tornado - WebSocketClientConnection - How to catch and handle connection failures? - python

I am a Tornado and also Websocket newbie. There are many resources how to implement a websocket server application with Tornado. However, I haven't found a complete example that contains a websocket client application built on top of Tornado.
Server Application (server.py)
#!/usr/bin/env python
# -*- coding: utf-8 -*-
import logging
import tornado.web
import tornado.websocket
import tornado.ioloop
import tornado.options
from tornado.options import define, options
define("port", default=3000, help="run on the given port", type=int)
class Application(tornado.web.Application):
def __init__(self):
handlers = [(r"/", MainHandler)]
settings = dict(debug=True)
tornado.web.Application.__init__(self, handlers, **settings)
class MainHandler(tornado.websocket.WebSocketHandler):
def check_origin(self, origin):
return True
def open(self):
logging.info("A client connected.")
def on_close(self):
logging.info("A device disconnected")
def main():
tornado.options.parse_command_line()
app = Application()
app.listen(options.port)
tornado.ioloop.IOLoop.instance().start()
if __name__ == "__main__":
main()
Client Application (client.py):
#!/usr/bin/env python
# -*- coding: utf-8 -*-
from tornado.ioloop import IOLoop
from tornado.websocket import websocket_connect
if __name__ == "__main__":
ws = websocket_connect("ws://localhost:3000")
IOLoop.instance().start()
Problems & Questions
I don't have any problems with server application. However, I can't say same thing for client application. Why?
If server goes down, client application doesn't react. It does not say "Hey, server is down buddy.". I write my own WebSocketClientConnection class (inherited by the original) and also websocket_connect function (same function, but this one uses my WebSocketClientConnection class). I override on_connection_close method in my WebSocketClientConnection class, so I can catch "server down on middle of the connection" failure. However, I cannot make it reconnect. How do I make it reconnect to server?
If server is already down before the connection, client application does not raise anything and seems perfectly. I don't know how to catch that failure. How do I catch this connection failure?
My WebSocketClientConnection class:
class MyWebSocketClientConnection(tornado.websocket.WebSocketClientConnection):
def on_connection_close(self):
super(MyWebSocketClientConnection, self).on_connection_close()
print "connection closed"
Thanks.

By the way, after I figured out, I did write an example Tornado WebSocket client/server pair to demonstrate how to do it.
https://github.com/ilkerkesen/tornado-websocket-client-example
I hope it helps someone.

websocket_connect returns a Future; to see whether it succeeded or failed you must examine the Future (probably by yielding it in a coroutine). Furthermore, after establishing the connection you should go into a read_message loop. Even if you don't expect the client to send any messages, you should still call read_message: this is how you will connections that are closed after being established (read_message will return None).
#gen.coroutine
def my_websocket_client(url):
ws = yield websocket_connect(url)
while True:
msg = yield ws.read_message()
if msg is None: break
# do stuff with msg

Related

Implement SSL in tornadoweb websockets server

Good Day to all,
I have a websocket server written in python that serves clients changes done to a postgresql database table. Presently it uses the standard ws protocol. I would Like to implement SSL wss but there seems to be no documentation on the tornadoweb site. Any help appreciated!
Code as follows:
import tornado.web
import tornado.websocket
import tornado.ioloop
import tornado.httpserver
import threading
import select
import psycopg2
import psycopg2.extensions
# This is a global variable to store all connected clients
websockets = []
# Connect to DB to listen for notifications
conn = psycopg2.connect("host=xxx.xxx.xxx.xxx dbname=mydb user=xxx port=yyy")
conn.set_isolation_level(psycopg2.extensions.ISOLATION_LEVEL_AUTOCOMMIT)
curs = conn.cursor()
curs.execute("LISTEN notifier;")
# Create the websocket
class WebSocketHandler(tornado.websocket.WebSocketHandler):
#tornado.web.asynchronous
# Add to global variable the connected client
def open(self):
self.set_nodelay(True)
# on disconnect remove client
def on_close(self):
# Search for it in the list. If it matches with us (the client) then remove it as he has quit the con
try:
for websocket in websockets:
if websocket[0] == self:
websockets.remove(websocket)
except ValueError:
print ValueError
def on_message(self, message):
if self not in websockets:
websockets.append([self,message])
def on_pong(self, data):
print data
# This is the function that polls the database for changes, then it sends change to clients through websockets
def db_listener():
while 1:
if select.select([conn],[],[],5) == ([],[],[]):
for websocket in websockets:
# ping the client regularly to avoid disconnect
websocket[0].ping("ping")
else:
conn.poll()
while conn.notifies:
notify = conn.notifies.pop()
details = notify.payload.split(",")
if len(details) > 1:
for websocket in websockets:
if details[34] in websocket[1]:
websocket[0].write_message(notify.payload)
application = tornado.web.Application([
(r"/websocket", WebSocketHandler),
])
if __name__ == "__main__":
# Start a separate thread for every client so that we do not block the main websockets program!
threading.Thread(target=db_listener).start()
application.listen(5252)
tornado.ioloop.IOLoop.instance().start()
Any suggestions?
Thanks to all!!
try this:
server = tornado.httpserver.HTTPServer(application, ssl_options = {
"certfile": "path-to-crt-file",
"keyfile": "path-to-key-file",
})
server.listen(5252)

