I have an issue where a method call is blocking and not releasing. Unfortunately, the bug as to why isn't exactly solvable, so the workaround at the moment is to build in a timeout.
I've tried to do this by registering a timer and have it raise an exception to break from the blocked call. However, that raises the exception in the timer thread, not the main thread.
It looks like this right now:
from threading import Timer
def timeoutSocket():
raise InterruptedError
socketDeadlockDetector = Timer(DEADLOCK_TIMEOUT, timeoutSocket)
socketDeadlockDetector.start()
# receive and unpack data
try:
packet = server.receive()
except InterruptedError:
print("Interrupted socket receive, continuing")
continue
socketDeadlockDetector.cancel()
server.receive() is the method that is blocking when it shouldn't. However, when I run this, the socketDeadlockDetector thread interrupts itself, without affecting the original thread.
Is there a way to pass this exception up to the parent?
Timer creates a thread to run the function. It doesn't do you any good to raise an exception because that's not the thread needing interruption. When you hit the timeout, you need to cancel whatever is blocking in the other thread. In this case its a socket, so killing the socket should do.
import struct
def timeoutSocket():
# enable linger with timeout 0 to send RESET on close
server.setsockopt(socket.SOL_SOCKET, socket.SO_LINGER, struct.pack('ii', 1, 0))
server.close()
Related
As a simple example, consider the network equivalent of /dev/zero, below. (Or more realistically, just a web server sending a large file.)
If a client disconnects early, you get a barrage of log messages:
WARNING:asyncio:socket.send() raised exception.
But I'm not finding any way to catch said exception. The hypothetical server continues reading gigabytes from disk and sending them to a dead socket, with no effort on the client's part, and you've got yourself a DoS attack.
The only thing I've found from the docs is to yield from a read, with an empty string indicating closure. But that's no good here because a normal client isn't going to send anything, blocking the write loop.
What's the right way to detect failed writes, or be notified that the TCP connection has been closed, with the streams API or otherwise?
Code:
from asyncio import *
import logging
#coroutine
def client_handler(reader, writer):
while True:
writer.write(bytes(1))
yield from writer.drain()
logging.basicConfig(level=logging.INFO)
loop = get_event_loop()
coro = start_server(client_handler, '', 12345)
server = loop.run_until_complete(coro)
loop.run_forever()
I did some digging into the asyncio source to expand on dano's answer on why the exceptions aren't being raised without explicitly passing control to the event loop. Here's what I've found.
Calling yield from wirter.drain() gives the control over to the StreamWriter.drain coroutine. This coroutine checks for and raises any exceptions that that the StreamReaderProtocol set on the StreamReader. But since we passed control over to drain, the protocol hasn't had the chance to set the exception yet. drain then gives control over to the FlowControlMixin._drain_helper coroutine. This coroutine the returns immediately because some more flags haven't been set yet, and the control ends up back with the coroutine that called yield from wirter.drain().
And so we have gone full circle without giving control to the event loop to allow it handle other coroutines and bubble up the exceptions to writer.drain().
yielding before a drain() gives the transport/protocol a chance to set the appropriate flags and exceptions.
Here's a mock up of what's going on, with all the nested calls collapsed:
import asyncio as aio
def set_exception(ctx, exc):
ctx["exc"] = exc
#aio.coroutine
def drain(ctx):
if ctx["exc"] is not None:
raise ctx["exc"]
return
#aio.coroutine
def client_handler(ctx):
i = 0
while True:
i += 1
print("write", i)
# yield # Uncommenting this allows the loop.call_later call to be scheduled.
yield from drain(ctx)
CTX = {"exc": None}
loop = aio.get_event_loop()
# Set the exception in 5 seconds
loop.call_later(5, set_exception, CTX, Exception("connection lost"))
loop.run_until_complete(client_handler(CTX))
loop.close()
This should probably fixed upstream in the Streams API by the asyncio developers.
This is a little bit strange, but you can actually allow an exception to reach the client_handler coroutine by forcing it to yield control to the event loop for one iteration:
import asyncio
import logging
#asyncio.coroutine
def client_handler(reader, writer):
while True:
writer.write(bytes(1))
yield # Yield to the event loop
yield from writer.drain()
logging.basicConfig(level=logging.INFO)
loop = asyncio.get_event_loop()
coro = asyncio.start_server(client_handler, '', 12345)
server = loop.run_until_complete(coro)
loop.run_forever()
If I do that, I get this output when I kill the client connection:
ERROR:asyncio:Task exception was never retrieved
future: <Task finished coro=<client_handler() done, defined at aio.py:4> exception=ConnectionResetError(104, 'Connection reset by peer')>
Traceback (most recent call last):
File "/usr/lib/python3.4/asyncio/tasks.py", line 238, in _step
result = next(coro)
File "aio.py", line 9, in client_handler
yield from writer.drain()
File "/usr/lib/python3.4/asyncio/streams.py", line 301, in drain
raise exc
File "/usr/lib/python3.4/asyncio/selector_events.py", line 700, in write
n = self._sock.send(data)
ConnectionResetError: [Errno 104] Connection reset by peer
I'm really not quite sure why you need to explicitly let the event loop get control for the exception to get through - don't have time at the moment to dig into it. I assume some bit needs to get flipped to indicate the connection dropped, and calling yield from writer.drain() (which can short-circuit going through the event loop) in a loop is preventing that from happening, but I'm really not sure. If I get a chance to investigate, I'll update the answer with that info.
