Hi all,
I am currently working on a HTTP Proxy. For maximum flexibility, I am
implementing the proxy at a low level : I am using the SocketServer library.
The server itself is very simple:
class MyTCPServer(SocketServer.TCPServer):
allow_reuse_address = 1
and the handler looks like:
class MyTCPHandler(SocketServer.BaseRequestHandler):
def handle(self):
# Prints ip and port
print "\n#### " + str(self.client_address) + " ####"
requestParser = HTTPRequestParser()
while True:
# Get packet
data = self.request.recv(4096)
if data == '':
break
# Parse request packet
if requestParser.got_new_chunk(data):
break
someStuff = ""
self.request.send(someStuff)
This is working fine, but I have a small performance issue.
The proxy is targeted by Firefox. Sometimes the first receive call on
the input socket takes a lot of time (up to 20s sometimes), which should
not be as this is a communication of two local sockets. This is
occurring maybe every 30 requests or so.
When looking at the logs I noticed that this weird behavior happens when
the client port is not contiguous to the previous ones. For example,
handling the fourth request takes a lot of time:
#### ('127.0.0.1', 49704) ####
#### ('127.0.0.1', 49705) ####
#### ('127.0.0.1', 49706) ####
#### ('127.0.0.1', 49674) ####
Do you have any idea what the problem could be ? I tried to manually
close self.request request, but I still have the problem. Is it related
to "allow_reuse_address = 1" ?
Thanks for your help !
Rémi
--
http://mail.python.org/mailman/listinfo/python-list