Attaching ZMQStream with existing tornado ioloop

I have an application where every websocket connection (within tornado open callback) creates a zmq.SUB socket to an existing zmq.FORWARDER device. Idea is to receive data from zmq as callbacks, which can then be relayed to frontend clients over websocket connection.
https://gist.github.com/abhinavsingh/6378134
ws.py
import zmq
from zmq.eventloop import ioloop
from zmq.eventloop.zmqstream import ZMQStream
ioloop.install()
from tornado.websocket import WebSocketHandler
from tornado.web import Application
from tornado.ioloop import IOLoop
ioloop = IOLoop.instance()
class ZMQPubSub(object):
def __init__(self, callback):
self.callback = callback
def connect(self):
self.context = zmq.Context()
self.socket = self.context.socket(zmq.SUB)
self.socket.connect('tcp://127.0.0.1:5560')
self.stream = ZMQStream(self.socket)
self.stream.on_recv(self.callback)
def subscribe(self, channel_id):
self.socket.setsockopt(zmq.SUBSCRIBE, channel_id)
class MyWebSocket(WebSocketHandler):
def open(self):
self.pubsub = ZMQPubSub(self.on_data)
self.pubsub.connect()
self.pubsub.subscribe("session_id")
print 'ws opened'
def on_message(self, message):
print message
def on_close(self):
print 'ws closed'
def on_data(self, data):
print data
def main():
application = Application([(r'/channel', MyWebSocket)])
application.listen(10001)
print 'starting ws on port 10001'
ioloop.start()
if __name__ == '__main__':
main()
forwarder.py
import zmq
def main():
try:
context = zmq.Context(1)
frontend = context.socket(zmq.SUB)
frontend.bind('tcp://*:5559')
frontend.setsockopt(zmq.SUBSCRIBE, '')
backend = context.socket(zmq.PUB)
backend.bind('tcp://*:5560')
print 'starting zmq forwarder'
zmq.device(zmq.FORWARDER, frontend, backend)
except KeyboardInterrupt:
pass
except Exception as e:
logger.exception(e)
finally:
frontend.close()
backend.close()
context.term()
if __name__ == '__main__':
main()
publish.py
import zmq
if __name__ == '__main__':
context = zmq.Context()
socket = context.socket(zmq.PUB)
socket.connect('tcp://127.0.0.1:5559')
socket.send('session_id helloworld')
print 'sent data for channel session_id'
However, my ZMQPubSub class doesn't seem like is receiving any data at all.
I further experimented and realized that I need to call ioloop.IOLoop.instance().start() after registering on_recv callback within ZMQPubSub. But, that will just block the execution.
I also tried passing main.ioloop instance to ZMQStream constructor but doesn't help either.
Is there a way by which I can bind ZMQStream to existing main.ioloop instance without blocking flow within MyWebSocket.open?
In your now complete example, simply change frontend in your forwarder to a PULL socket and your publisher socket to PUSH, and it should behave as you expect.
The general principles of socket choice that are relevant here:
use PUB/SUB when you want to send a message to everyone who is ready to receive it (may be no one)
use PUSH/PULL when you want to send a message to exactly one peer, waiting for them to be ready
it may appear initially that you just want PUB-SUB, but once you start looking at each socket pair, you realize that they are very different. The frontend-websocket connection is definitely PUB-SUB - you may have zero-to-many receivers, and you just want to send messages to everyone who happens to be available when a message comes through. But the backend side is different - there is only one receiver, and it definitely wants every message from the publishers.
So there you have it - backend should be PULL and frontend PUB. All your sockets:
PUSH -> [PULL-PUB] -> SUB
publisher.py: socket is PUSH, connected to backend in device.py
forwarder.py: backend is PULL, frontend is PUB
ws.py: SUB connects and subscribes to forwarder.frontend.
The relevant behavior that makes PUB/SUB fail on the backend in your case is the slow joiner syndrome, which is described in The Guide. Essentially, subscribers take a finite time to tell publishers about there subscriptions, so if you send a message immediately after opening a PUB socket, the odds are it hasn't been told that it has any subscribers yet, so it's just discarding messages.
ZeroMq subscribers have to subscribe on what messages they wish to receive; I don't see that in your code. I believe the Python way is this:
self.socket.setsockopt(zmq.SUBSCRIBE, "")

