Python多进程编程

http://blog.chinaunix.net/uid-26000296-id-4442717.html

Python在2.6版本新引进了多进程机制
一、 简介
多进程在Python中是一个包，它支持使用和线程模块类似的API来创建进程。 
多进程包支持本地和远程并发，并有高效的GIL锁机制。
因为这些原因，多进程模块允许程序员在指定机器上实现完整的多进程开发，无论是Unix还是Windows.

Warning:
这个包的某些功能要求在运行的OS上能实现共享信号量机制，
如果没有这个功能，多进程同步模块将不能运行，并在导入模块时出现ImportError, 
可以看问题3770描述，获得更多的信息；

Note：
这个包中的功能要求 __main__ 模块可以被导入。
这个在编程指导中有提到但也还值得在这里再指出来。
这意味着，在有些例子中，如 multiprocessing.Pool的例子中，它不会进行交互解释工作。
例子如下：

>>>
>>> from multiprocessing import Pool
>>> p = Pool(5)
>>> def f(x):
...     return x*x
...
>>> p.map(f, [1,2,3])

Process PoolWorker-1:
Process PoolWorker-2:
Process PoolWorker-3:
Traceback (most recent call last):
AttributeError: 'module' object has no attribute 'f'
AttributeError: 'module' object has no attribute 'f'
AttributeError: 'module' object has no attribute 'f'

(If you try this it will actually output three full tracebacks interleaved in a semi-random fashion, 
and then you may have to stop the master process somehow.)

1、Process类
In multiprocessing, processes are spawned by creating a Process object and then calling its start() method. Process follows 
the API of threading.Thread. A trivial example of a multiprocess program is
在 multiprocessing中，进程通过创建一个 Process对象被生成，然后调用它的 start()方法。
进程的API和 threading.Thread的类似。
一个简单的多进程程序示例如下：

from multiprocessing import Process

def f(name):
    print 'hello', name

if __name__ == '__main__':
    p = Process(target=f, args=('bob',))
    p.start()
    p.join()

下面的例子显示了独立的进程ID：

from multiprocessing import Process
import os

def info(title):
    print title
    print 'module name:', __name__
    if hasattr(os, 'getppid'):  # only available on Unix
        print 'parent process:', os.getppid()
    print 'process id:', os.getpid()

def f(name):
    info('function f')
    print 'hello', name

if __name__ == '__main__':
    info('main line')
    p = Process(target=f, args=('bob',))
    p.start()
    p.join()

2、 进程间交换对象
multiprocessing支持两种类型的进程间的通信通道:
2.1 队列
队列类很像是Queue.Queue的克隆，例如:

from multiprocessing import Process, Queue

def f(q):
    q.put([42, None, 'hello'])

if __name__ == '__main__':
    q = Queue()
    p = Process(target=f, args=(q,))
    p.start()
    print q.get()    # prints "[42, None, 'hello']"
    p.join()

队列是线程和进程安全的。

2.2 管道
Pipe()函数返回一对管道连接的对象，默认的它们是双向的。例如:

from multiprocessing import Process, Pipe

def f(conn):
    conn.send([42, None, 'hello'])
    conn.close()

if __name__ == '__main__':
    parent_conn, child_conn = Pipe()
    p = Process(target=f, args=(child_conn,))
    p.start()
    print parent_conn.recv()   # prints "[42, None, 'hello']"
    p.join()

通过Pipe()生成的两个连接对象，表示管道的两端。
每个连接对象都有 send() 和 recv()方法。
要注意的是：如果两个进程(或线程)，在同一时间尝试从管道的同一端读或写时，数据会被变脏。
当然，在同一时间，进程间使用不同的管道进行读写是没有任务风险的。

3、进程间的同步
multiprocessing包含了threading中所有同步机制。
例如，可以使用锁来确保在某个时刻，只有一个进程输出信息到标准输出:

from multiprocessing import Process, Lock

def f(l, i):
    l.acquire()
    print 'hello world', i
    l.release()

if __name__ == '__main__':
    lock = Lock()

    for num in range(10):
        Process(target=f, args=(lock, num)).start()

不使用锁时，不同进程的输出会混到一起。

4、进程间的共享 
如上面所提，当进行并行编程时，通常最好尽可能地避免使用共享，这一点在多进程编程中更是如此。
当然，如果实在要在进程间共享数据，multiprocessing也提供了两种方法来实现。
4.1 共享内存
数据可以使用Value或Array类型存储在共享内存映射中，示例如下:

from multiprocessing import Process, Value, Array

def f(n, a):
    n.value = 3.1415927
    for i in range(len(a)):
        a[i] = -a[i]

if __name__ == '__main__':
    num = Value('d', 0.0)
    arr = Array('i', range(10))

    p = Process(target=f, args=(num, arr))
    p.start()
    p.join()

    print num.value
    print arr[:]

将会输出:
3.1415927
[0, -1, -2, -3, -4, -5, -6, -7, -8, -9]

The 'd' and 'i' arguments used when creating num and arr are typecodes of the kind used by the array module:
  'd' indicates a double precision float and 
  'i' indicates a signed integer. 
These shared objects will be process and thread-safe.

为了更灵活地使用共享内存，可以使用multiprocessing.sharedctypes模块，
它支持从共享内存中分配任意ctype对象的创建。

4.2 Server process
Manager()函数返回一个管理对象，它控制了一个服务端进程，
用来保持Python对象，并允许其它进程使用代理来管理这些对象。
Manager()返回的管理者，支持类型包括，
list, dict, Namespace, Lock,RLock, Semaphore, BoundedSemaphore,Condition, Event, Queue, Value and Array。
示例如下:

from multiprocessing import Process, Manager

def f(d, l):
    d[1] = '1'
    d['2'] = 2
    d[0.25] = None
    l.reverse()

if __name__ == '__main__':
    manager = Manager()

    d = manager.dict()
    l = manager.list(range(10))

    p = Process(target=f, args=(d, l))
    p.start()
    p.join()

    print d
    print l

将会输出:
{0.25: None, 1: '1', '2': 2}
[9, 8, 7, 6, 5, 4, 3, 2, 1, 0]

Server process managers比使用共享内存对象更灵活，因为它支持任意对象类型。
同样的，单个的manager可以通过网络在不同机器上进程间共享。当然，会比共享内存慢。

5、 使用工作池
Pool类表示一个工作进程池，它有方法，可以以不同的方式将任务卸载到不同工作进程。
示例如下:

from multiprocessing import Pool

def f(x):
    return x*x

if __name__ == '__main__':
    pool = Pool(processes=4)              # start 4 worker processes
    result = pool.apply_async(f, [10])    # evaluate "f(10)" asynchronously
    print result.get(timeout=1)           # prints "100" unless your computer is *very* slow
    print pool.map(f, range(10))          # prints "[0, 1, 4,..., 81]"

Note that the methods of a pool should only ever be used by the process which created it.

二、 引用
multiprocessing包复制了绝大多数的threading模块的API。
1. 进程和异常
class multiprocessing.Process(group=None, target=None, name=None, args=(), kwargs={})

Process对象表示分离的进程被激活，Process类有和threading.Thread同样的方法

构建函数总是会被调用，参数设置中，
  group   为None；它的存在是为了兼容threading.Thread.
  target  为可调用的对象，它被 run()方法引用，它的默认值是 None,意即没有什么可调用的。
  name    是进程名，默认地，它会是一个唯一的名字，如‘Process-N1:N2:...Nk',N1...Nk是一个整数序列，由进程生成时确认。
  args    是传递给target的参数。
  kwargs  是一个关键词参数的字典。

如果有子类覆盖了构建函数，那么必须它会在做任务处理前调用基类的构建函数(Process.__init__())
run()
    表示进程激活。
    You may override this method in a subclass. 
    The standard run() method invokes the callable object passed to the object’s constructor as the target argument,
    if any, with sequential and keyword arguments taken from the args and kwargs arguments, respectively.
start()
    启动进程
    它必须被每个进程对象调用至少一次，它被安排到run方法激活的分立的进程对象中。
join([timeout])
    Block the calling thread until the process whose join() method is called terminates 
    or until the optional timeout occurs.
    If timeout is None then there is no timeout.
    A process can be joined many times.
    A process cannot join itself because this would cause a deadlock. 
    It is an error to attempt to join a process before it has been started.
name
    进程名
    The name is a string used for identification purposes only. It has no semantics.
    Multiple processes may be given the same name. The initial name is set by the constructor.
is_alive()
    返回进程是否仍在活动
    Roughly, a process object is alive from the moment the start() method returns until the child process terminates.
daemon
    The process’s daemon flag, a Boolean value. This must be set before start() is called.
    The initial value is inherited from the creating process.

    When a process exits, it attempts to terminate all of its daemonic child processes.
    Note that a daemonic process is not allowed to create child processes. 
    Otherwise a daemonic process would leave its children orphaned if it gets terminated when its parent process exits.       

    Additionally, these are not Unix daemons or services, 
    they are normal processes that will be terminated (and not joined) if non-daemonic processes have exited.
    In addition to the threading.Thread API, Process objects also support the following attributes and methods:
pid
    Return the process ID. Before the process is spawned, this will be None.
exitcode
    The child’s exit code. This will be None if the process has not yet terminated. 
    A negative value -N indicates that the child was terminated by signal N.
authkey
    The process’s authentication key (a byte string).
    When multiprocessing is initialized the main process is assigned a random string using os.urandom().
    When a Process object is created, it will inherit the authentication key of its parent process, 
    although this may be changed by setting authkey to another byte string.
    See Authentication keys.
terminate()
    Terminate the process. On Unix this is done using the SIGTERM signal; 
    on Windows TerminateProcess() is used. Note that exit     handlers and finally clauses, etc., will not be executed.

    Note that descendant processes of the process will not be terminated – they will simply become orphaned.

    Warning If this method is used when the associated process is using a pipe or queue 
    then the pipe or queue is liable to become corrupted and may become unusable by other process. 
    Similarly, if the process has acquired a lock or semaphore etc. 
    then terminating it is liable to cause other processes to deadlock.   
NOTE：the start(), join(), is_alive(), terminate() and exitcode methods 
should only be called by the process that created the process object.

Process的方法的使用示例:

>>>
>>> import multiprocessing, time, signal
>>> p = multiprocessing.Process(target=time.sleep, args=(1000,))
>>> print p, p.is_alive()
False

>>> p.start()
>>> print p, p.is_alive()
True

>>> p.terminate()
>>> time.sleep(0.1)
>>> print p, p.is_alive()
False

>>> p.exitcode == -signal.SIGTERM
True

exception multiprocessing.BufferTooShort
Exception raised by Connection.recv_bytes_into() when the supplied buffer object is too small for the message read.

