);
}
$this->model = null;
$this->mailingList = null;
$this->substitutionList = null;
$this->model = new $className($myTask['lang']);
$addresses = $this->model->getMailRecipients();
if(empty($addresses) || sizeof($addresses) == 0){
continue;
}
$this->model->prepare();
$this->substitutionList = $this->model->getDynamicParams();
}
else{
throw new exception('No recipient model was found');
}
foreach($addresses as $myMail){
$this->mailingList[$myMail['personal_email']] = $myMail['contact_name'];
}
$templatePath = dirname(__FILE__) . '/../';
$templatePath .= $lang ? self::$templatesDirEn . $myTask['html_email_path'] : self::$templatesDirHe . $myTask['html_email_path'];
$html = file_get_contents($templatePath);
$this->sendMail($html, $myTask['task_schedule_id']);
echo "model:" . $myTask['recipient_model'];
echo $this->log;
$this->log = "";
die("\r\n Child proccess has been executed successfully\r\n");
}
}
if($pid){
foreach($proccesses as $key => $val){
pcntl_waitpid($val, $status, WUNTRACED);
}
}
}
提前致谢， 奥列格。
最佳答案
简介
我看到你正在尝试发送邮件 $this->sendMail($html, $myTask['task_schedule_id']); 我认为尝试使用多个进程来发送邮件是一个非常糟糕的主意这个任务。您应该考虑为此任务使用消息队列，因为电子邮件可能非常慢。
使用队列系统
你应该使用 Gearman , ZeroMQ或 Beanstalkd为了这个任务。最坏的情况使用 memcached 实现您自己的简单消息队列。
这是一个典型的 Gearman 示例:https://stackoverflow.com/questions/13855907/when-to-send-auto-email-instantly-on-button-click-or-later
快速修复
删除所有这些代码并将其放入名为 execute_worker 的函数中，您可以在其中将任务推送给它
// Break Task to groups
$tasks = array_chunk(range("A", "Z"), 10);
foreach($tasks as $task) {
$pid = pcntl_fork();
if ($pid == - 1) {
throw new ErrorException('FORK FAILED, STATUS -1');
break;
}
if ($pid == 0) {
execute_worker($task); // In Child
exit(); // In Child
}
}
使用线程
您还可以使用 Worker或 Thread在 PHP 中使用 pThreads以加快处理速度。
简单项目
据说file_get_contents 与 curl 相比速度较慢，而且与允许处理多个 cURL 的 curl_multi_init 相比没有任何优势并行处理。
参见:
我们的目标是实现我们自己的多file_get_contents版本
多个 file_get_contents 示例
// My Storage
$s = new Storage();
// Threads Storage
$ts = array();
// Total Threads same as total pages
$pages = 100;
// Porpulate Threads (Don't Start Yet)
$i = 0;
while($i ++ < $pages) {
$ts[] = new Process($s, $i);
}
// Start the timer
$start = microtime(true);
// Lets start all our Threads
foreach($ts as $t) {
$t->start();
}
// Wait for all threads to compleate
foreach($ts as $t) {
$t->join();
}
printf("\n\nFound %s in %d pages", number_format($s->total), $pages);
printf("\nFinished %0.3f sec", microtime(true) - $start);
输出
php a.php
3:01:37: 3548 #START {"page":1}
3:01:37: 7064 #START {"page":2}
3:01:37: 10908 #START {"page":3}
3:01:37: 10424 #START {"page":4}
3:01:37: 11472 #START {"page":5}
3:01:37: 3876 #START {"page":6}
3:01:37: 7276 #START {"page":7}
3:01:37: 11484 #START {"page":8}
3:01:37: 932 #START {"page":9}
3:01:37: 11492 #START {"page":10}
3:01:37: 11500 #START {"page":11}
3:01:37: 11508 #START {"page":12}
3:01:37: 11504 #START {"page":13}
3:01:37: 11512 #START {"page":14}
3:01:37: 11516 #START {"page":15}
3:01:37: 11520 #START {"page":16}
3:01:37: 11524 #START {"page":17}
3:01:37: 11528 #START {"page":18}
3:01:37: 10816 #START {"page":19}
3:01:37: 7280 #START {"page":20}
3:01:37: 11556 #START {"page":21}