Tornado server stuck in loop, not accepting clients

I've written a code in tornado that connects to a server that is pushing an infinite data stream, processes the data stream and sends it out on a websocket server.
The problem is that the way I implemented it the server became blocked on a particular function and doesn't accept any more clients since it never exits the function serving the data to the websocket. I want the connection to the server and the data retrieved from it processed only once but send the processed data to all the clients that connect to my tornado server. Could someone please help me, I can't figure out a way to do it. Here's my code with processing of data removed:
import socket
import ssl
import tornado.httpserver
import tornado.websocket
import tornado.ioloop
import tornado.web
websockets = []
class WSHandler(tornado.websocket.WebSocketHandler):
def readData(self):
while True:
line = self.ssl_sock.read()
#PROCESS THE READ LINE AND CONVERT INTO RESULTING DATA
if(toSend):
self.write_message(result)
def makeConnection(self):
self.ssl_sock.connect(self.address)
self.readData()
def open(self):
print 'New connection was opened'
self.s=socket.socket(socket.AF_INET, socket.SOCK_STREAM)
self.ssl_sock=ssl.wrap_socket(self.s, cert_reqs=ssl.CERT_NONE)
self.address=('SERVER_ADDRESS',5000)
self.nodes=[]
self.edges=[]
if self not in websockets:
print ('added')
websockets.append(self)
if(len(websockets)==1):
print('executing make conn')
self.makeConnection()
else:
self.readData()
print('executing read data')
def on_message(self, message):
print 'Incoming message:', message
self.write_message("You said: " + message)
def on_close(self):
print 'Connection was closed...'
application = tornado.web.Application([
(r'/ws', WSHandler),
])
if __name__ == "__main__":
http_server = tornado.httpserver.HTTPServer(application)
http_server.listen(8888)
tornado.ioloop.IOLoop.instance().start()
Tornado is an asynchronous framework, that is, all your IO must be run within its event loop, otherwise the whole server gets stuck.
Try having a look at Tornado Async Client.