If e is an instance of BufferTooShort then e.args[0] will give the message as a byte string.

2. 管道和队列
对于多进程编程来说，通常是使用消息传递来实现进程间的通信，而避免使用同步原子，如锁来实现。
对于消息传递，可以使用 Pipe()(用于两个进程的连接)或一个队列(它允许多个生产都和消费者)。

Queue, multiprocessing.queues.SimpleQueue 和 JoinableQueue类型都是多生产者，多消费者的FIFO队列。 
它们之间的不同之处在于Queue没有task_done() 和 join()方法；

If you use JoinableQueue then you must call JoinableQueue.task_done() for each task removed from the queue or else the 
semaphore used to count the number of unfinished tasks may eventually overflow, raising an exception.
如果使用 JoinableQueue，就必须对每个任务调用 JoinableQueue.task_done()，以从队列或信号量中移除。

NOTE:
 that one can also create a shared queue by using a manager object – see Managers.
NOTE:
 multiprocessing uses the usual Queue.Empty and Queue.Full exceptions to signal a timeout. 
 They are not available in the multiprocessing namespace so you need to import them from Queue.
NOTE:
 When an object is put on a queue, the object is pickled and a background thread later flushes the pickled data to an  
 underlying pipe. This has some consequences which are a little surprising, 
 but should not cause any practical difficulties – 
 if they really bother you then you can instead use a queue created with a manager.
    1. After putting an object on an empty queue there may be an infinitesimal delay 
       before the queue’s empty() method returns False and get_nowait() can return without raising Queue.Empty.
    2. If multiple processes are enqueuing objects, 
       it is possible for the objects to be received at the other end out-of-order.        
       However, objects enqueued by the same process will always be in the expected order with respect to each other.
WARNING: 
  If a process is killed using Process.terminate() or os.kill() while it is trying to use a Queue, 
  then the data in the queue is likely to become corrupted. 
  This may cause any other process to get an exception when it tries to use the queue later on.
WARNING:
  As mentioned above, if a child process has put items on a queue (and it has not used JoinableQueue.cancel_join_thread), 
  then that process will not terminate until all buffered items have been flushed to the pipe.
  This means that if you try joining that process you may get a deadlock unless you are sure 
  that all items which have been put on the queue have been consumed. Similarly, 
  if the child process is non-daemonic then the parent process may hang on exit 
  when it tries to join all its non-daemonic children.
  Note that a queue created using a manager does not have this issue. See Programming guidelines.
队列的使用示例方法说明:
multiprocessing.Pipe([duplex])
    Returns a pair (conn1, conn2) of Connection objects representing the ends of a pipe.

    If duplex is True (the default) then the pipe is bidirectional. 
    If duplex is False then the pipe is unidirectional: 
       conn1 can only be used for receiving messages and conn2 can only be used for sending messages.

class multiprocessing.Queue([maxsize])
    Returns a process shared queue implemented using a pipe and a few locks/semaphores. 
    When a process first puts an item on the queue a feeder thread is started 
    which transfers objects from a buffer into the pipe.

    The usual Queue.Empty and Queue.Full exceptions from the standard library’s Queue module 
    are raised to signal timeouts.

    Queue implements all the methods of Queue.Queue except for task_done() and join().
qsize()
    Return the approximate size of the queue. 
    Because of multithreading/multiprocessing semantics, this number is not reliable.

    Note that this may raise NotImplementedError on Unix platforms like Mac OS X where sem_getvalue() is not implemented.
empty()
    Return True if the queue is empty, False otherwise. 
    Because of multithreading/multiprocessing semantics, this is not reliable.
full()
    Return True if the queue is full, False otherwise. 
    Because of multithreading/multiprocessing semantics, this is not reliable.
put(obj[, block[, timeout]])
    Put obj into the queue. If the optional argument block is True (the default) and timeout is None (the default), 
    block if necessary until a free slot is available. If timeout is a positive number, 
    it blocks at most timeout seconds and raises the Queue.
    Full exception if no free slot was available within that time. Otherwise (block is False), 
    put an item on the queue if a free slot is immediately available, 
    else raise the Queue.Full exception (timeout is ignored in that case).
put_nowait(obj)
    Equivalent to put(obj, False).
get([block[, timeout]])
    Remove and return an item from the queue. If optional args block is True (the default) 
    and timeout is None (the default), block if necessary until an item is available. 
    If timeout is a positive number, it blocks at most timeout seconds and raises     the Queue.Empty exception if no item 
    was available within that time. 
    Otherwise (block is False), return an item if one is immediately available, 
    else raise the Queue.Empty exception (timeout is ignored in that case).
get_nowait()
    Equivalent to get(False).
   Queue has a few additional methods not found in Queue.Queue. These methods are usually unnecessary for most code:
close()
    Indicate that no more data will be put on this queue by the current process. 
    The background thread will quit once it has flushed all buffered data to the pipe. 
    This is called automatically when the queue is garbage collected.
join_thread()
    Join the background thread. This can only be used after close() has been called. 
    It blocks until the background thread exits, ensuring that all data in the buffer has been flushed to the pipe.

    By default if a process is not the creator of the queue then on exit 
    it will attempt to join the queue’s background thread.     The process can call cancel_join_thread() to make 
join_thread() 
   do nothing.
cancel_join_thread()
    Prevent join_thread() from blocking. In particular, this prevents the background 
    thread from being joined automatically when the process exits – see join_thread().
    A better name for this method might be allow_exit_without_flush(). 
    It is likely to cause enqueued data to lost, and you almost certainly will not need to use it. 
    It is really only there if you need the current process to exit immediately without     waiting to flush enqueued data to 
    the underlying pipe, and you don’t care about lost data.

class multiprocessing.queues.SimpleQueue
It is a simplified Queue type, very close to a locked Pipe.
empty()
    Return True if the queue is empty, False otherwise.
get()
    Remove and return an item from the queue.
put(item)
    Put item into the queue.

class multiprocessing.JoinableQueue([maxsize])
JoinableQueue, a Queue subclass, is a queue which additionally has task_done() and join() methods.
task_done()
    Indicate that a formerly enqueued task is complete. Used by queue consumer threads. 
    For each get() used to fetch a task, a subsequent call to task_done() tells the queue 
    that the processing on the task is complete.
    If a join() is currently blocking, it will resume when all items have been processed 
    (meaning that a task_done() call was received for every item that had been put() into the queue).

    Raises a ValueError if called more times than there were items placed in the queue.
join()
    Block until all items in the queue have been gotten and processed.
    The count of unfinished tasks goes up whenever an item is added to the queue.
    The count goes down whenever a consumer thread calls task_done() to indicate 
    that the item was retrieved and all work on it is complete. When the count of unfinished tasks     drops to zero, join() unblocks.

3. 杂项
multiprocessing.active_children()
Return list of all live children of the current process.
Calling this has the side affect of “joining” any processes which have already finished.

multiprocessing.cpu_count()
Return the number of CPUs in the system. May raise NotImplementedError.

multiprocessing.current_process()
Return the Process object corresponding to the current process.

An analogue of threading.current_thread().

multiprocessing.freeze_support()
Add support for when a program which uses multiprocessing has been frozen to produce a Windows executable. (Has been tested 
with py2exe, PyInstaller and cx_Freeze.)
One needs to call this function straight after the if __name__ == '__main__' line of the main module. For example:

from multiprocessing import Process, freeze_support

def f():
    print 'hello world!'
if __name__ == '__main__':
    freeze_support()
    Process(target=f).start()

If the freeze_support() line is omitted then trying to run the frozen executable will raise RuntimeError.
If the module is being run normally by the Python interpreter then freeze_support() has no effect.

multiprocessing.set_executable()
Sets the path of the Python interpreter to use when starting a child process. (By default sys.executable is used). Embedders 
will probably need to do some thing like

set_executable(os.path.join(sys.exec_prefix, 'pythonw.exe'))
before they can create child processes. (Windows only)

Note multiprocessing contains no analogues of threading.active_count(), threading.enumerate(), threading.settrace(), 


threading.setprofile(), threading.Timer, or threading.local.


4.连接对象
Connection objects allow the sending and receiving of picklable objects or strings. They can be thought of as message 


oriented connected sockets.


Connection objects are usually created using Pipe() – see also Listeners and Clients.


class multiprocessing.Connection
send(obj)
Send an object to the other end of the connection which should be read using recv().


The object must be picklable. Very large pickles (approximately 32 MB+, though it depends on the OS) may raise a ValueError 


exception.


recv()
Return an object sent from the other end of the connection using send(). Blocks until there its something to receive. Raises 


EOFError if there is nothing left to receive and the other end was closed.


fileno()
Return the file descriptor or handle used by the connection.


close()
Close the connection.


This is called automatically when the connection is garbage collected.


poll([timeout])
Return whether there is any data available to be read.


If timeout is not specified then it will return immediately. If timeout is a number then this specifies the maximum time in 


seconds to block. If timeout is None then an infinite timeout is used.


send_bytes(buffer[, offset[, size]])
Send byte data from an object supporting the buffer interface as a complete message.


If offset is given then data is read from that position in buffer. If size is given then that many bytes will be read from 


buffer. Very large buffers (approximately 32 MB+, though it depends on the OS) may raise a ValueError exception


recv_bytes([maxlength])
Return a complete message of byte data sent from the other end of the connection as a string. Blocks until there is something 


to receive. Raises EOFError if there is nothing left to receive and the other end has closed.


If maxlength is specified and the message is longer than maxlength then IOError is raised and the connection will no longer 


be readable.


recv_bytes_into(buffer[, offset])
Read into buffer a complete message of byte data sent from the other end of the connection and return the number of bytes in 


the message. Blocks until there is something to receive. Raises EOFError if there is nothing left to receive and the other 


end was closed.


buffer must be an object satisfying the writable buffer interface. If offset is given then the message will be written into 


the buffer from that position. Offset must be a non-negative integer less than the length of buffer (in bytes).


If the buffer is too short then a BufferTooShort exception is raised and the complete message is available as e.args[0] where 


e is the exception instance.


For example:


>>>
>>> from multiprocessing import Pipe
>>> a, b = Pipe()
>>> a.send([1, 'hello', None])
>>> b.recv()


[1, 'hello', None]
>>> b.send_bytes('thank you')
>>> a.recv_bytes()
'thank you'


>>> import array
>>> arr1 = array.array('i', range(5))
>>> arr2 = array.array('i', [0] * 10)
>>> a.send_bytes(arr1)
>>> count = b.recv_bytes_into(arr2)
>>> assert count == len(arr1) * arr1.itemsize
>>> arr2
array('i', [0, 1, 2, 3, 4, 0, 0, 0, 0, 0])


Warning The Connection.recv() method automatically unpickles the data it receives, which can be a security risk unless you 


can trust the process which sent the message.
Therefore, unless the connection object was produced using Pipe() you should only use the recv() and send() methods after 


performing some sort of authentication. See Authentication keys.
Warning If a process is killed while it is trying to read or write to a pipe then the data in the pipe is likely to become 


corrupted, because it may become impossible to be sure where the message boundaries lie.