3:01:37: 11560 #START {"page":22}
3:01:37: 11564 #START {"page":23}
3:01:37: 11612 #START {"page":24}
3:01:37: 11616 #START {"page":25}
3:01:37: 11600 #START {"page":26}
3:01:37: 11608 #START {"page":27}
3:01:37: 11568 #START {"page":28}
3:01:37: 11452 #START {"page":29}
3:01:38: 11624 #START {"page":30}
3:01:38: 11628 #START {"page":31}
3:01:38: 11632 #START {"page":32}
3:01:38: 11636 #START {"page":33}
3:01:38: 11644 #START {"page":34}
3:01:38: 11648 #START {"page":35}
3:01:38: 11652 #START {"page":36}
3:01:38: 11656 #START {"page":37}
3:01:38: 11660 #START {"page":38}
3:01:38: 11664 #START {"page":39}
3:01:38: 11668 #START {"page":40}
3:01:38: 11672 #START {"page":41}
3:01:38: 11676 #START {"page":42}
3:01:38: 11680 #START {"page":43}
3:01:38: 11684 #START {"page":44}
3:01:38: 11688 #START {"page":45}
3:01:38: 11692 #START {"page":46}
3:01:38: 11696 #START {"page":47}
3:01:38: 11700 #START {"page":48}
3:01:38: 11704 #START {"page":49}
3:01:38: 11712 #START {"page":50}
3:01:38: 11708 #START {"page":51}
3:01:38: 11716 #START {"page":52}
3:01:38: 11720 #START {"page":53}
3:01:38: 11724 #START {"page":54}
3:01:38: 11728 #START {"page":55}
3:01:38: 11732 #START {"page":56}
3:01:38: 11736 #START {"page":57}
3:01:38: 11740 #START {"page":58}
3:01:38: 11744 #START {"page":59}
3:01:38: 11748 #START {"page":60}
3:01:38: 11752 #START {"page":61}
3:01:38: 11756 #START {"page":62}
3:01:38: 11760 #START {"page":63}
3:01:38: 11764 #START {"page":64}
3:01:38: 11768 #START {"page":65}
3:01:38: 11772 #START {"page":66}
3:01:38: 11776 #START {"page":67}
3:01:38: 11780 #START {"page":68}
3:01:38: 11784 #START {"page":69}
3:01:38: 11788 #START {"page":70}
3:01:38: 11792 #START {"page":71}
3:01:38: 11796 #START {"page":72}
3:01:38: 11800 #START {"page":73}
3:01:38: 11804 #START {"page":74}
3:01:38: 11808 #START {"page":75}
3:01:38: 11812 #START {"page":76}
3:01:38: 11816 #START {"page":77}
3:01:38: 11820 #START {"page":78}
3:01:38: 11824 #START {"page":79}
3:01:38: 11828 #START {"page":80}
3:01:38: 11832 #START {"page":81}
3:01:38: 11836 #START {"page":82}
3:01:38: 11840 #START {"page":83}
3:01:38: 11844 #START {"page":84}
3:01:38: 11848 #START {"page":85}
3:01:38: 11852 #START {"page":86}
3:01:38: 11856 #START {"page":87}
3:01:38: 11860 #START {"page":88}
3:01:38: 11864 #START {"page":89}
3:01:38: 11868 #START {"page":90}
3:01:38: 11872 #START {"page":91}
3:01:38: 11876 #START {"page":92}
3:01:38: 11880 #START {"page":93}
3:01:38: 11884 #START {"page":94}
3:01:38: 11888 #START {"page":95}
3:01:38: 11892 #START {"page":96}
3:01:38: 11896 #START {"page":97}
3:01:38: 11900 #START {"page":98}
3:01:38: 11904 #START {"page":99}
3:01:38: 11908 #START {"page":100}
3:01:38: 11508 #END {"page":12,"byte":1141,"count":155839}
3:01:38: 10424 #END {"page":4,"byte":1201,"count":553595}
3:01:38: 11516 #END {"page":15,"byte":1204,"count":119612}
3:01:38: 3548 #END {"page":1,"byte":1208,"count":6737525}
3:01:38: 11484 #END {"page":8,"byte":1160,"count":257021}
3:01:38: 11472 #END {"page":5,"byte":1175,"count":446411}
3:01:38: 10908 #END {"page":3,"byte":1222,"count":787301}
3:01:38: 11492 #END {"page":10,"byte":1175,"count":193958}
