線程及進程概念可自行學習
Unix/Linux操作系統提供了一個fork()系統調用,它非常特殊。普通的函數調用,調用一次,返回一次,但是fork()調用一次,返回兩次,因爲操作系統自動把當前進程(稱爲父進程)複製了一份(稱爲子進程),然後,分別在父進程和子進程內返回。
子進程永遠返回0,而父進程返回子進程的ID。這樣做的理由是,一個父進程可以fork出很多子進程,所以,父進程要記下每個子進程的ID,而子進程只需要調用getppid()就可以拿到父進程的ID。
常用方法:
multiprocessing.cpu_count() 計算當前計算機有幾個CPU可用
multiprocessing.active_children() 查看當前還活着的子進程
p.is_alive() 查看當前進程是否存活
p.join() 進程的阻塞,如果join中無參數,則等待進程運行完後繼續執行主函數,如果join有timeout參數,則超出timeout時間後繼續執行主函數,不等待進程返回結果
p.name() 輸出p進程的名字
p.pid() 輸出p進程的pid是多少
p.start() 開始p進程,與run()方法相同
Python的os模塊封裝了常見的系統調用,其中就包括fork,可以在Python程序中輕鬆創建子進程:
例子:
import os
print 'Process (%s) start...' % os.getpid()
pid = os.fork()
if pid==0:
print 'I am child process (%s) and my parent is %s.' % (os.getpid(), os.getppid())
else:
print 'I (%s) just created a child process (%s).' % (os.getpid(), pid)
輸出:
Process (876) start...
I (876) just created a child process (877).
I am child process (877) and my parent is 876.
有了fork調用,一個進程在接到新任務時就可以複製出一個子進程來處理新任務,常見的Apache服務器就是由父進程監聽端口,每當有新的http請求時,就fork出子進程來處理新的http請求。
multiprocessing
由於Python是跨平臺的,自然也應該提供一個跨平臺的多進程支持。multiprocessing模塊就是跨平臺版本的多進程模塊。
multiprocessing模塊提供了一個Process類來代表一個進程對象,下面的例子演示了啓動一個子進程並等待其結束:
例子:
from multiprocessing import Process
import os
# 子進程要執行的代碼
def run_proc(name):
print 'Run child process %s (%s)...' % (name, os.getpid())
if __name__=='__main__':
print 'Parent process %s.' % os.getpid()
p = Process(target=run_proc, args=('test',))
print 'Process will start.'
p.start()
p.join()
print 'Process end.'
輸出:
Parent process 928.
Process will start.
Run child process test (929)...
Process end.
創建子進程時,只需要傳入一個執行函數和函數的參數,創建一個Process實例,用start()方法啓動,這樣創建進程比fork()還要簡單。
join()方法可以等待子進程結束後再繼續往下運行,通常用於進程間的同步。
例子:
#創建子進程的方法
import time
import multiprocessing
def worker(name,interval):
print ("{0} start".format(name))
time.sleep(interval)
print ("{0} end".format(name))
if __name__ == "__main__":
print("main start")
print (multiprocessing.cpu_count())
#創建子進程,目標是那個函數,傳遞的參數都有哪些
p1 = multiprocessing.Process(target=worker, args=("worker1",2))
p2 = multiprocessing.Process(target=worker, args=("worker2",3))
p3 = multiprocessing.Process(target=worker, args=("worker3",4))
#啓動進程
p1.start()
p2.start()
p3.start()
for i in multiprocessing.active_children():
print ("The PID of {0} is {1}".format(i.name, i.pid))
print("main end")
輸出:
main start
4
The PID of Process-1 is 1588
The PID of Process-3 is 6216
The PID of Process-2 is 5724
main end
worker1 start
worker2 start
worker3 start
worker1 end
worker2 end
worker3 end
Pool
如果要啓動大量的子進程,可以用進程池的方式批量創建子進程:
例子:
from multiprocessing import Pool
import os, time, random
def long_time_task(name):
print 'Run task %s (%s)...' % (name, os.getpid())
start = time.time()
time.sleep(random.random() * 3)
end = time.time()
print 'Task %s runs %0.2f seconds.' % (name, (end - start))
if __name__=='__main__':
print 'Parent process %s.' % os.getpid()
p = Pool()
for i in range(5):
p.apply_async(long_time_task, args=(i,))
print 'Waiting for all subprocesses done...'
p.close()
p.join()
print 'All subprocesses done.'