5. 同步原子
Generally synchronization primitives are not as necessary in a multiprocess program as they are in a multithreaded program. 


See the documentation for threading module.


Note that one can also create synchronization primitives by using a manager object – see Managers.


class multiprocessing.BoundedSemaphore([value])
A bounded semaphore object: a clone of threading.BoundedSemaphore.


(On Mac OS X, this is indistinguishable from Semaphore because sem_getvalue() is not implemented on that platform).


class multiprocessing.Condition([lock])
A condition variable: a clone of threading.Condition.


If lock is specified then it should be a Lock or RLock object from multiprocessing.


class multiprocessing.Event
A clone of threading.Event. This method returns the state of the internal semaphore on exit, so it will always return True 


except if a timeout is given and the operation times out.


Changed in version 2.7: Previously, the method always returned None.


class multiprocessing.Lock
A non-recursive lock object: a clone of threading.Lock.


class multiprocessing.RLock
A recursive lock object: a clone of threading.RLock.


class multiprocessing.Semaphore([value])
A semaphore object: a clone of threading.Semaphore.


Note The acquire() method of BoundedSemaphore, Lock, RLock and Semaphore has a timeout parameter not supported by the 


equivalents in threading. The signature is acquire(block=True, timeout=None) with keyword parameters being acceptable. If 


block is True and timeout is not None then it specifies a timeout in seconds. If block is False then timeout is ignored.
On Mac OS X, sem_timedwait is unsupported, so calling acquire() with a timeout will emulate that function’s behavior using a 


sleeping loop.
Note If the SIGINT signal generated by Ctrl-C arrives while the main thread is blocked by a call to BoundedSemaphore.acquire


(), Lock.acquire(), RLock.acquire(), Semaphore.acquire(), Condition.acquire() or Condition.wait() then the call will be 


immediately interrupted and KeyboardInterrupt will be raised.
This differs from the behaviour of threading where SIGINT will be ignored while the equivalent blocking calls are in 


progress.


6. 共享类型对象
It is possible to create shared objects using shared memory which can be inherited by child processes.


multiprocessing.Value(typecode_or_type, *args[, lock])
Return a ctypes object allocated from shared memory. By default the return value is actually a synchronized wrapper for the 


object.


typecode_or_type determines the type of the returned object: it is either a ctypes type or a one character typecode of the 


kind used by the array module. *args is passed on to the constructor for the type.


If lock is True (the default) then a new recursive lock object is created to synchronize access to the value. If lock is a 


Lock or RLock object then that will be used to synchronize access to the value. If lock is False then access to the returned 


object will not be automatically protected by a lock, so it will not necessarily be “process-safe”.


Operations like += which involve a read and write are not atomic. So if, for instance, you want to atomically increment a 


shared value it is insufficient to just do


counter.value += 1
Assuming the associated lock is recursive (which it is by default) you can instead do


with counter.get_lock():
    counter.value += 1
Note that lock is a keyword-only argument.


multiprocessing.Array(typecode_or_type, size_or_initializer, *, lock=True)
Return a ctypes array allocated from shared memory. By default the return value is actually a synchronized wrapper for the 


array.


typecode_or_type determines the type of the elements of the returned array: it is either a ctypes type or a one character 


typecode of the kind used by the array module. If size_or_initializer is an integer, then it determines the length of the 


array, and the array will be initially zeroed. Otherwise, size_or_initializer is a sequence which is used to initialize the 


array and whose length determines the length of the array.


If lock is True (the default) then a new lock object is created to synchronize access to the value. If lock is a Lock or 


RLock object then that will be used to synchronize access to the value. If lock is False then access to the returned object 


will not be automatically protected by a lock, so it will not necessarily be “process-safe”.


Note that lock is a keyword only argument.


Note that an array of ctypes.c_char has value and raw attributes which allow one to use it to store and retrieve strings.


6.1. 多进程共享类型模块
The multiprocessing.sharedctypes module provides functions for allocating ctypes objects from shared memory which can be 


inherited by child processes.


Note Although it is possible to store a pointer in shared memory remember that this will refer to a location in the address 


space of a specific process. However, the pointer is quite likely to be invalid in the context of a second process and trying 


to dereference the pointer from the second process may cause a crash.
multiprocessing.sharedctypes.RawArray(typecode_or_type, size_or_initializer)
Return a ctypes array allocated from shared memory.


typecode_or_type determines the type of the elements of the returned array: it is either a ctypes type or a one character 


typecode of the kind used by the array module. If size_or_initializer is an integer then it determines the length of the 


array, and the array will be initially zeroed. Otherwise size_or_initializer is a sequence which is used to initialize the 


array and whose length determines the length of the array.


Note that setting and getting an element is potentially non-atomic – use Array() instead to make sure that access is 


automatically synchronized using a lock.


multiprocessing.sharedctypes.RawValue(typecode_or_type, *args)
Return a ctypes object allocated from shared memory.


typecode_or_type determines the type of the returned object: it is either a ctypes type or a one character typecode of the 


kind used by the array module. *args is passed on to the constructor for the type.


Note that setting and getting the value is potentially non-atomic – use Value() instead to make sure that access is 


automatically synchronized using a lock.


Note that an array of ctypes.c_char has value and raw attributes which allow one to use it to store and retrieve strings – 


see documentation for ctypes.


multiprocessing.sharedctypes.Array(typecode_or_type, size_or_initializer, *args[, lock])
The same as RawArray() except that depending on the value of lock a process-safe synchronization wrapper may be returned 


instead of a raw ctypes array.


If lock is True (the default) then a new lock object is created to synchronize access to the value. If lock is a Lock or 


RLock object then that will be used to synchronize access to the value. If lock is False then access to the returned object 


will not be automatically protected by a lock, so it will not necessarily be “process-safe”.


Note that lock is a keyword-only argument.


multiprocessing.sharedctypes.Value(typecode_or_type, *args[, lock])
The same as RawValue() except that depending on the value of lock a process-safe synchronization wrapper may be returned 


instead of a raw ctypes object.


If lock is True (the default) then a new lock object is created to synchronize access to the value. If lock is a Lock or 


RLock object then that will be used to synchronize access to the value. If lock is False then access to the returned object 


will not be automatically protected by a lock, so it will not necessarily be “process-safe”.


Note that lock is a keyword-only argument.


multiprocessing.sharedctypes.copy(obj)
Return a ctypes object allocated from shared memory which is a copy of the ctypes object obj.


multiprocessing.sharedctypes.synchronized(obj[, lock])
Return a process-safe wrapper object for a ctypes object which uses lock to synchronize access. If lock is None (the default) 


then a multiprocessing.RLock object is created automatically.


A synchronized wrapper will have two methods in addition to those of the object it wraps: get_obj() returns the wrapped 


object and get_lock() returns the lock object used for synchronization.


Note that accessing the ctypes object through the wrapper can be a lot slower than accessing the raw ctypes object.


The table below compares the syntax for creating shared ctypes objects from shared memory with the normal ctypes syntax. (In 


the table MyStruct is some subclass of ctypes.Structure.)


ctypes sharedctypes using type sharedctypes using typecode
c_double(2.4) RawValue(c_double, 2.4) RawValue(‘d’, 2.4)
MyStruct(4, 6) RawValue(MyStruct, 4, 6)  
(c_short * 7)() RawArray(c_short, 7) RawArray(‘h’, 7)
(c_int * 3)(9, 2, 8) RawArray(c_int, (9, 2, 8)) RawArray(‘i’, (9, 2, 8))
Below is an example where a number of ctypes objects are modified by a child process:


from multiprocessing import Process, Lock
from multiprocessing.sharedctypes import Value, Array
from ctypes import Structure, c_double


class Point(Structure):
    _fields_ = [('x', c_double), ('y', c_double)]


def modify(n, x, s, A):
    n.value **= 2
    x.value **= 2
    s.value = s.value.upper()
    for a in A:
        a.x **= 2
        a.y **= 2


if __name__ == '__main__':
    lock = Lock()


    n = Value('i', 7)
    x = Value(c_double, 1.0/3.0, lock=False)
    s = Array('c', 'hello world', lock=lock)
    A = Array(Point, [(1.875,-6.25), (-5.75,2.0), (2.375,9.5)], lock=lock)


    p = Process(target=modify, args=(n, x, s, A))
    p.start()
    p.join()


    print n.value
    print x.value
    print s.value
    print [(a.x, a.y) for a in A]
The results printed are


49
0.1111111111111111
HELLO WORLD
[(3.515625, 39.0625), (33.0625, 4.0), (5.640625, 90.25)]


7. 管理者
Managers provide a way to create data which can be shared between different processes. A manager object controls a server 


process which manages shared objects. Other processes can access the shared objects by using proxies.


multiprocessing.Manager()
Returns a started SyncManager object which can be used for sharing objects between processes. The returned manager object 


corresponds to a spawned child process and has methods which will create shared objects and return corresponding proxies.


Manager processes will be shutdown as soon as they are garbage collected or their parent process exits. The manager classes 


are defined in the multiprocessing.managers module:


class multiprocessing.managers.BaseManager([address[, authkey]])
Create a BaseManager object.


Once created one should call start() or get_server().serve_forever() to ensure that the manager object refers to a started 


manager process.


address is the address on which the manager process listens for new connections. If address is None then an arbitrary one is 


chosen.


authkey is the authentication key which will be used to check the validity of incoming connections to the server process. If 


authkey is None then current_process().authkey. Otherwise authkey is used and it must be a string.


start([initializer[, initargs]])
Start a subprocess to start the manager. If initializer is not None then the subprocess will call initializer(*initargs) when 


it starts.


get_server()
Returns a Server object which represents the actual server under the control of the Manager. The Server object supports the 


serve_forever() method:


>>>
>>> from multiprocessing.managers import BaseManager
>>> manager = BaseManager(address=('', 50000), authkey='abc')
>>> server = manager.get_server()
>>> server.serve_forever()
Server additionally has an address attribute.


connect()
Connect a local manager object to a remote manager process:


>>>
>>> from multiprocessing.managers import BaseManager
>>> m = BaseManager(address=('127.0.0.1', 5000), authkey='abc')
>>> m.connect()
shutdown()
Stop the process used by the manager. This is only available if start() has been used to start the server process.