3:01:38: 11504 #END {"page":13,"byte":1130,"count":141450}
3:01:38: 11528 #END {"page":18,"byte":1102,"count":95511}
3:01:38: 11524 #END {"page":17,"byte":1147,"count":102727}
3:01:38: 11560 #END {"page":22,"byte":1111,"count":73536}
3:01:38: 11556 #END {"page":21,"byte":1101,"count":78097}
3:01:38: 11500 #END {"page":11,"byte":1201,"count":172820}
3:01:38: 932 #END {"page":9,"byte":1159,"count":222922}
3:01:38: 11520 #END {"page":16,"byte":1135,"count":110510}
3:01:38: 7064 #END {"page":2,"byte":1165,"count":1264444}
3:01:38: 11512 #END {"page":14,"byte":1123,"count":129721}
3:01:38: 11612 #END {"page":24,"byte":1115,"count":65012}
3:01:38: 11600 #END {"page":26,"byte":1134,"count":58928}
3:01:38: 7276 #END {"page":7,"byte":1189,"count":301469}
3:01:38: 10816 #END {"page":19,"byte":1120,"count":89609}
3:01:38: 11616 #END {"page":25,"byte":1052,"count":61793}
3:01:38: 3876 #END {"page":6,"byte":1188,"count":362101}
3:01:38: 7280 #END {"page":20,"byte":1079,"count":83632}
3:01:38: 11564 #END {"page":23,"byte":1076,"count":68909}
3:01:38: 11632 #END {"page":32,"byte":1095,"count":44013}
3:01:38: 11652 #END {"page":36,"byte":1042,"count":37185}
3:01:38: 11452 #END {"page":29,"byte":1097,"count":50532}
3:01:38: 11636 #END {"page":33,"byte":1097,"count":42148}
3:01:38: 11644 #END {"page":34,"byte":1124,"count":40236}
3:01:38: 11664 #END {"page":39,"byte":1078,"count":32792}
3:01:38: 11668 #END {"page":40,"byte":1017,"count":31487}
3:01:38: 11608 #END {"page":27,"byte":1117,"count":55561}
3:01:38: 11628 #END {"page":31,"byte":1076,"count":46133}
3:01:38: 11624 #END {"page":30,"byte":1111,"count":48265}
3:01:38: 11568 #END {"page":28,"byte":1076,"count":52851}
3:01:38: 11656 #END {"page":37,"byte":1068,"count":35590}
3:01:38: 11688 #END {"page":45,"byte":1062,"count":26060}
3:01:38: 11680 #END {"page":43,"byte":1081,"count":28013}
3:01:38: 11672 #END {"page":41,"byte":1086,"count":30320}
3:01:38: 11724 #END {"page":54,"byte":1060,"count":19900}
3:01:38: 11716 #END {"page":52,"byte":1069,"count":21079}
3:01:38: 11732 #END {"page":56,"byte":1038,"count":18748}
3:01:38: 11692 #END {"page":46,"byte":1033,"count":25230}
3:01:38: 11696 #END {"page":47,"byte":1098,"count":24469}
3:01:38: 11728 #END {"page":55,"byte":1003,"count":19353}
3:01:38: 11648 #END {"page":35,"byte":1105,"count":38651}
3:01:38: 11660 #END {"page":38,"byte":1075,"count":34037}
3:01:38: 11700 #END {"page":48,"byte":1059,"count":23725}
3:01:39: 11720 #END {"page":53,"byte":1028,"count":20463}
3:01:39: 11704 #END {"page":49,"byte":1006,"count":22966}
3:01:39: 11712 #END {"page":50,"byte":988,"count":22369}
3:01:39: 11676 #END {"page":42,"byte":1113,"count":29144}
3:01:39: 11748 #END {"page":60,"byte":1054,"count":17002}
3:01:39: 11684 #END {"page":44,"byte":1041,"count":26999}
3:01:39: 11756 #END {"page":62,"byte":1024,"count":16165}
3:01:39: 11760 #END {"page":63,"byte":1036,"count":15814}
3:01:39: 11740 #END {"page":58,"byte":1075,"count":17833}
3:01:39: 11736 #END {"page":57,"byte":1064,"count":18293}
3:01:39: 11752 #END {"page":61,"byte":1077,"count":16607}
3:01:39: 11708 #END {"page":51,"byte":1045,"count":21668}
