|
1
|
|
|
<?php |
|
2
|
|
|
// +---------------------------------------------------------------------- |
|
3
|
|
|
// | PHPSpider [ A PHP Framework For Crawler ] |
|
4
|
|
|
// +---------------------------------------------------------------------- |
|
5
|
|
|
// | Copyright (c) 2006-2014 https://doc.phpspider.org All rights reserved. |
|
6
|
|
|
// +---------------------------------------------------------------------- |
|
7
|
|
|
// | Licensed ( http://www.apache.org/licenses/LICENSE-2.0 ) |
|
8
|
|
|
// +---------------------------------------------------------------------- |
|
9
|
|
|
// | Author: Seatle Yang <[email protected]> |
|
10
|
|
|
// +---------------------------------------------------------------------- |
|
11
|
|
|
|
|
12
|
|
|
//---------------------------------- |
|
13
|
|
|
// Worker多进程操作类 |
|
14
|
|
|
//---------------------------------- |
|
15
|
|
|
|
|
16
|
|
|
class worker |
|
17
|
|
|
{ |
|
18
|
|
|
// worker进程数 |
|
19
|
|
|
public $count = 0; |
|
20
|
|
|
// worker id,worker进程从1开始,0被master进程所使用 |
|
21
|
|
|
public $worker_id = 0; |
|
22
|
|
|
// worker 进程ID |
|
23
|
|
|
public $worker_pid = 0; |
|
24
|
|
|
// 进程用户 |
|
25
|
|
|
public $user = ''; |
|
26
|
|
|
// 进程名 |
|
27
|
|
|
public $title = ''; |
|
28
|
|
|
// 每个进程是否只运行一次 |
|
29
|
|
|
public $run_once = true; |
|
30
|
|
|
// 是否输出日志 |
|
31
|
|
|
public $log_show = false; |
|
32
|
|
|
// master进程启动回调 |
|
33
|
|
|
public $on_start = false; |
|
34
|
|
|
// master进程停止回调 |
|
35
|
|
|
public $on_stop = false; |
|
36
|
|
|
// worker进程启动回调 |
|
37
|
|
|
public $on_worker_start = false; |
|
38
|
|
|
// worker进程停止回调 |
|
39
|
|
|
public $on_worker_stop = false; |
|
40
|
|
|
// master进程ID |
|
41
|
|
|
protected static $_master_pid = 0; |
|
42
|
|
|
// worker进程ID |
|
43
|
|
|
protected static $_worker_pids = array(); |
|
44
|
|
|
// master、worker进程启动时间 |
|
45
|
|
|
public $time_start = 0; |
|
46
|
|
|
// master、worker进程运行状态 [starting|running|shutdown|reload] |
|
47
|
|
|
protected static $_status = "starting"; |
|
48
|
|
|
|
|
49
|
|
|
|
|
50
|
|
|
public function __construct() |
|
51
|
|
|
{ |
|
52
|
|
|
self::$_master_pid = posix_getpid(); |
|
53
|
|
|
// 产生时钟云,添加后父进程才可以收到信号 |
|
54
|
|
|
declare(ticks = 1); |
|
55
|
|
|
$this->install_signal(); |
|
|
|
|
|
|
56
|
|
|
} |
|
57
|
|
|
|
|
58
|
|
|
/** |
|
59
|
|
|
* 安装信号处理函数 |
|
60
|
|
|
* @return void |
|
61
|
|
|
*/ |
|
62
|
|
|
protected function install_signal() |