Sockjs-tornado and rpc

I'm trying to add a sockjs-tornado server to my site, and all worked fine until I decided to connect it to my other apps via MsgPack (using msgpack-rpc-python). And now works sockjs server either RPC server. Depending on wich of them start there loop first.
I think that I need to use one tornado.ioloop for both of them. But do not know how to achieve it. Or may be there is another way to add rpc to a tornado server?
Here is a sockjs-tornado sample code with msgpack-rpc-python:
import tornado.ioloop
import tornado.web
import sockjs.tornado
import msgpackrpc
class RPCServer(object):
def sum(self, x, y):
return x + y
class IndexHandler(tornado.web.RequestHandler):
"""Regular HTTP handler to serve the chatroom page"""
def get(self):
self.render('index.html')
class ChatConnection(sockjs.tornado.SockJSConnection):
"""Chat connection implementation"""
# Class level variable
participants = set()
def on_open(self, info):
# Send that someone joined
self.broadcast(self.participants, "Someone joined.")
# Add client to the clients list
self.participants.add(self)
def on_message(self, message):
# Broadcast message
self.broadcast(self.participants, message)
def on_close(self):
# Remove client from the clients list and broadcast leave message
self.participants.remove(self)
self.broadcast(self.participants, "Someone left.")
if __name__ == "__main__":
# 1. Create chat router
ChatRouter = sockjs.tornado.SockJSRouter(ChatConnection, '/chat')
# 1.5 Create MsgPack RPC Server
rpc = msgpackrpc.Server(RPCServer())
# 2. Create Tornado application
app = tornado.web.Application(
[(r"/", IndexHandler)] + ChatRouter.urls
)
# 3. Make Tornado app listen on port 5000
app.listen(5000)
# 3.5 Make MsgPack RPC Server listen on port 5001
rpc.listen(msgpackrpc.Address('localhost', 5001))
# 4. Start IOLoop
tornado.ioloop.IOLoop.instance().start()
# 5. Never executed
rpc.start()
`
Any suggestions or examples are welcome!
This happens because both start() calls start Tornado IOLoop and they won't exit until IOLoop stopped.
Yes, you have to use one IOLoop. As msgpackrpc.Server accepts Loop class instance and Loop incapsulates IOLoop, try this:
if __name__ == '__main__':
io_loop = tornado.ioloop.IOLoop.instance()
loop = msgpackrpc.Loop(io_loop)
rpc = msgpackrpc.Server(RPCServer(), loop=loop)
# ... sockjs-tornado initialisation. No need to call rpc.start()
io_loop.start()

Bottle web framework - How to stop?