This can be called multiple times.


register(typeid[, callable[, proxytype[, exposed[, method_to_typeid[, create_method]]]]])
A classmethod which can be used for registering a type or callable with the manager class.


typeid is a “type identifier” which is used to identify a particular type of shared object. This must be a string.


callable is a callable used for creating objects for this type identifier. If a manager instance will be created using the 


from_address() classmethod or if the create_method argument is False then this can be left as None.


proxytype is a subclass of BaseProxy which is used to create proxies for shared objects with this typeid. If None then a 


proxy class is created automatically.


exposed is used to specify a sequence of method names which proxies for this typeid should be allowed to access using 


BaseProxy._callmethod(). (If exposed is None then proxytype._exposed_ is used instead if it exists.) In the case where no 


exposed list is specified, all “public methods” of the shared object will be accessible. (Here a “public method” means 


any attribute which has a __call__() method and whose name does not begin with '_'.)


method_to_typeid is a mapping used to specify the return type of those exposed methods which should return a proxy. It maps 


method names to typeid strings. (If method_to_typeid is None then proxytype._method_to_typeid_ is used instead if it exists.) 


If a method’s name is not a key of this mapping or if the mapping is None then the object returned by the method will be 


copied by value.


create_method determines whether a method should be created with name typeid which can be used to tell the server process to 


create a new shared object and return a proxy for it. By default it is True.


BaseManager instances also have one read-only property:


address
The address used by the manager.


class multiprocessing.managers.SyncManager
A subclass of BaseManager which can be used for the synchronization of processes. Objects of this type are returned by 


multiprocessing.Manager().


It also supports creation of shared lists and dictionaries.


BoundedSemaphore([value])
Create a shared threading.BoundedSemaphore object and return a proxy for it.


Condition([lock])
Create a shared threading.Condition object and return a proxy for it.


If lock is supplied then it should be a proxy for a threading.Lock or threading.RLock object.


Event()
Create a shared threading.Event object and return a proxy for it.


Lock()
Create a shared threading.Lock object and return a proxy for it.


Namespace()
Create a shared Namespace object and return a proxy for it.


Queue([maxsize])
Create a shared Queue.Queue object and return a proxy for it.


RLock()
Create a shared threading.RLock object and return a proxy for it.


Semaphore([value])
Create a shared threading.Semaphore object and return a proxy for it.


Array(typecode, sequence)
Create an array and return a proxy for it.


Value(typecode, value)
Create an object with a writable value attribute and return a proxy for it.


dict()
dict(mapping)
dict(sequence)
Create a shared dict object and return a proxy for it.


list()
list(sequence)
Create a shared list object and return a proxy for it.


Note Modifications to mutable values or items in dict and list proxies will not be propagated through the manager, because 


the proxy has no way of knowing when its values or items are modified. To modify such an item, you can re-assign the modified 


object to the container proxy:
# create a list proxy and append a mutable object (a dictionary)
lproxy = manager.list()
lproxy.append({})
# now mutate the dictionary
d = lproxy[0]
d['a'] = 1
d['b'] = 2
# at this point, the changes to d are not yet synced, but by
# reassigning the dictionary, the proxy is notified of the change
lproxy[0] = d


7.1. Namespace objects


A namespace object has no public methods, but does have writable attributes. Its representation shows the values of its 


attributes.


However, when using a proxy for a namespace object, an attribute beginning with '_' will be an attribute of the proxy and not 


an attribute of the referent:


>>>
>>> manager = multiprocessing.Manager()
>>> Global = manager.Namespace()
>>> Global.x = 10
>>> Global.y = 'hello'
>>> Global._z = 12.3    # this is an attribute of the proxy
>>> print Global
Namespace(x=10, y='hello')


7.2. Customized managers


To create one’s own manager, one creates a subclass of BaseManager and uses the register() classmethod to register new types 


or callables with the manager class. For example:


from multiprocessing.managers import BaseManager


class MathsClass(object):
    def add(self, x, y):
        return x + y
    def mul(self, x, y):
        return x * y


class MyManager(BaseManager):
    pass


MyManager.register('Maths', MathsClass)


if __name__ == '__main__':
    manager = MyManager()
    manager.start()
    maths = manager.Maths()
    print maths.add(4, 3)         # prints 7
    print maths.mul(7, 8)         # prints 56


7.3. Using a remote manager


It is possible to run a manager server on one machine and have clients use it from other machines (assuming that the 


firewalls involved allow it).


Running the following commands creates a server for a single shared queue which remote clients can access:


>>>
>>> from multiprocessing.managers import BaseManager
>>> import Queue
>>> queue = Queue.Queue()
>>> class QueueManager(BaseManager): pass
>>> QueueManager.register('get_queue', callable=lambda:queue)
>>> m = QueueManager(address=('', 50000), authkey='abracadabra')
>>> s = m.get_server()
>>> s.serve_forever()
One client can access the server as follows:


>>>
>>> from multiprocessing.managers import BaseManager
>>> class QueueManager(BaseManager): pass
>>> QueueManager.register('get_queue')
>>> m = QueueManager(address=('foo.bar.org', 50000), authkey='abracadabra')
>>> m.connect()
>>> queue = m.get_queue()
>>> queue.put('hello')
Another client can also use it:


>>>
>>> from multiprocessing.managers import BaseManager
>>> class QueueManager(BaseManager): pass
>>> QueueManager.register('get_queue')
>>> m = QueueManager(address=('foo.bar.org', 50000), authkey='abracadabra')
>>> m.connect()
>>> queue = m.get_queue()
>>> queue.get()
'hello'
Local processes can also access that queue, using the code from above on the client to access it remotely:


>>>
>>> from multiprocessing import Process, Queue
>>> from multiprocessing.managers import BaseManager
>>> class Worker(Process):
...     def __init__(self, q):
...         self.q = q
...         super(Worker, self).__init__()
...     def run(self):
...         self.q.put('local hello')
...
>>> queue = Queue()
>>> w = Worker(queue)
>>> w.start()
>>> class QueueManager(BaseManager): pass
...
>>> QueueManager.register('get_queue', callable=lambda: queue)
>>> m = QueueManager(address=('', 50000), authkey='abracadabra')
>>> s = m.get_server()
>>> s.serve_forever()


8. 代理对象
A proxy is an object which refers to a shared object which lives (presumably) in a different process. The shared object is 


said to be the referent of the proxy. Multiple proxy objects may have the same referent.


A proxy object has methods which invoke corresponding methods of its referent (although not every method of the referent will 


necessarily be available through the proxy). A proxy can usually be used in most of the same ways that its referent can:


>>>
>>> from multiprocessing import Manager
>>> manager = Manager()
>>> l = manager.list([i*i for i in range(10)])
>>> print l
[0, 1, 4, 9, 16, 25, 36, 49, 64, 81]
>>> print repr(l)

>>> l[4]
16
>>> l[2:5]
[4, 9, 16]
Notice that applying str() to a proxy will return the representation of the referent, whereas applying repr() will return the 


representation of the proxy.


An important feature of proxy objects is that they are picklable so they can be passed between processes. Note, however, that 


if a proxy is sent to the corresponding manager’s process then unpickling it will produce the referent itself. This means, 


for example, that one shared object can contain a second:


>>>
>>> a = manager.list()
>>> b = manager.list()
>>> a.append(b)         # referent of a now contains referent of b
>>> print a, b
[[]] []
>>> b.append('hello')
>>> print a, b
[['hello']] ['hello']
Note The proxy types in multiprocessing do nothing to support comparisons by value. So, for instance, we have:
>>>
>>> manager.list([1,2,3]) == [1,2,3]
False
One should just use a copy of the referent instead when making comparisons.
class multiprocessing.managers.BaseProxy
Proxy objects are instances of subclasses of BaseProxy.


_callmethod(methodname[, args[, kwds]])
Call and return the result of a method of the proxy’s referent.


If proxy is a proxy whose referent is obj then the expression


proxy._callmethod(methodname, args, kwds)
will evaluate the expression


getattr(obj, methodname)(*args, **kwds)
in the manager’s process.


The returned value will be a copy of the result of the call or a proxy to a new shared object – see documentation for the 


method_to_typeid argument of BaseManager.register().


If an exception is raised by the call, then is re-raised by _callmethod(). If some other exception is raised in the manager’


s process then this is converted into a RemoteError exception and is raised by _callmethod().


Note in particular that an exception will be raised if methodname has not been exposed


An example of the usage of _callmethod():


>>>
>>> l = manager.list(range(10))
>>> l._callmethod('__len__')
10
>>> l._callmethod('__getslice__', (2, 7))   # equiv to `l[2:7]`
[2, 3, 4, 5, 6]
>>> l._callmethod('__getitem__', (20,))     # equiv to `l[20]`
Traceback (most recent call last):
...
IndexError: list index out of range
_getvalue()
Return a copy of the referent.


If the referent is unpicklable then this will raise an exception.


__repr__()
Return a representation of the proxy object.


__str__()
Return the representation of the referent.


8.1. Cleanup


A proxy object uses a weakref callback so that when it gets garbage collected it deregisters itself from the manager which 


owns its referent.


A shared object gets deleted from the manager process when there are no longer any proxies referring to it.


9. Process Pools
One can create a pool of processes which will carry out tasks submitted to it with the Pool class.


class multiprocessing.Pool([processes[, initializer[, initargs[, maxtasksperchild]]]])
A process pool object which controls a pool of worker processes to which jobs can be submitted. It supports asynchronous 


results with timeouts and callbacks and has a parallel map implementation.


processes is the number of worker processes to use. If processes is None then the number returned by cpu_count() is used. If 


initializer is not None then each worker process will call initializer(*initargs) when it starts.


Note that the methods of the pool object should only be called by the process which created the pool.


New in version 2.7: maxtasksperchild is the number of tasks a worker process can complete before it will exit and be replaced 


with a fresh worker process, to enable unused resources to be freed. The default maxtasksperchild is None, which means worker 


processes will live as long as the pool.


Note Worker processes within a Pool typically live for the complete duration of the Pool’s work queue. A frequent pattern 


found in other systems (such as Apache, mod_wsgi, etc) to free resources held by workers is to allow a worker within a pool 


to complete only a set amount of work before being exiting, being cleaned up and a new process spawned to replace the old 


one. The maxtasksperchild argument to the Pool exposes this ability to the end user.
apply(func[, args[, kwds]])
Equivalent of the apply() built-in function. It blocks until the result is ready, so apply_async() is better suited for 


performing work in parallel. Additionally, func is only executed in one of the workers of the pool.


apply_async(func[, args[, kwds[, callback]]])
A variant of the apply() method which returns a result object.