|
63
|
|
|
{ |
|
64
|
|
|
// stop |
|
65
|
|
|
pcntl_signal(SIGINT, array($this, 'signal_handler'), false); |
|
66
|
|
|
// reload |
|
67
|
|
|
pcntl_signal(SIGUSR1, array($this, 'signal_handler'), false); |
|
68
|
|
|
// status |
|
69
|
|
|
pcntl_signal(SIGUSR2, array($this, 'signal_handler'), false); |
|
70
|
|
|
// ignore |
|
71
|
|
|
pcntl_signal(SIGPIPE, SIG_IGN, false); |
|
72
|
|
|
// install signal handler for dead kids |
|
73
|
|
|
// pcntl_signal(SIGCHLD, array($this, 'signal_handler')); |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
|
|
/** |
|
77
|
|
|
* 卸载信号处理函数 |
|
78
|
|
|
* @return void |
|
79
|
|
|
*/ |
|
80
|
|
|
protected function uninstall_signal() |
|
81
|
|
|
{ |
|
82
|
|
|
// uninstall stop signal handler |
|
83
|
|
|
pcntl_signal(SIGINT, SIG_IGN, false); |
|
84
|
|
|
// uninstall reload signal handler |
|
85
|
|
|
pcntl_signal(SIGUSR1, SIG_IGN, false); |
|
86
|
|
|
// uninstall status signal handler |
|
87
|
|
|
pcntl_signal(SIGUSR2, SIG_IGN, false); |
|
88
|
|
|
} |
|
89
|
|
|
|
|
90
|
|
|
/** |
|
91
|
|
|
* 信号处理函数,会被其他类调用到,所以要设置为public |
|
92
|
|
|
* @param int $signal |
|
93
|
|
|
*/ |
|
94
|
|
|
public function signal_handler($signal) { |
|
95
|
|
|
switch ($signal) { |
|
96
|
|
|
// stop 2 |
|
97
|
|
|
case SIGINT: |
|
98
|
|
|
// master进程和worker进程都会调用 |
|
99
|
|
|
$this->stop_all(); |
|
|
|
|
|
|
100
|
|
|
break; |
|
101
|
|
|
// reload 30 |
|
102
|
|
|
case SIGUSR1: |
|
103
|
|
|
echo "reload\n"; |
|
104
|
|
|
break; |
|
105
|
|
|
// show status 31 |
|
106
|
|
|
case SIGUSR2: |
|
107
|
|
|
echo "status\n"; |
|
108
|
|
|
break; |
|
109
|
|
|
} |
|
110
|
|
|
} |
|
111
|
|
|
|
|
112
|
|
|
/** |
|
113
|
|
|
* 运行worker实例 |
|
114
|
|
|
*/ |
|
115
|
|
|
public function run() |
|
116
|
|
|
{ |
|
117
|
|
|
$this->time_start = microtime(true); |
|
118
|
|
|
$this->worker_id = 0; |
|
|
|
|
|
|
119
|
|
|
$this->worker_pid = posix_getpid(); |
|
|
|
|
|
|
120
|
|
|
$this->set_process_title($this->title); |
|
|
|
|
|
|
121
|
|
|
|
|
122
|
|
|
// 这里赋值,worker进程也会克隆到 |
|
123
|
|
|
if ($this->log_show) |
|
|
|
|
|
|
124
|
|
|
{ |
|
125
|
|
|
log::$log_show = true; |
|
126
|
|
|
} |
|
127
|
|
|
|
|
128
|
|
|
if ($this->on_start) |
|
|
|
|
|
|
129
|
|
|
{ |
|
130
|
|
|
call_user_func($this->on_start, $this); |
|
131
|
|
|
} |
|
132
|
|
|
|
|
133
|
|
|
// worker进程从1开始,0被master进程所使用 |
|
134
|
|
|
for ($i = 1; $i <= $this->count; $i++) |
|
|
|
|
|
|
135
|
|
|
{ |