When starting a bottle webserver without a thread or a subprocess, there's no problem. To exit the bottle app -> CTRL + c.
In a thread, how can I programmatically stop the bottle web server ?
I didn't find a stop() method or something like that in the documentation. Is there a reason ?
For the default (WSGIRef) server, this is what I do (actually it is a cleaner approach of Vikram Pudi's suggestion):
from bottle import Bottle, ServerAdapter
class MyWSGIRefServer(ServerAdapter):
server = None
def run(self, handler):
from wsgiref.simple_server import make_server, WSGIRequestHandler
if self.quiet:
class QuietHandler(WSGIRequestHandler):
def log_request(*args, **kw): pass
self.options['handler_class'] = QuietHandler
self.server = make_server(self.host, self.port, handler, **self.options)
self.server.serve_forever()
def stop(self):
# self.server.server_close() <--- alternative but causes bad fd exception
self.server.shutdown()
app = Bottle()
#app.route('/')
def index():
return 'Hello world'
#app.route('/stop') # not working from here, it has to come from another thread
def stopit():
server.stop()
server = MyWSGIRefServer(port=80)
try:
app.run(server=server)
except:
print('Bye')
When I want to stop the bottle application, from another thread, I do the following:
server.stop()
I had trouble closing a bottle server from within a request as bottle seems to run requests in subprocesses.
I eventually found the solution was to do:
sys.stderr.close()
inside the request (that got passed up to the bottle server and axed it).
An updated version of mike's answer.
from bottlepy.bottle import WSGIRefServer, run
from threading import Thread
import time
class MyServer(WSGIRefServer):
def run(self, app): # pragma: no cover
from wsgiref.simple_server import WSGIRequestHandler, WSGIServer
from wsgiref.simple_server import make_server
import socket
class FixedHandler(WSGIRequestHandler):
def address_string(self): # Prevent reverse DNS lookups please.
return self.client_address[0]
def log_request(*args, **kw):
if not self.quiet:
return WSGIRequestHandler.log_request(*args, **kw)
handler_cls = self.options.get('handler_class', FixedHandler)
server_cls = self.options.get('server_class', WSGIServer)
if ':' in self.host: # Fix wsgiref for IPv6 addresses.
if getattr(server_cls, 'address_family') == socket.AF_INET:
class server_cls(server_cls):
address_family = socket.AF_INET6
srv = make_server(self.host, self.port, app, server_cls, handler_cls)
self.srv = srv ### THIS IS THE ONLY CHANGE TO THE ORIGINAL CLASS METHOD!
srv.serve_forever()
def shutdown(self): ### ADD SHUTDOWN METHOD.
self.srv.shutdown()
# self.server.server_close()
def begin():
run(server=server)
server = MyServer(host="localhost", port=8088)
Thread(target=begin).start()
time.sleep(2) # Shut down server after 2 seconds
server.shutdown()
The class WSGIRefServer is entirely copied with only 1 line added to the run() method is added. Also add a simple shutdown() method. Unfortunately, this is necessary because of the way bottle creates the run() method.
You can make your thread a daemon by setting the daemon property to True before calling start.
mythread = threading.Thread()
mythread.daemon = True
mythread.start()
A deamon thread will stop whenever the main thread that it is running in is killed or dies. The only problem is that you won't be able to make the thread run any code on exit and if the thread is in the process of doing something, it will be stopped immediately without being able to finish the method it is running.
There's no way in Python to actually explicitly stop a thread. If you want to have more control over being able to stop your server you should look into Python Processes from the multiprocesses module.
Since bottle doesn't provide a mechanism, it requires a hack. This is perhaps the cleanest one if you are using the default WSGI server:
In bottle's code the WSGI server is started with:
srv.serve_forever()
If you have started bottle in its own thread, you can stop it using:
srv.shutdown()
To access the srv variable in your code, you need to edit the bottle source code and make it global. After changing the bottle code, it would look like:
srv = None #make srv global
class WSGIRefServer(ServerAdapter):
def run(self, handler): # pragma: no cover
global srv #make srv global
...
Here's one option: provide custom server (same as default), that records itself:
import bottle
class WSGI(bottle.WSGIRefServer):
instances = []
def run(self, *args, **kw):
self.instances.append(self)
super(WSGI, self).run(*args, **kw)
# some other thread:
bottle.run(host=ip_address, port=12345, server=WSGI)
# control thread:
logging.warn("servers are %s", WSGI.instances)
This is exactly the same method than sepero's and mike's answer, but now much simpler with Bottle version 0.13+:
from bottle import W, run, route
from threading import Thread
import time
#route('/')
def index():
return 'Hello world'
def shutdown():
time.sleep(5)
server.srv.shutdown()
server = WSGIRefServer(port=80)
Thread(target=shutdown).start()
run(server=server)
Also related: https://github.com/bottlepy/bottle/issues/1229 and https://github.com/bottlepy/bottle/issues/1230.
Another example with a route http://localhost/stop to do the shutdown:
from bottle import WSGIRefServer, run, route
from threading import Thread
#route('/')
def index():
return 'Hello world'
#route('/stop')
def stopit():
Thread(target=shutdown).start()
def shutdown():
server.srv.shutdown()
server = WSGIRefServer(port=80)
run(server=server)
PS: it requires at least Bottle 0.13dev.
Console log of Bottle server tells us that the official way of shutting down the server is "Hit Ctrl-C":
Bottle v0.12.19 server starting up (using WSGIRefServer())...
Listening on http://localhost:8080/
Hit Ctrl-C to quit.
Why not simply follow it programmatically?
Hitting "Ctrl-C" is nothing but sending SIGINT to the process, and we can achieve it with just built-in modules:
Get current PID with os.getpid().