3:01:39: 11768 #END {"page":65,"byte":1041,"count":15021}
3:01:39: 11764 #END {"page":64,"byte":1063,"count":15405}
3:01:39: 11744 #END {"page":59,"byte":1052,"count":17394}
3:01:39: 11800 #END {"page":73,"byte":1025,"count":12361}
3:01:39: 11792 #END {"page":71,"byte":1053,"count":13051}
3:01:39: 11796 #END {"page":72,"byte":1092,"count":12721}
3:01:39: 11784 #END {"page":69,"byte":1031,"count":13677}
3:01:39: 11780 #END {"page":68,"byte":1019,"count":13967}
3:01:39: 11772 #END {"page":66,"byte":1068,"count":14644}
3:01:39: 11816 #END {"page":77,"byte":1045,"count":11185}
3:01:39: 11804 #END {"page":74,"byte":1062,"count":12071}
3:01:39: 11824 #END {"page":79,"byte":1047,"count":10719}
3:01:39: 11820 #END {"page":78,"byte":1035,"count":10940}
3:01:39: 11788 #END {"page":70,"byte":987,"count":13354}
3:01:39: 11776 #END {"page":67,"byte":1036,"count":14278}
3:01:39: 11828 #END {"page":80,"byte":1013,"count":10519}
3:01:39: 11832 #END {"page":81,"byte":1052,"count":10318}
3:01:39: 11812 #END {"page":76,"byte":991,"count":11465}
3:01:39: 11808 #END {"page":75,"byte":1043,"count":11769}
3:01:39: 11860 #END {"page":88,"byte":1018,"count":8991}
3:01:39: 11852 #END {"page":86,"byte":971,"count":9362}
3:01:39: 11868 #END {"page":90,"byte":1006,"count":8641}
3:01:39: 11840 #END {"page":83,"byte":1026,"count":9922}
3:01:39: 11872 #END {"page":91,"byte":980,"count":8464}
3:01:39: 11892 #END {"page":96,"byte":936,"count":7727}
3:01:39: 11836 #END {"page":82,"byte":1052,"count":10117}
3:01:39: 11844 #END {"page":84,"byte":973,"count":9739}
3:01:39: 11864 #END {"page":89,"byte":1033,"count":8821}
3:01:39: 11856 #END {"page":87,"byte":994,"count":9169}
3:01:39: 11848 #END {"page":85,"byte":1040,"count":9544}
3:01:39: 11896 #END {"page":97,"byte":988,"count":7562}
3:01:39: 11876 #END {"page":92,"byte":1003,"count":8294}
3:01:39: 11888 #END {"page":95,"byte":995,"count":7860}
3:01:39: 11880 #END {"page":93,"byte":1052,"count":8143}
3:01:39: 11900 #END {"page":98,"byte":977,"count":7418}
3:01:39: 11904 #END {"page":99,"byte":999,"count":7270}
3:01:39: 11884 #END {"page":94,"byte":931,"count":8002}
3:01:39: 11908 #END {"page":100,"byte":977,"count":7144}
Found 14,075,927 in 100 pages
Finished 1.489 sec
耗时
Found 14,075,927 in 100 pages
Finished 1.489 sec
使用的类
class Process extends Thread {
public function __construct($storage, $page) {
$this->storage = $storage;
$this->page = $page;
// $this->start();
}
public function run() {
$format = "%s: %1u %s\t%s\n";
$formatTime = "g:i:s";
$sleep = mt_rand(0, 1); // Just for Demo
printf($format, date($formatTime), $this->getThreadId(), "#START", "{\"page\":$this->page}");
// Do something useful
$data = file_get_contents(sprintf("http://api.stackoverflow.com/1.1/tags?pagesize=100&page=%s", $this->page));
// Decode the Data from API
$json = json_decode(gzdecode($data));
// Lets Build A profile
$profile = array();
$profile['page'] = $this->page;
$profile['byte'] = strlen($data);
// Do more work
$profile['count'] = array_sum(array_map(function ($v) {
return $v->count;
}, $json->tags));
$this->storage->total = bcadd($this->storage->total, $profile['count']);