The stream based API doesn't have a callback you can specify for when the connection is closed. But the Protocol API does, so use it instead: https://docs.python.org/3/library/asyncio-protocol.html#connection-callbacks
I've seen many topics about this particular problem but i still can't figure why i'm not catching a SIGINT in my main Thread.
Here is my code:
def connect(self, retry=100):
tries=retry
logging.info('connecting to %s' % self.path)
while True:
try:
self.sp = serial.Serial(self.path, 115200)
self.pileMessage = pilemessage.Pilemessage()
self.pileData = pilemessage.Pilemessage()
self.reception = reception.Reception(self.sp,self.pileMessage,self.pileData)
self.reception.start()
self.collisionlistener = collisionListener.CollisionListener(self)
self.message = messageThread.Message(self.pileMessage,self.collisionlistener)
self.datastreaminglistener = dataStreamingListener.DataStreamingListener(self)
self.datastreaming = dataStreaming.Data(self.pileData,self.datastreaminglistener)
return
except serial.serialutil.SerialException:
logging.info('retrying')
if not retry:
raise SpheroError('failed to connect after %d tries' % (tries-retry))
retry -= 1
def disconnect(self):
self.reception.stop()
self.message.stop()
self.datastreaming.stop()
while not self.pileData.isEmpty():
self.pileData.pop()
self.datastreaminglistener.remove()
while not self.pileMessage.isEmpty():
self.pileMessage.pop()
self.collisionlistener.remove()
self.sp.close()
if __name__ == '__main__':
import time
try:
logging.getLogger().setLevel(logging.DEBUG)
s = Sphero("/dev/rfcomm0")
s.connect()
s.set_motion_timeout(65525)
s.set_rgb(0,255,0)
s.set_back_led_output(255)
s.configure_locator(0,0)
except KeyboardInterrupt:
s.disconnect()
In the main function I call Connect() which is launching Threads over which i don't have direct controll.
When I launch this script I would like to be able to stop it when hitting Control+C by calling the "disconnect()" function which stops all the other threads.
In the code i provided it doesn't work because there is no thread in the main function. But I already tryied putting all the instuctions from Main() in a Thread with a While loop without success.
Is there a simple way to solve my problem ?
Thanx
Your indentation is messed up, but there's enough to go on.
Your main thread isn't catching SIGINT because it's not alive. There is nothing that stops your main thread from continuing past the try block, seeing no more code, and closing up shop.
I am not familiar with Sphero. I just attempted to google its docs and was linked to a bunch of 404 pages, so I'll tell you what you would normally do in a threaded environment - join your threads to the main thread so that the main thread can't finish execution before the worker threads.
for t in my_thread_list:
t.join() #main thread can't get past here until all the threads finish
If your Sphero object doesn't provide join-like functionality, you could hack something in that blocks, i.e.
raw_input('Press Enter to disconnect')
s.disconnect()
I have this server draft that forks new children after a new client connection. Then depends on a client’s command child server does some work inside the function handler(connection).
In the meantime, I want to stop the parent server and before that let the parent wait for all working children.
The question is where shall I place this signal function for the Ctrl+C keyboard interrupt option.
signal.signal(signal.SIGINT, signal_handler)
children_list = []
conn = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
conn.bind((HOST, PORT))
conn.listen(5)
print("Listening on TCP port %s" % PORT)
def reaper(pids):
while children_list:
pid,stat = os.waitpid(0, os.WNOHANG)
if not pid:
break
pids.remove(pid)
def handler(connection):
cmd = connection.recv(socksize)
def signal_handler(signal, frame):
print 'You pressed Ctrl+C!'
sys.exit(0)
def accept():
while 1:
global connection
connection, address = conn.accept()
print "welcome new client!"
reaper(children_list)
pid = os.fork()
if pid:#parent
children_list.append(pid)
connection.close()
else:#child
handler(connection)
accept()
The question is where shall I place this signal function for the Ctrl+C keyboard interrupt option. signal.signal(signal.SIGINT, signal_handler)
Anywhere you want, as long as it's after you define signal_handler and before you call accept.
I'm not sure that signal handler is actually what you want. What you're actually doing will, on most platforms, exit immediately, just as you're asking it to, causing the children to be reparented to init or launchd or similar. But you want to actually wait for all children. So, you can't call exit.
In POSIX, you're not allowed to call waitpid inside a signal hander. Python signal handlers aren't real signal handlers, so this might not be relevant—but I don't think that fact is guaranteed anywhere. If you want to take the risk, you could try waiting for all of the children right there in signal_handler before calling exit and see if it works on your platform.
However, the standard POSIX way to do it is to have the signal handler set some global flag and return. The main program will then wake up from accept or whatever other blocking call it's waiting on with an EINTR error, at which point it can check the flag—if true, time to wait on the children and quit; otherwise, go back into the loop and accept again.