Kill the process with os.kill(). Remember passing SIGINT so it will be exactly same as "Hit Ctrl-C".
Wrap the 'kill' in another thread and start it after a few seconds so the client won't get error.
Here is the server code:
from bottle import route, run
import os
import signal
from threading import Thread
import time
#route('/hello')
def return_hello():
return 'Hello'
#route('/stop')
def handle_stop_request():
# Handle "stop server" request from client: start a new thread to stop the server
Thread(target=shutdown_server).start()
return ''
def shutdown_server():
time.sleep(2)
pid = os.getpid() # Get process ID of the current Python script
os.kill(pid, signal.SIGINT)
# Kill the current script process with SIGINT, which does same as "Ctrl-C"
run(host='localhost', port=8080)
Here is the client code:
import requests
def request_service(service_key):
url = f'http://127.0.0.1:8080/{service_key}'
response = requests.get(url)
content = response.content.decode('utf-8')
print(content)
request_service('hello')
request_service('stop')
Note that in function "handle_stop_request" we didn't stop the server immediately but rather started a thread then returned empty string. With this mechanism, when a client requests "http://127.0.0.1:8080/stop", it can get response (the empty string) normally. After that, the server will shutdown. If we otherwise shutdown the server in function "handle_stop_request", the server will close the connection before returning to the client, and hence the client will get "ConnectionError".
Server side output:
Bottle v0.12.19 server starting up (using WSGIRefServer())...
Listening on http://localhost:8080/
Hit Ctrl-C to quit.
127.0.0.1 - - [23/Nov/2021 11:18:08] "GET /hello HTTP/1.1" 200 5
127.0.0.1 - - [23/Nov/2021 11:18:08] "GET /stop HTTP/1.1" 200 0
Client side output:
Hello
The code was tested under Python 3.7 and Bottle 0.12.
I suppose that the bottle webserver runs forever until it terminates. There are no methonds like stop().
But you can make something like this:
from bottle import route, run
import threading, time, os, signal, sys, operator
class MyThread(threading.Thread):
def __init__(self, target, *args):
threading.Thread.__init__(self, target=target, args=args)
self.start()
class Watcher:
def __init__(self):
self.child = os.fork()
if self.child == 0:
return
else:
self.watch()
def watch(self):
try:
os.wait()
except KeyboardInterrupt:
print 'KeyBoardInterrupt'
self.kill()
sys.exit()
def kill(self):
try:
os.kill(self.child, signal.SIGKILL)
except OSError: pass
def background_process():
while 1:
print('background thread running')
time.sleep(1)
#route('/hello/:name')
def index(name='World'):
return '<b>Hello %s!</b>' % name
def main():
Watcher()
MyThread(background_process)
run(host='localhost', port=8080)
if __name__ == "__main__":
main()
Then you can use Watcher.kill() when you need to kill your server.
Here is the code of run() function of the bottle:
try:
app = app or default_app()
if isinstance(app, basestring):
app = load_app(app)
if not callable(app):
raise ValueError("Application is not callable: %r" % app)
for plugin in plugins or []:
app.install(plugin)
if server in server_names:
server = server_names.get(server)
if isinstance(server, basestring):
server = load(server)
if isinstance(server, type):
server = server(host=host, port=port, **kargs)
if not isinstance(server, ServerAdapter):
raise ValueError("Unknown or unsupported server: %r" % server)
server.quiet = server.quiet or quiet
if not server.quiet:
stderr("Bottle server starting up (using %s)...\n" % repr(server))
stderr("Listening on http://%s:%d/\n" % (server.host, server.port))
stderr("Hit Ctrl-C to quit.\n\n")
if reloader:
lockfile = os.environ.get('BOTTLE_LOCKFILE')
bgcheck = FileCheckerThread(lockfile, interval)
with bgcheck:
server.run(app)
if bgcheck.status == 'reload':
sys.exit(3)
else:
server.run(app)
except KeyboardInterrupt:
pass
except (SyntaxError, ImportError):
if not reloader: raise
if not getattr(server, 'quiet', False): print_exc()
sys.exit(3)
finally:
if not getattr(server, 'quiet', False): stderr('Shutdown...\n')
As you can see there are no other way to get off the run loop, except some exceptions.
The server.run function depends on the server you use, but there are no universal quit-method anyway.
This equally kludgy hack has the advantage that is doesn't have you copy-paste any code from bottle.py:
# The global server instance.
server = None
def setup_monkey_patch_for_server_shutdown():
"""Setup globals to steal access to the server reference.
This is required to initiate shutdown, unfortunately.
(Bottle could easily remedy that.)"""
# Save the original function.
from wsgiref.simple_server import make_server
# Create a decorator that will save the server upon start.
def stealing_make_server(*args, **kw):
global server
server = make_server(*args, **kw)
return server
# Patch up wsgiref itself with the decorated function.
import wsgiref.simple_server
wsgiref.simple_server.make_server = stealing_make_server
setup_monkey_patch_for_server_shutdown()
def shutdown():
"""Request for the server to shutdown."""
server.shutdown()
I've found this solution to be the easiest, but it does require that the "psutil" package is installed, to get the current process. It also requires the "signals" module, but that's part of the standard library.
#route('/shutdown')
def shutdown():
current_process = psutil.Process()
current_process.send_signal(signal.CTRL_C_EVENT)
return 'Shutting down the web server'
Hope that's of use to someone!
This question was top in my google search, so i will post my answer:
When the server is started with the Bottle() class, it has a method close() to stop the server. From the source code:
""" Close the application and all installed plugins. """
For example:
class Server:
def __init__(self, host, port):
self._host = host
self._port = port
self._app = Bottle()
def stop(self):
# close ws server
self._app.close()
def foo(self):
# More methods, routes...
Calling stop method will stop de server.

Categories