pystuck 0.8.5

依赖项

• IPython

• rpyc(3.2.3，可能与其他版本兼容)

使用说明

pystuck 在没有客户端连接时不会消耗资源。

run_server 函数所做的一切就是启动一个线程，该线程在 accept 上阻塞（等待客户端连接），因此它可以用于生产环境。

在生产环境中使用 pystuck 有两个缺点

1. 修改变量和模块不是线程安全的。

2. 潜在的安全漏洞 - 没有任何东西可以阻止未授权的用户连接到受权的运行中的 Python 进程并使用远程访问做任何事情。

Python 在做什么？！

test.py 挂起了，你不想知道它在哪吗？

import pystuck; pystuck.run_server().

while True:
    with lock: # could block
         sock.recv(1024) # could block

从 shell 中运行 pystuck 显示有趣的内容

% pystuck
<_MainThread(MainThread, started -1215396160)>
  File "test.py", line 9, in <module>
    with lock: # could block

...

它挂起等待锁！实际上它打印了两个与 pystuck 相关的线程，忽略它们。

谁拥有了锁？！

# it seldom happens that a thread doesn't release the lock or is stuck while holding it.
# we want to know which thread... bear with me now.
rlock.acquire()

再次调用pystuck

% pystuck
<_MainThread(MainThread, started -1215396160)>
  File "test.py", line 9, in <module>
    with lock: # could block

<Thread(Thread-1, started -1219450000)>
  File "test.py", line 12, in <module>
    do_math()

<Thread(Thread-2, started -1219540000)>
  File "test.py", line 14, in <module>
    foo()
  File "test.py", line 20, in foo
    do_network()

use the 'modules' dictionary to access remote modules (like 'os', or '__main__')

In [0]: modules['sys']._current_frames() # gets a mapping between a thread to its frame (top of stack)
Out[0]: {-1215396160: <frame object at 0x8a07154>,
         -1219450000: <frame object at 0x8a29154>,
         -1219540000: <frame object at 0x8b39154>}

In [1]: _[-1215396160] # get the stuck thread's frame
Out[1]: <frame object at 0x8a07154>

In [2]: _.f_locals # local variables of the function
Out[2]: {'lock': <_RLock owner=-1219450000 count=1>}

In [3]: # our stuck thread is probably waiting for thread 1219450000 to finish do_math.. figures

Pyrc能为您做什么

Pyrc是一个用于在进程之间通信Python对象和过程调用的库。以下是您可以对已运行的服务器执行的某些操作。

% pystuck

(stacks appear here...)

In [1]: modules['sys'].stdout = file("/tmp/log.txt", "w") # tunnel the script's stdout to log

In [2]: modules['__main__'].global_var = 100 # change and inspect variables in the __main__ module (the name of the script when invoked like this: python script.py)

In [3]: socket = modules['socket'].socket() # create a socket object opened by the server script!