If callback is specified then it should be a callable which accepts a single argument. When the result becomes ready callback 


is applied to it (unless the call failed). callback should complete immediately since otherwise the thread which handles the 


results will get blocked.


map(func, iterable[, chunksize])
A parallel equivalent of the map() built-in function (it supports only one iterable argument though). It blocks until the 


result is ready.


This method chops the iterable into a number of chunks which it submits to the process pool as separate tasks. The 


(approximate) size of these chunks can be specified by setting chunksize to a positive integer.


map_async(func, iterable[, chunksize[, callback]])
A variant of the map() method which returns a result object.


If callback is specified then it should be a callable which accepts a single argument. When the result becomes ready callback 


is applied to it (unless the call failed). callback should complete immediately since otherwise the thread which handles the 


results will get blocked.


imap(func, iterable[, chunksize])
An equivalent of itertools.imap().


The chunksize argument is the same as the one used by the map() method. For very long iterables using a large value for 


chunksize can make the job complete much faster than using the default value of 1.


Also if chunksize is 1 then the next() method of the iterator returned by the imap() method has an optional timeout 


parameter: next(timeout) will raise multiprocessing.TimeoutError if the result cannot be returned within timeout seconds.


imap_unordered(func, iterable[, chunksize])
The same as imap() except that the ordering of the results from the returned iterator should be considered arbitrary. (Only 


when there is only one worker process is the order guaranteed to be “correct”.)


close()
Prevents any more tasks from being submitted to the pool. Once all the tasks have been completed the worker processes will 


exit.


terminate()
Stops the worker processes immediately without completing outstanding work. When the pool object is garbage collected 


terminate() will be called immediately.


join()
Wait for the worker processes to exit. One must call close() or terminate() before using join().


class multiprocessing.pool.AsyncResult
The class of the result returned by Pool.apply_async() and Pool.map_async().


get([timeout])
Return the result when it arrives. If timeout is not None and the result does not arrive within timeout seconds then 


multiprocessing.TimeoutError is raised. If the remote call raised an exception then that exception will be reraised by get().


wait([timeout])
Wait until the result is available or until timeout seconds pass.


ready()
Return whether the call has completed.


successful()
Return whether the call completed without raising an exception. Will raise AssertionError if the result is not ready.


The following example demonstrates the use of a pool:


from multiprocessing import Pool


def f(x):
    return x*x


if __name__ == '__main__':
    pool = Pool(processes=4)              # start 4 worker processes


    result = pool.apply_async(f, (10,))    # evaluate "f(10)" asynchronously
    print result.get(timeout=1)           # prints "100" unless your computer is *very* slow


    print pool.map(f, range(10))          # prints "[0, 1, 4,..., 81]"


    it = pool.imap(f, range(10))
    print it.next()                       # prints "0"
    print it.next()                       # prints "1"
    print it.next(timeout=1)              # prints "4" unless your computer is *very* slow


    import time
    result = pool.apply_async(time.sleep, (10,))
    print result.get(timeout=1)           # raises TimeoutError


10. Listeners and Clients
Usually message passing between processes is done using queues or by using Connection objects returned by Pipe().


However, the multiprocessing.connection module allows some extra flexibility. It basically gives a high level message 


oriented API for dealing with sockets or Windows named pipes, and also has support for digest authentication using the hmac 


module.


multiprocessing.connection.deliver_challenge(connection, authkey)
Send a randomly generated message to the other end of the connection and wait for a reply.


If the reply matches the digest of the message using authkey as the key then a welcome message is sent to the other end of 


the connection. Otherwise AuthenticationError is raised.


multiprocessing.connection.answer_challenge(connection, authkey)
Receive a message, calculate the digest of the message using authkey as the key, and then send the digest back.


If a welcome message is not received, then AuthenticationError is raised.


multiprocessing.connection.Client(address[, family[, authenticate[, authkey]]])
Attempt to set up a connection to the listener which is using address address, returning a Connection.


The type of the connection is determined by family argument, but this can generally be omitted since it can usually be 


inferred from the format of address. (See Address Formats)


If authenticate is True or authkey is a string then digest authentication is used. The key used for authentication will be 


either authkey or current_process().authkey) if authkey is None. If authentication fails then AuthenticationError is raised. 


See Authentication keys.


class multiprocessing.connection.Listener([address[, family[, backlog[, authenticate[, authkey]]]]])
A wrapper for a bound socket or Windows named pipe which is ‘listening’ for connections.


address is the address to be used by the bound socket or named pipe of the listener object.


Note If an address of ‘0.0.0.0’ is used, the address will not be a connectable end point on Windows. If you require a 


connectable end-point, you should use ‘127.0.0.1’.
family is the type of socket (or named pipe) to use. This can be one of the strings 'AF_INET' (for a TCP socket), 'AF_UNIX' 


(for a Unix domain socket) or 'AF_PIPE' (for a Windows named pipe). Of these only the first is guaranteed to be available. If 


family is None then the family is inferred from the format of address. If address is also None then a default is chosen. This 


default is the family which is assumed to be the fastest available. See Address Formats. Note that if family is 'AF_UNIX' and 


address is None then the socket will be created in a private temporary directory created using tempfile.mkstemp().


If the listener object uses a socket then backlog (1 by default) is passed to the listen() method of the socket once it has 


been bound.


If authenticate is True (False by default) or authkey is not None then digest authentication is used.


If authkey is a string then it will be used as the authentication key; otherwise it must be None.


If authkey is None and authenticate is True then current_process().authkey is used as the authentication key. If authkey is 


None and authenticate is False then no authentication is done. If authentication fails then AuthenticationError is raised. 


See Authentication keys.


accept()
Accept a connection on the bound socket or named pipe of the listener object and return a Connection object. If 


authentication is attempted and fails, then AuthenticationError is raised.


close()
Close the bound socket or named pipe of the listener object. This is called automatically when the listener is garbage 


collected. However it is advisable to call it explicitly.


Listener objects have the following read-only properties:


address
The address which is being used by the Listener object.


last_accepted
The address from which the last accepted connection came. If this is unavailable then it is None.


The module defines two exceptions:


exception multiprocessing.connection.AuthenticationError
Exception raised when there is an authentication error.


Examples


The following server code creates a listener which uses 'secret password' as an authentication key. It then waits for a 


connection and sends some data to the client:


from multiprocessing.connection import Listener
from array import array


address = ('localhost', 6000)     # family is deduced to be 'AF_INET'
listener = Listener(address, authkey='secret password')


conn = listener.accept()
print 'connection accepted from', listener.last_accepted


conn.send([2.25, None, 'junk', float])


conn.send_bytes('hello')


conn.send_bytes(array('i', [42, 1729]))


conn.close()
listener.close()
The following code connects to the server and receives some data from the server:


from multiprocessing.connection import Client
from array import array


address = ('localhost', 6000)
conn = Client(address, authkey='secret password')


print conn.recv()                 # => [2.25, None, 'junk', float]


print conn.recv_bytes()            # => 'hello'


arr = array('i', [0, 0, 0, 0, 0])
print conn.recv_bytes_into(arr)     # => 8
print arr                         # => array('i', [42, 1729, 0, 0, 0])


conn.close()


10.1. Address Formats


An 'AF_INET' address is a tuple of the form (hostname, port) where hostname is a string and port is an integer.


An 'AF_UNIX' address is a string representing a filename on the filesystem.


An 'AF_PIPE' address is a string of the form
r'\\.\pipe\PipeName'. To use Client() to connect to a named pipe on a remote computer called ServerName one should use an 


address of the form r'\\ServerName\pipe\PipeName' instead.


Note that any string beginning with two backslashes is assumed by default to be an 'AF_PIPE' address rather than an 'AF_UNIX' 


address.


11. Authentication keys
When one uses Connection.recv, the data received is automatically unpickled. Unfortunately unpickling data from an untrusted 


source is a security risk. Therefore Listener and Client() use the hmac module to provide digest authentication.


An authentication key is a string which can be thought of as a password: once a connection is established both ends will 


demand proof that the other knows the authentication key. (Demonstrating that both ends are using the same key does not 


involve sending the key over the connection.)


If authentication is requested but do authentication key is specified then the return value of current_process().authkey is 


used (see Process). This value will automatically inherited by any Process object that the current process creates. This 


means that (by default) all processes of a multi-process program will share a single authentication key which can be used 


when setting up connections between themselves.


Suitable authentication keys can also be generated by using os.urandom().


12. Logging
Some support for logging is available. Note, however, that the logging package does not use process shared locks so it is 


possible (depending on the handler type) for messages from different processes to get mixed up.


multiprocessing.get_logger()
Returns the logger used by multiprocessing. If necessary, a new one will be created.


When first created the logger has level logging.NOTSET and no default handler. Messages sent to this logger will not by 


default propagate to the root logger.


Note that on Windows child processes will only inherit the level of the parent process’s logger – any other customization 


of the logger will not be inherited.


multiprocessing.log_to_stderr()
This function performs a call to get_logger() but in addition to returning the logger created by get_logger, it adds a 


handler which sends output to sys.stderr using format '[%(levelname)s/%(processName)s] %(message)s'.


Below is an example session with logging turned on:


>>>
>>> import multiprocessing, logging
>>> logger = multiprocessing.log_to_stderr()
>>> logger.setLevel(logging.INFO)
>>> logger.warning('doomed')
[WARNING/MainProcess] doomed
>>> m = multiprocessing.Manager()
[INFO/SyncManager-...] child process calling self.run()
[INFO/SyncManager-...] created temp directory /.../pymp-...
[INFO/SyncManager-...] manager serving at '/.../listener-...'
>>> del m
[INFO/MainProcess] sending shutdown message to manager
[INFO/SyncManager-...] manager exiting with exitcode 0
In addition to having these two logging functions, the multiprocessing also exposes two additional logging level attributes. 


These are SUBWARNING and SUBDEBUG. The table below illustrates where theses fit in the normal level hierarchy.


Level Numeric value
SUBWARNING 25
SUBDEBUG 5
For a full table of logging levels, see the logging module.