|
136
|
|
|
$this->fork_one_worker($i); |
|
|
|
|
|
|
137
|
|
|
} |
|
138
|
|
|
$this->monitor_workers(); |
|
|
|
|
|
|
139
|
|
|
} |
|
140
|
|
|
|
|
141
|
|
|
/** |
|
142
|
|
|
* 创建一个子进程 |
|
143
|
|
|
* @param Worker $worker |
|
144
|
|
|
* @throws Exception |
|
145
|
|
|
*/ |
|
146
|
|
|
public function fork_one_worker($worker_id) |
|
147
|
|
|
{ |
|
148
|
|
|
//$sockets = stream_socket_pair(STREAM_PF_UNIX, STREAM_SOCK_STREAM, STREAM_IPPROTO_IP); |
|
149
|
|
|
$pid = pcntl_fork(); |
|
150
|
|
|
|
|
151
|
|
|
// 主进程记录子进程pid |
|
152
|
|
|
if($pid > 0) |
|
153
|
|
|
{ |
|
154
|
|
|
self::$_worker_pids[$worker_id] = $pid; |
|
|
|
|
|
|
155
|
|
|
} |
|
156
|
|
|
// 子进程运行 |
|
157
|
|
|
elseif(0 === $pid) |
|
158
|
|
|
{ |
|
159
|
|
|
$this->time_start = microtime(true); |
|
160
|
|
|
$this->worker_id = $worker_id; |
|
|
|
|
|
|
161
|
|
|
$this->worker_pid = posix_getpid(); |
|
|
|
|
|
|
162
|
|
|
$this->set_process_title($this->title); |
|
|
|
|
|
|
163
|
|
|
$this->set_process_user($this->user); |
|
|
|
|
|
|
164
|
|
|
// 清空master进程克隆过来的worker进程ID |
|
165
|
|
|
self::$_worker_pids = array(); |
|
166
|
|
|
//$this->uninstall_signal(); |
|
167
|
|
|
|
|
168
|
|
|
// 设置worker进程的运行状态为运行中 |
|
169
|
|
|
self::$_status = "running"; |
|
170
|
|
|
|
|
171
|
|
|
// 注册进程退出回调,用来检查是否有错误(子进程里面注册) |
|
172
|
|
|
register_shutdown_function(array($this, 'check_errors')); |
|
173
|
|
|
|
|
174
|
|
|
// 如果设置了worker进程启动回调函数 |
|
175
|
|
|
if ($this->on_worker_start) |
|
|
|
|
|
|
176
|
|
|
{ |
|
177
|
|
|
call_user_func($this->on_worker_start, $this); |
|
178
|
|
|
} |
|
179
|
|
|
|
|
180
|
|
|
// 停止当前worker实例 |
|
181
|
|
|
$this->stop(); |
|
|
|
|
|
|
182
|
|
|
// 这里用0表示正常退出 |
|
183
|
|
|
exit(0); |
|
|
|
|
|
|
184
|
|
|
} |
|
185
|
|
|
else |
|
186
|
|
|
{ |
|
187
|
|
|
log::add("fork one worker fail", "Error"); |
|
188
|
|
|
exit; |
|
|
|
|
|
|
189
|
|
|
} |
|
190
|
|
|
} |
|
191
|
|
|
|
|
192
|
|
|
/** |
|
193
|
|
|
* 尝试设置运行当前进程的用户 |
|
194
|
|
|
* |
|
195
|
|
|
* @param $user_name |
|
196
|
|
|
*/ |
|
197
|
|
|
protected static function set_process_user($user_name) |
|
198
|
|
|
{ |
|
199
|
|
|
// 用户名为空 或者 当前用户不是root用户 |
|
200
|
|
|
if(empty($user_name) || posix_getuid() !== 0) |
|
201
|
|
|
{ |
|
202
|
|
|
return; |
|
203
|
|
|
} |
|
204
|
|
|
$user_info = posix_getpwnam($user_name); |
|
205
|
|
|
if($user_info['uid'] != posix_getuid() || $user_info['gid'] != posix_getgid()) |
|
206
|
|
|
{ |
|
207