輸出:
Parent process 669.
Waiting for all subprocesses done...
Run task 0 (671)...
Run task 1 (672)...
Run task 2 (673)...
Run task 3 (674)...
Task 2 runs 0.14 seconds.
Run task 4 (673)...
Task 1 runs 0.27 seconds.
Task 3 runs 0.86 seconds.
Task 0 runs 1.41 seconds.
Task 4 runs 1.91 seconds.
All subprocesses done.
代碼解讀:
對Pool對象調用join()方法會等待所有子進程執行完畢,調用join()之前必須先調用close(),調用close()之後就不能繼續添加新的Process了。
請注意輸出的結果,task 0,1,2,3是立刻執行的,而task 4要等待前面某個task完成後才執行,這是因爲Pool的默認大小在我的電腦上是4,因此,最多同時執行4個進程。這是Pool有意設計的限制,並不是操作系統的限制。如果改成:
p = Pool(5)
就可以同時跑5個進程。
由於Pool的默認大小是CPU的核數,如果你不幸擁有8核CPU,你要提交至少9個子進程才能看到上面的等待效果。
進程間通信
Process之間肯定是需要通信的,操作系統提供了很多機制來實現進程間的通信。Python的multiprocessing模塊包裝了底層的機制,提供了Queue、Pipes等多種方式來交換數據。
我們以Queue爲例,在父進程中創建兩個子進程,一個往Queue裏寫數據,一個從Queue裏讀數據:
例子:
from multiprocessing import Process, Queue
import os, time, random
# 寫數據進程執行的代碼:
def write(q):
for value in ['A', 'B', 'C']:
print 'Put %s to queue...' % value
q.put(value)
time.sleep(random.random())
# 讀數據進程執行的代碼:
def read(q):
while True:
value = q.get(True)
print 'Get %s from queue.' % value
if __name__=='__main__':
# 父進程創建Queue,並傳給各個子進程:
q = Queue()
pw = Process(target=write, args=(q,))
pr = Process(target=read, args=(q,))
# 啓動子進程pw,寫入:
pw.start()
# 啓動子進程pr,讀取:
pr.start()
# 等待pw結束:
pw.join()
# pr進程裏是死循環,無法等待其結束,只能強行終止:
pr.terminate()
輸出:
Put A to queue...
Get A from queue.
Put B to queue...
Get B from queue.
Put C to queue...
Get C from queue.
在Unix/Linux下,multiprocessing模塊封裝了fork()調用,使我們不需要關注fork()的細節。由於Windows沒有fork調用,因此,multiprocessing需要“模擬”出fork的效果,父進程所有Python對象都必須通過pickle序列化再傳到子進程去,所有,如果multiprocessing在Windows下調用失敗了,要先考慮是不是pickle失敗了。
多進程鎖
例子:
import multiprocessing
import time
def add(number, value, lock):
#獲取鎖
lock.acquire()
#異常的捕獲
try:
print ("add{0} number = {1}".format(value, number))
for i in xrange(1, 6):
number += value
time.sleep(1)
print ("add{0} number = {1}".format(value, number))
except Exception as e:
raise e
finally:
#釋放鎖
lock.release()
if __name__ == "__main__":
#鎖的實例化
lock = multiprocessing.Lock()
number = 0
#進程包含進程鎖,p1和p2進程分別去搶鎖,先搶到的先運行
p1 = multiprocessing.Process(target=add, args=(number, 1, lock))
p2 = multiprocessing.Process(target=add, args=(number, 3, lock))
p1.start()
p2.start()
print ("main end")
輸出:
main end
add3 number = 0
add3 number = 3
add3 number = 6
add3 number = 9
add3 number = 12
add3 number = 15
add1 number = 0
add1 number = 1
add1 number = 2
add1 number = 3
add1 number = 4
add1 number = 5
例子:
import multiprocessing
import time
def add(number, value, lock):
#使用with lock寫法來自動加鎖及釋放,與acquire和release相同
with lock:
print ("add{0} number = {1}".format(value, number))
for i in xrange(1, 6):
number += value
time.sleep(1)
print ("add{0} number = {1}".format(value, number))
if __name__ == "__main__":
#鎖的實例化
lock = multiprocessing.Lock()
number = 0
#進程包含進程鎖,p1和p2進程分別去搶鎖,先搶到的先運行
p1 = multiprocessing.Process(target=add, args=(number, 1, lock))
p2 = multiprocessing.Process(target=add, args=(number, 3, lock))
p1.start()
p2.start()
print ("main end")
輸出:
main end
add1 number = 0
add1 number = 1
add1 number = 2
add1 number = 3
add1 number = 4
add1 number = 5
add3 number = 0
add3 number = 3
add3 number = 6
add3 number = 9
add3 number = 12
add3 number = 15
共享內存
import multiprocessing
import time
def add(number, add_value):
try:
print ("add{0} number = {1}".format(add_value, number.value))
for i in xrange(1, 6):
number.value += add_value
time.sleep(1)
print ("add{0} number = {1}".format(add_value, number.value))
except Exception as e :
raise e
if __name__ == "__main__":
#number共享內存的實例化,number.value纔可以使用共享內存操作,分別有value和array
number = multiprocessing.Value('i', 0)
p1 = multiprocessing.Process(target=add, args=(number, 1))
p2 = multiprocessing.Process(target=add, args=(number, 3))
p1.start()
p2.start()
print ("main end")
輸出:
main end
add1 number = 0
add3 number = 1
add1 number = 4
add3 number = 5
add1 number = 8
add3 number = 9
add1 number = 12
add3 number = 13
add1 number = 16
add3 number = 17
add1 number = 20
add3 number = 20
多進程manager管理
manager可以接收多種類型的數據,相比較array和value功能更豐富
例子:
import multiprocessing
def worker(d, l):
l += range(11,16)
for i in xrange(1,6):
key = "key {0}".format(i)
value = "value {0}".format(i)
d[key] = value
if __name__ == "__main__":
#實例化manager
manager = multiprocessing.Manager()
#接收字典類型的數據
d = manager.dict()
#接收列表類型的數據
l = manager.list()
p = multiprocessing.Process(target=worker, args=(d, l))
p.start()
p.join()
print (d)
print (l)
輸出:
{'key 1': 'value 1', 'key 2': 'value 2', 'key 3': 'value 3', 'key 4': 'value 4', 'key 5': 'value 5'}
[11, 12, 13, 14, 15]
進程池
與MySQL連接池含義類似,創建連接池後所有進程都從進程池連接,超出進程池數量的進程會排隊等待
例子:
import multiprocessing
import time
def fun1(message):
print ("start {0}".format(message))
time.sleep(1)
print ("end {0}".format(message))
if __name__ == "__main__":
# 實例化進程池
pool = multiprocessing.Pool(2)
for i in xrange(1,10):
message = "number is {0}".format(i)
# apply_async是將進程池跑滿,多進程同時操作
pool.apply_async(func=fun1,args=(message,))
pool.close()
# 等待所有進程關閉,在join前需要close
pool.join()
輸出:
start number is 1
start number is 2
end number is 1
start number is 3
end number is 2
start number is 4
end number is 4
start number is 5
end number is 3
start number is 6
end number is 6
start number is 7
end number is 5
start number is 8
end number is 8
end number is 7
start number is 9
end number is 9
例子:
import multiprocessing
import time
def fun1(message):
print ("start {0}".format(message))
time.sleep(1)
print ("end {0}".format(message))
if __name__ == "__main__":
# 實例化進程池
pool = multiprocessing.Pool(2)
for i in xrange(1,10):
message = "number is {0}".format(i)
# apply是單進程,只有一個進程在運行
pool.apply(func=fun1,args=(message,))
pool.close()
# 等待所有進程關閉,在join前需要close
pool.join()
輸出:
start number is 1
end number is 1
start number is 2
end number is 2
start number is 3
end number is 3
start number is 4
end number is 4
start number is 5
end number is 5
start number is 6
end number is 6
start number is 7
end number is 7
start number is 8
end number is 8
start number is 9
end number is 9