These additional logging levels are used primarily for certain debug messages within the multiprocessing module. Below is the 


same example as above, except with SUBDEBUG enabled:


>>>
>>> import multiprocessing, logging
>>> logger = multiprocessing.log_to_stderr()
>>> logger.setLevel(multiprocessing.SUBDEBUG)
>>> logger.warning('doomed')
[WARNING/MainProcess] doomed
>>> m = multiprocessing.Manager()
[INFO/SyncManager-...] child process calling self.run()
[INFO/SyncManager-...] created temp directory /.../pymp-...
[INFO/SyncManager-...] manager serving at '/.../pymp-djGBXN/listener-...'
>>> del m
[SUBDEBUG/MainProcess] finalizer calling ...
[INFO/MainProcess] sending shutdown message to manager
[DEBUG/SyncManager-...] manager received shutdown message
[SUBDEBUG/SyncManager-...] calling <finalize object,="" callback="unlink," ...
[SUBDEBUG/SyncManager-...] finalizer calling ...
[SUBDEBUG/SyncManager-...] calling 
[SUBDEBUG/SyncManager-...] finalizer calling ...
[INFO/SyncManager-...] manager exiting with exitcode 0


13. The multiprocessing.dummy module
multiprocessing.dummy replicates the API of multiprocessing but is no more than a wrapper around the threading module.


三、 Programming guidelines
There are certain guidelines and idioms which should be adhered to when using multiprocessing.


1. All platforms
Avoid shared state


As far as possible one should try to avoid shifting large amounts of data between processes.


It is probably best to stick to using queues or pipes for communication between processes rather than using the lower level 


synchronization primitives from the threading module.


Picklability


Ensure that the arguments to the methods of proxies are picklable.
Thread safety of proxies


Do not use a proxy object from more than one thread unless you protect it with a lock.


(There is never a problem with different processes using the same proxy.)


Joining zombie processes


On Unix when a process finishes but has not been joined it becomes a zombie. There should never be very many because each 


time a new process starts (or active_children() is called) all completed processes which have not yet been joined will be 


joined. Also calling a finished process’s Process.is_alive will join the process. Even so it is probably good practice to 


explicitly join all the processes that you start.
Better to inherit than pickle/unpickle


On Windows many types from multiprocessing need to be picklable so that child processes can use them. However, one should 


generally avoid sending shared objects to other processes using pipes or queues. Instead you should arrange the program so 


that a process which needs access to a shared resource created elsewhere can inherit it from an ancestor process.
Avoid terminating processes


Using the Process.terminate method to stop a process is liable to cause any shared resources (such as locks, semaphores, 


pipes and queues) currently being used by the process to become broken or unavailable to other processes.


Therefore it is probably best to only consider using Process.terminate on processes which never use any shared resources.


Joining processes that use queues


Bear in mind that a process that has put items in a queue will wait before terminating until all the buffered items are fed 


by the “feeder” thread to the underlying pipe. (The child process can call the cancel_join_thread() method of the queue to 


avoid this behaviour.)


This means that whenever you use a queue you need to make sure that all items which have been put on the queue will 


eventually be removed before the process is joined. Otherwise you cannot be sure that processes which have put items on the 


queue will terminate. Remember also that non-daemonic processes will be automatically be joined.


An example which will deadlock is the following:


from multiprocessing import Process, Queue


def f(q):
    q.put('X' * 1000000)


if __name__ == '__main__':
    queue = Queue()
    p = Process(target=f, args=(queue,))
    p.start()
    p.join()                    # this deadlocks
    obj = queue.get()
A fix here would be to swap the last two lines round (or simply remove the p.join() line).


Explicitly pass resources to child processes


On Unix a child process can make use of a shared resource created in a parent process using a global resource. However, it is 


better to pass the object as an argument to the constructor for the child process.


Apart from making the code (potentially) compatible with Windows this also ensures that as long as the child process is still 


alive the object will not be garbage collected in the parent process. This might be important if some resource is freed when 


the object is garbage collected in the parent process.


So for instance


from multiprocessing import Process, Lock


def f():
    ... do something using "lock" ...


if __name__ == '__main__':
   lock = Lock()
   for i in range(10):
        Process(target=f).start()
should be rewritten as


from multiprocessing import Process, Lock


def f(l):
    ... do something using "l" ...


if __name__ == '__main__':
   lock = Lock()
   for i in range(10):
        Process(target=f, args=(lock,)).start()
Beware of replacing sys.stdin with a “file like object”


multiprocessing originally unconditionally called:


os.close(sys.stdin.fileno())
in the multiprocessing.Process._bootstrap() method — this resulted in issues with processes-in-processes. This has been 


changed to:


sys.stdin.close()
sys.stdin = open(os.devnull)
Which solves the fundamental issue of processes colliding with each other resulting in a bad file descriptor error, but 


introduces a potential danger to applications which replace sys.stdin() with a “file-like object” with output buffering. 


This danger is that if multiple processes call close() on this file-like object, it could result in the same data being 


flushed to the object multiple times, resulting in corruption.


If you write a file-like object and implement your own caching, you can make it fork-safe by storing the pid whenever you 


append to the cache, and discarding the cache when the pid changes. For example:


@property
def cache(self):
    pid = os.getpid()
    if pid != self._pid:
        self._pid = pid
        self._cache = []
    return self._cache
For more information, see issue 5155, issue 5313 and issue 5331


2. Windows
Since Windows lacks os.fork() it has a few extra restrictions:


More picklability


Ensure that all arguments to Process.__init__() are picklable. This means, in particular, that bound or unbound methods 


cannot be used directly as the target argument on Windows — just define a function and use that instead.


Also, if you subclass Process then make sure that instances will be picklable when the Process.start method is called.


Global variables


Bear in mind that if code run in a child process tries to access a global variable, then the value it sees (if any) may not 


be the same as the value in the parent process at the time that Process.start was called.


However, global variables which are just module level constants cause no problems.


Safe importing of main module


Make sure that the main module can be safely imported by a new Python interpreter without causing unintended side effects 


(such a starting a new process).


For example, under Windows running the following module would fail with a RuntimeError:


from multiprocessing import Process


def foo():
    print 'hello'


p = Process(target=foo)
p.start()
Instead one should protect the “entry point” of the program by using if __name__ == '__main__': as follows:


from multiprocessing import Process, freeze_support


def foo():
    print 'hello'


if __name__ == '__main__':
    freeze_support()
    p = Process(target=foo)
    p.start()
(The freeze_support() line can be omitted if the program will be run normally instead of frozen.)


This allows the newly spawned Python interpreter to safely import the module and then run the module’s foo() function.


Similar restrictions apply if a pool or manager is created in the main module.


四、 Examples
Demonstration of how to create and use customized managers and proxies:


#
# This module shows how to use arbitrary callables with a subclass of
# `BaseManager`.
#
# Copyright (c) 2006-2008, R Oudkerk
# All rights reserved.
#


from multiprocessing import freeze_support
from multiprocessing.managers import BaseManager, BaseProxy
import operator


##


class Foo(object):
    def f(self):
        print 'you called Foo.f()'
    def g(self):
        print 'you called Foo.g()'
    def _h(self):
        print 'you called Foo._h()'


# A simple generator function
def baz():
    for i in xrange(10):
        yield i*i


# Proxy type for generator objects
class GeneratorProxy(BaseProxy):
    _exposed_ = ('next', '__next__')
    def __iter__(self):
        return self
    def next(self):
        return self._callmethod('next')
    def __next__(self):
        return self._callmethod('__next__')


# Function to return the operator module
def get_operator_module():
    return operator


##


class MyManager(BaseManager):
    pass


# register the Foo class; make `f()` and `g()` accessible via proxy
MyManager.register('Foo1', Foo)


# register the Foo class; make `g()` and `_h()` accessible via proxy
MyManager.register('Foo2', Foo, exposed=('g', '_h'))


# register the generator function baz; use `GeneratorProxy` to make proxies
MyManager.register('baz', baz, proxytype=GeneratorProxy)


# register get_operator_module(); make public functions accessible via proxy
MyManager.register('operator', get_operator_module)


##


def test():
    manager = MyManager()
    manager.start()


    print '-' * 20


    f1 = manager.Foo1()
    f1.f()
    f1.g()
    assert not hasattr(f1, '_h')
    assert sorted(f1._exposed_) == sorted(['f', 'g'])


    print '-' * 20


    f2 = manager.Foo2()
    f2.g()
    f2._h()
    assert not hasattr(f2, 'f')
    assert sorted(f2._exposed_) == sorted(['g', '_h'])


    print '-' * 20


    it = manager.baz()
    for i in it:
        print '<%d>' % i,
    print


    print '-' * 20


    op = manager.operator()
    print 'op.add(23, 45) =', op.add(23, 45)
    print 'op.pow(2, 94) =', op.pow(2, 94)
    print 'op.getslice(range(10), 2, 6) =', op.getslice(range(10), 2, 6)
    print 'op.repeat(range(5), 3) =', op.repeat(range(5), 3)
    print 'op._exposed_ =', op._exposed_


##


if __name__ == '__main__':
    freeze_support()
    test()
Using Pool:


#
# A test of `multiprocessing.Pool` class
#
# Copyright (c) 2006-2008, R Oudkerk
# All rights reserved.
#


import multiprocessing
import time
import random
import sys


#
# Functions used by test code
#


def calculate(func, args):
    result = func(*args)
    return '%s says that %s%s = %s' % (
        multiprocessing.current_process().name,
        func.__name__, args, result
        )


def calculatestar(args):
    return calculate(*args)


def mul(a, b):
    time.sleep(0.5*random.random())
    return a * b


def plus(a, b):
    time.sleep(0.5*random.random())
    return a + b


def f(x):
    return 1.0 / (x-5.0)


def pow3(x):
    return x**3


def noop(x):
    pass


#
# Test code
#


def test():
    print 'cpu_count() = %d\n' % multiprocessing.cpu_count()


    #
    # Create pool
    #


    PROCESSES = 4
    print 'Creating pool with %d processes\n' % PROCESSES
    pool = multiprocessing.Pool(PROCESSES)
    print 'pool = %s' % pool
    print


    #
    # Tests
    #


    TASKS = [(mul, (i, 7)) for i in range(10)] + \
            [(plus, (i, 8)) for i in range(10)]


    results = [pool.apply_async(calculate, t) for t in TASKS]
    imap_it = pool.imap(calculatestar, TASKS)
    imap_unordered_it = pool.imap_unordered(calculatestar, TASKS)


    print 'Ordered results using pool.apply_async():'
    for r in results:
        print '\t', r.get()
    print


    print 'Ordered results using pool.imap():'
    for x in imap_it:
        print '\t', x
    print


    print 'Unordered results using pool.imap_unordered():'
    for x in imap_unordered_it:
        print '\t', x
    print


    print 'Ordered results using pool.map() --- will block till complete:'
    for x in pool.map(calculatestar, TASKS):
        print '\t', x
    print