|
|
|
if(!posix_setgid($user_info['gid']) || !posix_setuid($user_info['uid'])) |
|
208
|
|
|
{ |
|
209
|
|
|
log::add('Can not run woker as '.$user_name." , You shuld be root", "Error"); |
|
210
|
|
|
} |
|
211
|
|
|
} |
|
212
|
|
|
} |
|
213
|
|
|
|
|
214
|
|
|
/** |
|
215
|
|
|
* 设置当前进程的名称,在ps aux命令中有用 |
|
216
|
|
|
* 注意 需要php>=5.5或者安装了protitle扩展 |
|
217
|
|
|
* @param string $title |
|
218
|
|
|
* @return void |
|
219
|
|
|
*/ |
|
220
|
|
|
protected function set_process_title($title) |
|
221
|
|
|
{ |
|
222
|
|
|
if (!empty($title)) |
|
223
|
|
|
{ |
|
224
|
|
|
// 需要扩展 |
|
225
|
|
|
if(extension_loaded('proctitle') && function_exists('setproctitle')) |
|
226
|
|
|
{ |
|
227
|
|
|
@setproctitle($title); |
|
228
|
|
|
} |
|
229
|
|
|
// >=php 5.5 |
|
230
|
|
|
elseif (function_exists('cli_set_process_title')) |
|
231
|
|
|
{ |
|
232
|
|
|
cli_set_process_title($title); |
|
233
|
|
|
} |
|
234
|
|
|
} |
|
235
|
|
|
} |
|
236
|
|
|
|
|
237
|
|
|
/** |
|
238
|
|
|
* 监控所有子进程的退出事件及退出码 |
|
239
|
|
|
* @return void |
|
240
|
|
|
*/ |
|
241
|
|
|
public function monitor_workers() |
|
242
|
|
|
{ |
|
243
|
|
|
// 设置master进程的运行状态为运行中 |
|
244
|
|
|
self::$_status = "running"; |
|
245
|
|
|
while(1) |
|
246
|
|
|
{ |
|
247
|
|
|
// pcntl_signal_dispatch 子进程无法接受到信号 |
|
248
|
|
|
// 如果有信号到来,尝试触发信号处理函数 |
|
249
|
|
|
//pcntl_signal_dispatch(); |
|
250
|
|
|
// 挂起进程,直到有子进程退出或者被信号打断 |
|
251
|
|
|
$status = 0; |
|
252
|
|
|
$pid = pcntl_wait($status, WUNTRACED); |
|
253
|
|
|
// 如果有信号到来,尝试触发信号处理函数 |
|
254
|
|
|
//pcntl_signal_dispatch(); |
|
255
|
|
|
|
|
256
|
|
|
// 子进程退出信号 |
|
257
|
|
|
if($pid > 0) |
|
258
|
|
|
{ |
|
259
|
|
|
//echo "worker[".$pid."] stop\n"; |
|
260
|
|
|
//$this->stop(); |
|
261
|
|
|
|
|
262
|
|
|
// 如果不是正常退出,是被kill等杀掉的 |
|
263
|
|
|
if($status !== 0) |
|
264
|
|
|
{ |
|
265
|
|
|
log::add("worker {$pid} exit with status $status", "Warning"); |
|
266
|
|
|
} |
|
267
|
|
|
|
|
268
|
|
|
// key 和 value 互换 |
|
269
|
|
|
$worker_pids = array_flip(self::$_worker_pids); |
|
|
|
|
|
|
270
|
|
|
// 通过 pid 得到 worker_id |
|
271
|
|
|
$worker_id = $worker_pids[$pid]; |
|
272
|
|
|
// 这里不unset掉,是为了进程重启 |
|
273
|
|
|
self::$_worker_pids[$worker_id] = 0; |
|
274
|
|
|
//unset(self::$_worker_pids[$pid]); |
|
275
|
|
|
|
|
276
|
|
|
// 再生成一个worker |
|
277
|
|
|
if (!$this->run_once) |
|
|
|
|
|
|
278
|
|
|
{ |
|
279
|
|
|
$this->fork_one_worker($worker_id); |
|
280
|
|
|
} |
|
281
|