// Print Information
printf($format, date($formatTime), $this->getThreadId(), "#END\t", json_encode($profile));
}
}
class Storage extends Stackable {
public $total = 0;
public function run() {
}
}
结论
file_get_contents 是否仅在 1.489 秒 内通过我糟糕的连接获取了 100 页。 是的是的。在我的实时服务器上测试了相同的代码，我用了不到 0.939 秒 来获取 200 页面。
您的应用程序可以在很多方面变得更快，您只需在正确的地方使用正确的技术。
关于php - pcntl 多次运行相同的代码，需要帮助，我们在Stack Overflow上找到一个类似的问题： https://stackoverflow.com/questions/16383803/
总的来说，我对ruby还比较陌生，我正在为我正在创建的对象编写一些rspec测试用例。许多测试用例都非常基础，我只是想确保正确填充和返回值。我想知道是否有办法使用循环结构来执行此操作。不必为我要测试的每个方法都设置一个assertEquals。例如:describeitem,"TestingtheItem"doit"willhaveanullvaluetostart"doitem=Item.new#HereIcoulddotheitem.name.shouldbe_nil#thenIcoulddoitem.category.shouldbe_nilendend但我想要一些方法来使用
当我使用Bundler时，是否需要在我的Gemfile中将其列为依赖项？毕竟，我的代码中有些地方需要它。例如，当我进行Bundler设置时:require"bundler/setup" 最佳答案 没有。您可以尝试，但首先您必须用鞋带将自己抬离地面。 关于ruby-我需要将Bundler本身添加到Gemfile中吗？，我们在StackOverflow上找到一个类似的问题： https://stackoverflow.com/questions/4758609/
如何在buildr项目中使用Ruby？我在很多不同的项目中使用过Ruby、JRuby、Java和Clojure。我目前正在使用我的标准Ruby开发一个模拟应用程序，我想尝试使用Clojure后端(我确实喜欢功能代码)以及JRubygui和测试套件。我还可以看到在未来的不同项目中使用Scala作为后端。我想我要为我的项目尝试一下buildr(http://buildr.apache.org/)，但我注意到buildr似乎没有设置为在项目中使用JRuby代码本身!这看起来有点傻，因为该工具旨在统一通用的JVM语言并且是在ruby中构建的。除了将输出的jar包含在一个独特的、仅限ruby
我的代码目前看起来像这样numbers=[1,2,3,4,5]defpop_threepop=[]3.times{pop有没有办法在一行中完成pop_three方法中的内容？我基本上想做类似numbers.slice(0,3)的事情，但要删除切片中的数组项。嗯...嗯，我想我刚刚意识到我可以试试slice! 最佳答案 是numbers.pop(3)或者numbers.shift(3)如果你想要另一边。 关于ruby-多次弹出/移动ruby数组，我们在StackOverflow上找到一
在rails源中:https://github.com/rails/rails/blob/master/activesupport/lib/active_support/lazy_load_hooks.rb可以看到以下内容@load_hooks=Hash.new{|h,k|h[k]=[]}在IRB中，它只是初始化一个空哈希。和做有什么区别@load_hooks=Hash.new 最佳答案 查看rubydocumentationforHashnew→new_hashclicktotogglesourcenew(obj)→new_has
在选择我想要运行操作的频率时，唯一的选项是“每天”、“每小时”和“每10分钟”。谢谢!我想为我的Rails3.1应用程序运行调度程序。 最佳答案 这不是一个优雅的解决方案，但您可以安排它每天运行，并在实际开始工作之前检查日期是否为当月的第一天。 关于ruby-如何每月在Heroku运行一次Scheduler插件？，我们在StackOverflow上找到一个类似的问题： https://stackoverflow.com/questions/8692687/
exe应该在我打开页面时运行。异步进程需要运行。有什么方法可以在ruby中使用两个参数异步运行exe吗？我已经尝试过ruby命令-system()、exec()但它正在等待过程完成。我需要用参数启动exe，无需等待进程完成是否有任何rubygems会支持我的问题？ 最佳答案 您可以使用Process.spawn和Process.wait2:pid=Process.spawn'your.exe','--option'#Later...pid,status=Process.wait2pid您的程序将作为解释器的子进程执行。除
我尝试运行2.x应用程序。我使用rvm并为此应用程序设置其他版本的ruby:$rvmuseree-1.8.7-head我尝试运行服务器，然后出现很多错误:$script/serverNOTE:Gem.source_indexisdeprecated,useSpecification.Itwillberemovedonorafter2011-11-01.Gem.source_indexcalledfrom/Users/serg/rails_projects_terminal/work_proj/spohelp/config/../vendor/rails/railties/lib/r
我注意到像bundler这样的项目在每个specfile中执行requirespec_helper我还注意到rspec使用选项--require，它允许您在引导rspec时要求一个文件。您还可以将其添加到.rspec文件中，因此只要您运行不带参数的rspec就会添加它。使用上述方法有什么缺点可以解释为什么像bundler这样的项目选择在每个规范文件中都需要spec_helper吗？ 最佳答案 我不在Bundler上工作，所以我不能直接谈论他们的做法。并非所有项目都checkin.rspec文件。原因是这个文件，通常按照当前的惯例，只
我实际上是在尝试使用RVM在我的OSX10.7.5上更新ruby，并在输入以下命令后:rvminstallruby我得到了以下回复:Searchingforbinaryrubies,thismighttakesometime.Checkingrequirementsforosx.Installingrequirementsforosx.Updatingsystem.......Errorrunning'requirements_osx_brew_update_systemruby-2.0.0-p247',pleaseread/Users/username/.rvm/log/138121