    #
    # Simple benchmarks
    #


    N = 100000
    print 'def pow3(x): return x**3'


    t = time.time()
    A = map(pow3, xrange(N))
    print '\tmap(pow3, xrange(%d)):\n\t\t%s seconds' % \
          (N, time.time() - t)


    t = time.time()
    B = pool.map(pow3, xrange(N))
    print '\tpool.map(pow3, xrange(%d)):\n\t\t%s seconds' % \
          (N, time.time() - t)


    t = time.time()
    C = list(pool.imap(pow3, xrange(N), chunksize=N//8))
    print '\tlist(pool.imap(pow3, xrange(%d), chunksize=%d)):\n\t\t%s' \
          ' seconds' % (N, N//8, time.time() - t)


    assert A == B == C, (len(A), len(B), len(C))
    print


    L = [None] * 1000000
    print 'def noop(x): pass'
    print 'L = [None] * 1000000'


    t = time.time()
    A = map(noop, L)
    print '\tmap(noop, L):\n\t\t%s seconds' % \
          (time.time() - t)


    t = time.time()
    B = pool.map(noop, L)
    print '\tpool.map(noop, L):\n\t\t%s seconds' % \
          (time.time() - t)


    t = time.time()
    C = list(pool.imap(noop, L, chunksize=len(L)//8))
    print '\tlist(pool.imap(noop, L, chunksize=%d)):\n\t\t%s seconds' % \
          (len(L)//8, time.time() - t)


    assert A == B == C, (len(A), len(B), len(C))
    print


    del A, B, C, L


    #
    # Test error handling
    #


    print 'Testing error handling:'


    try:
        print pool.apply(f, (5,))
    except ZeroDivisionError:
        print '\tGot ZeroDivisionError as expected from pool.apply()'
    else:
        raise AssertionError('expected ZeroDivisionError')


    try:
        print pool.map(f, range(10))
    except ZeroDivisionError:
        print '\tGot ZeroDivisionError as expected from pool.map()'
    else:
        raise AssertionError('expected ZeroDivisionError')


    try:
        print list(pool.imap(f, range(10)))
    except ZeroDivisionError:
        print '\tGot ZeroDivisionError as expected from list(pool.imap())'
    else:
        raise AssertionError('expected ZeroDivisionError')


    it = pool.imap(f, range(10))
    for i in range(10):
        try:
            x = it.next()
        except ZeroDivisionError:
            if i == 5:
                pass
        except StopIteration:
            break
        else:
            if i == 5:
                raise AssertionError('expected ZeroDivisionError')


    assert i == 9
    print '\tGot ZeroDivisionError as expected from IMapIterator.next()'
    print


    #
    # Testing timeouts
    #


    print 'Testing ApplyResult.get() with timeout:',
    res = pool.apply_async(calculate, TASKS[0])
    while 1:
        sys.stdout.flush()
        try:
            sys.stdout.write('\n\t%s' % res.get(0.02))
            break
        except multiprocessing.TimeoutError:
            sys.stdout.write('.')
    print
    print


    print 'Testing IMapIterator.next() with timeout:',
    it = pool.imap(calculatestar, TASKS)
    while 1:
        sys.stdout.flush()
        try:
            sys.stdout.write('\n\t%s' % it.next(0.02))
        except StopIteration:
            break
        except multiprocessing.TimeoutError:
            sys.stdout.write('.')
    print
    print


    #
    # Testing callback
    #


    print 'Testing callback:'


    A = []
    B = [56, 0, 1, 8, 27, 64, 125, 216, 343, 512, 729]


    r = pool.apply_async(mul, (7, 8), callback=A.append)
    r.wait()


    r = pool.map_async(pow3, range(10), callback=A.extend)
    r.wait()


    if A == B:
        print '\tcallbacks succeeded\n'
    else:
        print '\t*** callbacks failed\n\t\t%s != %s\n' % (A, B)


    #
    # Check there are no outstanding tasks
    #


    assert not pool._cache, 'cache = %r' % pool._cache


    #
    # Check close() methods
    #


    print 'Testing close():'


    for worker in pool._pool:
        assert worker.is_alive()


    result = pool.apply_async(time.sleep, [0.5])
    pool.close()
    pool.join()


    assert result.get() is None


    for worker in pool._pool:
        assert not worker.is_alive()


    print '\tclose() succeeded\n'


    #
    # Check terminate() method
    #


    print 'Testing terminate():'


    pool = multiprocessing.Pool(2)
    DELTA = 0.1
    ignore = pool.apply(pow3, [2])
    results = [pool.apply_async(time.sleep, [DELTA]) for i in range(100)]
    pool.terminate()
    pool.join()


    for worker in pool._pool:
        assert not worker.is_alive()


    print '\tterminate() succeeded\n'


    #
    # Check garbage collection
    #


    print 'Testing garbage collection:'


    pool = multiprocessing.Pool(2)
    DELTA = 0.1
    processes = pool._pool
    ignore = pool.apply(pow3, [2])
    results = [pool.apply_async(time.sleep, [DELTA]) for i in range(100)]


    results = pool = None


    time.sleep(DELTA * 2)


    for worker in processes:
        assert not worker.is_alive()


    print '\tgarbage collection succeeded\n'




if __name__ == '__main__':
    multiprocessing.freeze_support()


    assert len(sys.argv) in (1, 2)


    if len(sys.argv) == 1 or sys.argv[1] == 'processes':
        print ' Using processes '.center(79, '-')
    elif sys.argv[1] == 'threads':
        print ' Using threads '.center(79, '-')
        import multiprocessing.dummy as multiprocessing
    else:
        print 'Usage:\n\t%s [processes | threads]' % sys.argv[0]
        raise SystemExit(2)


    test()
Synchronization types like locks, conditions and queues:


#
# A test file for the `multiprocessing` package
#
# Copyright (c) 2006-2008, R Oudkerk
# All rights reserved.
#


import time, sys, random
from Queue import Empty


import multiprocessing               # may get overwritten




#### TEST_VALUE


def value_func(running, mutex):
    random.seed()
    time.sleep(random.random()*4)


    mutex.acquire()
    print '\n\t\t\t' + str(multiprocessing.current_process()) + ' has finished'
    running.value -= 1
    mutex.release()


def test_value():
    TASKS = 10
    running = multiprocessing.Value('i', TASKS)
    mutex = multiprocessing.Lock()


    for i in range(TASKS):
        p = multiprocessing.Process(target=value_func, args=(running, mutex))
        p.start()


    while running.value > 0:
        time.sleep(0.08)
        mutex.acquire()
        print running.value,
        sys.stdout.flush()
        mutex.release()


    print
    print 'No more running processes'




#### TEST_QUEUE


def queue_func(queue):
    for i in range(30):
        time.sleep(0.5 * random.random())
        queue.put(i*i)
    queue.put('STOP')


def test_queue():
    q = multiprocessing.Queue()


    p = multiprocessing.Process(target=queue_func, args=(q,))
    p.start()


    o = None
    while o != 'STOP':
        try:
            o = q.get(timeout=0.3)
            print o,
            sys.stdout.flush()
        except Empty:
            print 'TIMEOUT'


    print




#### TEST_CONDITION


def condition_func(cond):
    cond.acquire()
    print '\t' + str(cond)
    time.sleep(2)
    print '\tchild is notifying'
    print '\t' + str(cond)
    cond.notify()
    cond.release()


def test_condition():
    cond = multiprocessing.Condition()


    p = multiprocessing.Process(target=condition_func, args=(cond,))
    print cond


    cond.acquire()
    print cond
    cond.acquire()
    print cond


    p.start()


    print 'main is waiting'
    cond.wait()
    print 'main has woken up'


    print cond
    cond.release()
    print cond
    cond.release()


    p.join()
    print cond




#### TEST_SEMAPHORE


def semaphore_func(sema, mutex, running):
    sema.acquire()


    mutex.acquire()
    running.value += 1
    print running.value, 'tasks are running'
    mutex.release()


    random.seed()
    time.sleep(random.random()*2)


    mutex.acquire()
    running.value -= 1
    print '%s has finished' % multiprocessing.current_process()
    mutex.release()


    sema.release()


def test_semaphore():
    sema = multiprocessing.Semaphore(3)
    mutex = multiprocessing.RLock()
    running = multiprocessing.Value('i', 0)


    processes = [
        multiprocessing.Process(target=semaphore_func,
                                args=(sema, mutex, running))
        for i in range(10)
        ]


    for p in processes:
        p.start()


    for p in processes:
        p.join()




#### TEST_JOIN_TIMEOUT


def join_timeout_func():
    print '\tchild sleeping'
    time.sleep(5.5)
    print '\n\tchild terminating'


def test_join_timeout():
    p = multiprocessing.Process(target=join_timeout_func)
    p.start()


    print 'waiting for process to finish'


    while 1:
        p.join(timeout=1)
        if not p.is_alive():
            break
        print '.',
        sys.stdout.flush()




#### TEST_EVENT


def event_func(event):
    print '\t%r is waiting' % multiprocessing.current_process()
    event.wait()
    print '\t%r has woken up' % multiprocessing.current_process()


def test_event():
    event = multiprocessing.Event()


    processes = [multiprocessing.Process(target=event_func, args=(event,))
                 for i in range(5)]


    for p in processes:
        p.start()


    print 'main is sleeping'
    time.sleep(2)


    print 'main is setting event'
    event.set()


    for p in processes:
        p.join()




#### TEST_SHAREDVALUES


def sharedvalues_func(values, arrays, shared_values, shared_arrays):
    for i in range(len(values)):
        v = values[i][1]
        sv = shared_values[i].value
        assert v == sv


    for i in range(len(values)):
        a = arrays[i][1]
        sa = list(shared_arrays[i][:])
        assert a == sa


    print 'Tests passed'


def test_sharedvalues():
    values = [
        ('i', 10),
        ('h', -2),
        ('d', 1.25)
        ]
    arrays = [
        ('i', range(100)),
        ('d', [0.25 * i for i in range(100)]),
        ('H', range(1000))
        ]


    shared_values = [multiprocessing.Value(id, v) for id, v in values]
    shared_arrays = [multiprocessing.Array(id, a) for id, a in arrays]


    p = multiprocessing.Process(
        target=sharedvalues_func,
        args=(values, arrays, shared_values, shared_arrays)
        )
    p.start()
    p.join()


    assert p.exitcode == 0




####


def test(namespace=multiprocessing):
    global multiprocessing


    multiprocessing = namespace


    for func in [ test_value, test_queue, test_condition,
                  test_semaphore, test_join_timeout, test_event,
                  test_sharedvalues ]:


        print '\n\t######## %s\n' % func.__name__
        func()


    ignore = multiprocessing.active_children()      # cleanup any old processes
    if hasattr(multiprocessing, '_debug_info'):
        info = multiprocessing._debug_info()
        if info:
            print info
            raise ValueError('there should be no positive refcounts left')




if __name__ == '__main__':
    multiprocessing.freeze_support()


    assert len(sys.argv) in (1, 2)


    if len(sys.argv) == 1 or sys.argv[1] == 'processes':
        print ' Using processes '.center(79, '-')
        namespace = multiprocessing
    elif sys.argv[1] == 'manager':
        print ' Using processes and a manager '.center(79, '-')
        namespace = multiprocessing.Manager()
        namespace.Process = multiprocessing.Process
        namespace.current_process = multiprocessing.current_process
        namespace.active_children = multiprocessing.active_children
    elif sys.argv[1] == 'threads':
        print ' Using threads '.center(79, '-')
        import multiprocessing.dummy as namespace
    else:
        print 'Usage:\n\t%s [processes | manager | threads]' % sys.argv[0]
        raise SystemExit(2)


    test(namespace)
An example showing how to use queues to feed tasks to a collection of worker processes and collect the results:


#
# Simple example which uses a pool of workers to carry out some tasks.
#
# Notice that the results will probably not come out of the output
# queue in the same in the same order as the corresponding tasks were
# put on the input queue.  If it is important to get the results back
# in the original order then consider using `Pool.map()` or
# `Pool.imap()` (which will save on the amount of code needed anyway).
#
# Copyright (c) 2006-2008, R Oudkerk
# All rights reserved.
#


import time
import random


from multiprocessing import Process, Queue, current_process, freeze_support


#
# Function run by worker processes
#


def worker(input, output):
    for func, args in iter(input.get, 'STOP'):
        result = calculate(func, args)
        output.put(result)


#
# Function used to calculate result
#


def calculate(func, args):
    result = func(*args)
    return '%s says that %s%s = %s' % \
        (current_process().name, func.__name__, args, result)


#
# Functions referenced by tasks
#


def mul(a, b):
    time.sleep(0.5*random.random())
    return a * b


def plus(a, b):
    time.sleep(0.5*random.random())
    return a + b


#
#
#


def test():
    NUMBER_OF_PROCESSES = 4
    TASKS1 = [(mul, (i, 7)) for i in range(20)]
    TASKS2 = [(plus, (i, 8)) for i in range(10)]


    # Create queues
    task_queue = Queue()
    done_queue = Queue()


    # Submit tasks
    for task in TASKS1:
        task_queue.put(task)


    # Start worker processes
    for i in range(NUMBER_OF_PROCESSES):
        Process(target=worker, args=(task_queue, done_queue)).start()


    # Get and print results
    print 'Unordered results:'
    for i in range(len(TASKS1)):
        print '\t', done_queue.get()


    # Add more tasks using `put()`
    for task in TASKS2:
        task_queue.put(task)


    # Get and print some more results
    for i in range(len(TASKS2)):
        print '\t', done_queue.get()


    # Tell child processes to stop
    for i in range(NUMBER_OF_PROCESSES):
        task_queue.put('STOP')




if __name__ == '__main__':
    freeze_support()
    test()
An example of how a pool of worker processes can each run a SimpleHTTPServer.HttpServer instance while sharing a single 


listening socket.


#
# Example where a pool of http servers share a single listening socket
#
# On Windows this module depends on the ability to pickle a socket
# object so that the worker processes can inherit a copy of the server
# object.  (We import `multiprocessing.reduction` to enable this pickling.)
#
# Not sure if we should synchronize access to `socket.accept()` method by
# using a process-shared lock -- does not seem to be necessary.
#
# Copyright (c) 2006-2008, R Oudkerk
# All rights reserved.
#


import os
import sys


from multiprocessing import Process, current_process, freeze_support
from BaseHTTPServer import HTTPServer
from SimpleHTTPServer import SimpleHTTPRequestHandler


if sys.platform == 'win32':
    import multiprocessing.reduction    # make sockets pickable/inheritable




def note(format, *args):
    sys.stderr.write('[%s]\t%s\n' % (current_process().name, format%args))




class RequestHandler(SimpleHTTPRequestHandler):
    # we override log_message() to show which process is handling the request
    def log_message(self, format, *args):
        note(format, *args)


def serve_forever(server):
    note('starting server')
    try:
        server.serve_forever()
    except KeyboardInterrupt:
        pass




def runpool(address, number_of_processes):
    # create a single server object -- children will each inherit a copy
    server = HTTPServer(address, RequestHandler)


    # create child processes to act as workers
    for i in range(number_of_processes-1):
        Process(target=serve_forever, args=(server,)).start()


    # main process also acts as a worker
    serve_forever(server)




def test():
    DIR = os.path.join(os.path.dirname(__file__), '..')
    ADDRESS = ('localhost', 8000)
    NUMBER_OF_PROCESSES = 4


    print 'Serving at http://%s:%d using %d worker processes' % \
          (ADDRESS[0], ADDRESS[1], NUMBER_OF_PROCESSES)
    print 'To exit press Ctrl-' + ['C', 'Break'][sys.platform=='win32']


    os.chdir(DIR)
    runpool(ADDRESS, NUMBER_OF_PROCESSES)




if __name__ == '__main__':
    freeze_support()
    test()
Some simple benchmarks comparing multiprocessing with threading:


#
# Simple benchmarks for the multiprocessing package
#
# Copyright (c) 2006-2008, R Oudkerk
# All rights reserved.
#


import time, sys, multiprocessing, threading, Queue, gc


if sys.platform == 'win32':
    _timer = time.clock
else:
    _timer = time.time


delta = 1




#### TEST_QUEUESPEED


def queuespeed_func(q, c, iterations):
    a = '0' * 256
    c.acquire()
    c.notify()
    c.release()


    for i in xrange(iterations):
        q.put(a)


    q.put('STOP')


def test_queuespeed(Process, q, c):
    elapsed = 0
    iterations = 1


    while elapsed < delta:
        iterations *= 2


        p = Process(target=queuespeed_func, args=(q, c, iterations))
        c.acquire()
        p.start()
        c.wait()
        c.release()


        result = None
        t = _timer()


        while result != 'STOP':
            result = q.get()


        elapsed = _timer() - t


        p.join()


    print iterations, 'objects passed through the queue in', elapsed, 'seconds'
    print 'average number/sec:', iterations/elapsed




#### TEST_PIPESPEED


def pipe_func(c, cond, iterations):
    a = '0' * 256
    cond.acquire()
    cond.notify()
    cond.release()


    for i in xrange(iterations):
        c.send(a)


    c.send('STOP')


def test_pipespeed():
    c, d = multiprocessing.Pipe()
    cond = multiprocessing.Condition()
    elapsed = 0
    iterations = 1


    while elapsed < delta:
        iterations *= 2


        p = multiprocessing.Process(target=pipe_func,
                                    args=(d, cond, iterations))
        cond.acquire()
        p.start()
        cond.wait()
        cond.release()


        result = None
        t = _timer()


        while result != 'STOP':
            result = c.recv()


        elapsed = _timer() - t
        p.join()


    print iterations, 'objects passed through connection in',elapsed,'seconds'
    print 'average number/sec:', iterations/elapsed




#### TEST_SEQSPEED


def test_seqspeed(seq):
    elapsed = 0
    iterations = 1


    while elapsed < delta:
        iterations *= 2


        t = _timer()


        for i in xrange(iterations):
            a = seq[5]


        elapsed = _timer()-t


    print iterations, 'iterations in', elapsed, 'seconds'
    print 'average number/sec:', iterations/elapsed




#### TEST_LOCK


def test_lockspeed(l):
    elapsed = 0
    iterations = 1


    while elapsed < delta:
        iterations *= 2


        t = _timer()


        for i in xrange(iterations):
            l.acquire()
            l.release()


        elapsed = _timer()-t


    print iterations, 'iterations in', elapsed, 'seconds'
    print 'average number/sec:', iterations/elapsed




#### TEST_CONDITION


def conditionspeed_func(c, N):
    c.acquire()
    c.notify()


    for i in xrange(N):
        c.wait()
        c.notify()


    c.release()


def test_conditionspeed(Process, c):
    elapsed = 0
    iterations = 1


    while elapsed < delta:
        iterations *= 2


        c.acquire()
        p = Process(target=conditionspeed_func, args=(c, iterations))
        p.start()


        c.wait()


        t = _timer()


        for i in xrange(iterations):
            c.notify()
            c.wait()


        elapsed = _timer()-t


        c.release()
        p.join()


    print iterations * 2, 'waits in', elapsed, 'seconds'
    print 'average number/sec:', iterations * 2 / elapsed


####


def test():
    manager = multiprocessing.Manager()


    gc.disable()


    print '\n\t######## testing Queue.Queue\n'
    test_queuespeed(threading.Thread, Queue.Queue(),
                    threading.Condition())
    print '\n\t######## testing multiprocessing.Queue\n'
    test_queuespeed(multiprocessing.Process, multiprocessing.Queue(),
                    multiprocessing.Condition())
    print '\n\t######## testing Queue managed by server process\n'
    test_queuespeed(multiprocessing.Process, manager.Queue(),
                    manager.Condition())
    print '\n\t######## testing multiprocessing.Pipe\n'
    test_pipespeed()


    print


    print '\n\t######## testing list\n'
    test_seqspeed(range(10))
    print '\n\t######## testing list managed by server process\n'
    test_seqspeed(manager.list(range(10)))
    print '\n\t######## testing Array("i", ..., lock=False)\n'
    test_seqspeed(multiprocessing.Array('i', range(10), lock=False))
    print '\n\t######## testing Array("i", ..., lock=True)\n'
    test_seqspeed(multiprocessing.Array('i', range(10), lock=True))


    print


    print '\n\t######## testing threading.Lock\n'
    test_lockspeed(threading.Lock())
    print '\n\t######## testing threading.RLock\n'
    test_lockspeed(threading.RLock())
    print '\n\t######## testing multiprocessing.Lock\n'
    test_lockspeed(multiprocessing.Lock())
    print '\n\t######## testing multiprocessing.RLock\n'
    test_lockspeed(multiprocessing.RLock())
    print '\n\t######## testing lock managed by server process\n'
    test_lockspeed(manager.Lock())
    print '\n\t######## testing rlock managed by server process\n'
    test_lockspeed(manager.RLock())


    print


    print '\n\t######## testing threading.Condition\n'
    test_conditionspeed(threading.Thread, threading.Condition())
    print '\n\t######## testing multiprocessing.Condition\n'
    test_conditionspeed(multiprocessing.Process, multiprocessing.Condition())
    print '\n\t######## testing condition managed by a server process\n'
    test_conditionspeed(multiprocessing.Process, manager.Condition())


    gc.enable()


if __name__ == '__main__':
    multiprocessing.freeze_support()
    test()