But all of this shouldn't be necessary anyway. The default SIGINT handler should just raise a KeyboardInterrupt, which means that—if you don't install a custom signal handler—all you really need to do is catch that exception. In other words, just replace the last line with this:
try:
accept()
except KeyboardInterrupt: # or maybe BaseException
# wait for children, blocking instead of WNOHANG of course
sys.exit(0)
I am developing a multi-threaded application in python. I have following scenario.
There are 2-3 producer threads which communicate with DB and get some data in large chunks and fill them up in a queue
There is an intermediate worker which breaks large chunks fetched by producer threads into smaller ones and fill them up in another queue.
There are 5 consumer threads which consume queue created by intermediate worker thread.
objects of data sources are accessed by producer threads through their API. these data sources are completely separate. So these producer understands only presence or absence of data which is supposed to be given out by data source object.
I create threads of these three types and i make main thread wait for completion of these threads by calling join() on them.
Now for such a setup I want a common error handler which senses failure of any thread, any exception and decides what to do. For e.g if I press ctrl+c after I start my application, main thread dies but producer, consumer threads continue to run. I would like that once ctrl+c is pressed entire application should shut down. Similarly if some DB error occurs in data source module, then producer thread should get notified of that.
This is what I have done so far:
I have created a class ThreadManager, it's object is passed to all threads. I have written an error handler method and passed it to sys.excepthook. This handler should catch exceptions, error and then it should call methods of ThreadManager class to control the running threads. Here is snippet:
class Producer(threading.Thread):
....
def produce():
data = dataSource.getData()
class DataSource:
....
def getData():
raise Exception("critical")
def customHandler(exceptionType, value, stackTrace):
print "In custom handler"
sys.excepthook = customHandler
Now when a thread of producer class calls getData() of DataSource class, exception is thrown. But this exception is never caught by my customHandler method.
What am I missing? Also in such scenario what other strategy can I apply? Please help. Thank you for having enough patience to read all this :)
What you need is a decorator. In essence you are modifying your original function and putting in inside a try-except:
def exception_decorator(func):
def _function(*args):
try:
result = func(*args)
except:
print('*** ESC default handler ***')
os._exit(1)
return result
return _function
If your thread function is called myfunc, then you add the following line above your function definition
#exception_decorator
def myfunc():
pass;
Can't you just catch "KeyboardInterrupt" when pressing Ctrl+C and do:
for thread in threading.enumerate():
thread._Thread__stop()
thread._Thread__delete()
while len(threading.enumerate()) > 1:
time.sleep(1)
os._exit(0)
and have a flag in each threaded class which is self.alive
you could theoretically call thread.alive = False and have it stop gracefully?
for thread in threading.enumerate():
thread.alive = False
time.sleep(5) # Grace period
thread._Thread__stop()
thread._Thread__delete()
while len(threading.enumerate()) > 1:
time.sleep(1)
os._exit(0)
example:
import os
from threading import *
from time import sleep
class worker(Thread):
def __init__(self):
self.alive = True
Thread.__init__(self)
self.start()
def run(self):
while self.alive:
sleep(0.1)
runner = worker()
try:
raw_input('Press ctrl+c!')
except:
pass
for thread in enumerate():
thread.alive = False
sleep(1)
try:
thread._Thread__stop()
thread._Thread__delete()
except:
pass
# There will always be 1 thread alive and that's the __main__ thread.
while len(enumerate()) > 1:
sleep(1)
os._exit(0)
Try going about it by changing the internal system exception handler?
import sys
origExcepthook = sys.excepthook
def uberexcept(exctype, value, traceback):
if exctype == KeyboardInterrupt:
print "Gracefully shutting down all the threads"
# enumerate() thingie here.
else:
origExcepthook(exctype, value, traceback)
sys.excepthook = uberexcept
How can I call shutdown() in a SocketServer after receiving a certain message "exit"? As I know, the call to serve_forever() will block the server.
Thanks!
Use the source, Luke!
Excerpt from SocketServer.py:
def serve_forever(self, poll_interval=0.5):
"""Handle one request at a time until shutdown.
Polls for shutdown every poll_interval seconds. Ignores
self.timeout. If you need to do periodic tasks, do them in
another thread.
"""
self.__is_shut_down.clear()
try:
while not self.__shutdown_request:
# XXX: Consider using another file descriptor or
# connecting to the socket to wake this up instead of
# polling. Polling reduces our responsiveness to a
# shutdown request and wastes cpu at all other times.
r, w, e = select.select([self], [], [], poll_interval)
if self in r:
self._handle_request_noblock()
finally:
self.__shutdown_request = False
self.__is_shut_down.set()
def shutdown(self):
"""Stops the serve_forever loop.
Blocks until the loop has finished. This must be called while
serve_forever() is running in another thread, or it will
deadlock.
"""
self.__shutdown_request = True
self.__is_shut_down.wait()
No the serve_forever is checking a flag on a regular basis (by default 0.5 sec). Calling shutdown will raise this flag and cause the serve_forever to end.