|
|
|
|
282
|
|
|
// 如果所有子进程都退出了,触发主进程退出函数 |
|
283
|
|
|
$all_worker_stop = true; |
|
284
|
|
|
foreach (self::$_worker_pids as $_worker_pid) |
|
285
|
|
|
{ |
|
286
|
|
|
// 只要有一个worker进程还存在进程ID,就不算退出 |
|
287
|
|
|
if ($_worker_pid != 0) |
|
288
|
|
|
{ |
|
289
|
|
|
$all_worker_stop = false; |
|
290
|
|
|
} |
|
291
|
|
|
} |
|
292
|
|
|
if ($all_worker_stop) |
|
293
|
|
|
{ |
|
294
|
|
|
if ($this->on_stop) |
|
|
|
|
|
|
295
|
|
|
{ |
|
296
|
|
|
call_user_func($this->on_stop, $this); |
|
297
|
|
|
} |
|
298
|
|
|
exit(0); |
|
|
|
|
|
|
299
|
|
|
} |
|
300
|
|
|
} |
|
301
|
|
|
// 其他信号 |
|
302
|
|
|
else |
|
303
|
|
|
{ |
|
304
|
|
|
// worker进程接受到master进行信号退出的,会到这里来 |
|
305
|
|
|
if ($this->on_stop) |
|
306
|
|
|
{ |
|
307
|
|
|
call_user_func($this->on_stop, $this); |
|
308
|
|
|
} |
|
309
|
|
|
exit(0); |
|
|
|
|
|
|
310
|
|
|
} |
|
311
|
|
|
} |
|
312
|
|
|
} |
|
313
|
|
|
|
|
314
|
|
|
/** |
|
315
|
|
|
* 执行关闭流程(所有进程) |
|
316
|
|
|
* 事件触发,非正常程序执行完毕 |
|
317
|
|
|
* @return void |
|
318
|
|
|
*/ |
|
319
|
|
|
public function stop_all() |
|
320
|
|
|
{ |
|
321
|
|
|
// 设置master、worker进程的运行状态为关闭状态 |
|
322
|
|
|
self::$_status = "shutdown"; |
|
323
|
|
|
// master进程 |
|
324
|
|
|
if(self::$_master_pid === posix_getpid()) |
|
|
|
|
|
|
325
|
|
|
{ |
|
326
|
|
|
// 循环给worker进程发送关闭信号 |
|
327
|
|
|
foreach (self::$_worker_pids as $worker_pid) |
|
|
|
|
|
|
328
|
|
|
{ |
|
329
|
|
|
posix_kill($worker_pid, SIGINT); |
|
330
|
|
|
} |
|
331
|
|
|
} |
|
332
|
|
|
// worker进程 |
|
333
|
|
|
else |
|
334
|
|
|
{ |
|
335
|
|
|
// 接收到master进程发送的关闭信号之后退出,这里应该考虑业务的完整性,不能强行exit |
|
336
|
|
|
$this->stop(); |
|
337
|
|
|
exit(0); |
|
|
|
|
|
|
338
|
|
|
} |
|
339
|
|
|
} |
|
340
|
|
|
|
|
341
|
|
|
/** |
|
342
|
|
|
* 停止当前worker实例 |
|
343
|
|
|
* 正常运行结束和接受信号退出,都会调用这个方法 |
|
344
|
|
|
* @return void |
|
345
|
|
|
*/ |
|
346
|
|
|
public function stop() |
|
347
|
|
|
{ |
|
348
|
|
|
if ($this->on_worker_stop) |
|
|
|
|
|
|
349
|
|
|
{ |
|
350
|
|
|
call_user_func($this->on_worker_stop, $this); |
|
351
|
|
|
} |
|
352
|
|
|
// 设置worker进程的运行状态为关闭 |
|
353
|
|
|
self::$_status = "shutdown"; |
|
354
|
|
|
} |
|
355
|
|
|
|
|
356
|
|
|
/** |
|
357
|
|
|
* 检查错误,PHP exit之前会执行 |
|
358
|
|
|
* @return void |
|
359
|
|
|
*/ |
|
360
|
|
|
public function check_errors() |
|
361
|
|
|
{ |
|
362
|
|
|
// 如果当前worker进程不是正常退出 |
|
363
|
|
|
if(self::$_status != "shutdown") |
|
|
|
|
|
|
364
|
|
|
{ |
|
365
|
|
|
$error_msg = "WORKER EXIT UNEXPECTED "; |
|
366
|
|
|
$errors = error_get_last(); |
|
367
|
|
|
if($errors && ($errors['type'] === E_ERROR || |
|
|
|
|
|
|
368
|
|
|
$errors['type'] === E_PARSE || |
|
369
|
|
|
$errors['type'] === E_CORE_ERROR || |
|
370
|
|
|
$errors['type'] === E_COMPILE_ERROR || |
|
371
|
|
|
$errors['type'] === E_RECOVERABLE_ERROR )) |
|
372
|
|
|
{ |
|
373
|
|
|
$error_msg .= $this->get_error_type($errors['type']) . " {$errors['message']} in {$errors['file']} on line {$errors['line']}"; |
|
|
|
|
|
|
374
|
|
|
} |
|
375
|
|
|
log::add($error_msg, 'Error'); |
|
376
|
|
|
} |
|
377
|
|
|
} |
|
378
|
|
|
|
|
379
|
|
|
/** |
|
380
|
|
|
* 获取错误类型对应的意义 |
|
381
|
|
|
* @param integer $type |
|
382
|
|
|
* @return string |
|
383
|
|
|
*/ |
|
384
|
|
|
protected function get_error_type($type) |
|
385
|
|
|
{ |
|
386
|
|
|
switch($type) |
|
387
|
|
|
{ |
|
388
|
|
|
case E_ERROR: // 1 // |
|
389
|
|
|
return 'E_ERROR'; |
|
390
|
|
|
case E_WARNING: // 2 // |
|
391
|
|
|
return 'E_WARNING'; |
|
392
|
|
|
case E_PARSE: // 4 // |
|
393
|
|
|
return 'E_PARSE'; |
|
394
|
|
|
case E_NOTICE: // 8 // |
|
395
|
|
|
return 'E_NOTICE'; |
|
396
|
|
|
case E_CORE_ERROR: // 16 // |
|
397
|
|
|
return 'E_CORE_ERROR'; |
|
398
|
|
|
case E_CORE_WARNING: // 32 // |
|
399
|
|
|
return 'E_CORE_WARNING'; |
|
400
|
|
|
case E_COMPILE_ERROR: // 64 // |
|
401
|
|
|
return 'E_COMPILE_ERROR'; |
|
402
|
|
|
case E_COMPILE_WARNING: // 128 // |
|
403
|
|
|
return 'E_COMPILE_WARNING'; |
|
404
|
|
|
case E_USER_ERROR: // 256 // |
|
405
|
|
|
return 'E_USER_ERROR'; |
|
406
|
|
|
case E_USER_WARNING: // 512 // |
|
407
|
|
|
return 'E_USER_WARNING'; |
|
408
|
|
|
case E_USER_NOTICE: // 1024 // |
|
409
|
|
|
return 'E_USER_NOTICE'; |
|
410
|
|
|
case E_STRICT: // 2048 // |
|
411
|
|
|
return 'E_STRICT'; |
|
412
|
|
|
case E_RECOVERABLE_ERROR: // 4096 // |
|
413
|
|
|
return 'E_RECOVERABLE_ERROR'; |
|
414
|
|
|
case E_DEPRECATED: // 8192 // |
|
415
|
|
|
return 'E_DEPRECATED'; |
|
416
|
|
|
case E_USER_DEPRECATED: // 16384 // |
|
417
|
|
|
return 'E_USER_DEPRECATED'; |
|
418
|
|
|
} |
|
419
|
|
|
return ""; |
|
420
|
|
|
} |
|
421
|
|
|
} |
|
422
|
|
|
|
This check looks for calls to methods that do not seem to exist on a given type. It looks for the method on the type itself as well as in inherited classes or implemented interfaces.
This is most likely a typographical error or the method has been renamed.