I am in the process of adding client/server UDP support to thekogans stream library and have run into a problem on Windows. Here is what I am doing;
server udp socket is bound to 0.0.0.0:8854.
server udp socket has IP_PKTINFO = true.
server udp socket has SO_REUSEADDR = true.
server udp socket starts an overlapped WSARecvMsg operation.
client binds to 0.0.0.0:0 and connects to 127.0.0.1:8854.
client sends a message using WSASend.
server socket receives the message and creates a new UDP socket with the following attributes:
SO_REUSEADDR = true
bind to address returned by IP_PKTINFO (127.0.0.1:8854).
connect to whatever address was returned by WSARecvMsg.
client and the new server UDP socket exchange a bunch of messages (using WSASend and WSARecv).
Here is the behavior I am seeing:
the first connection between client and server works flawlessly.
I then have the client exit and restart.
all other packets from the client are dropped.
if I set a timeout on the new server UDP socket (127.0.0.1:8854) and it times out and is closed, then the client can connect again. In other words, the scheme seems to work but only one client at a time. If the server has a concrete (not wildcard) socket created for the same port, no other client can send it messages.
Some more information that may be helpful: The server is async and uses IOCP. This code (using epoll and kqueue) works perfectly on Linux and OS X. I feel like I am missing some flag somewhere that winsock needs set but I can't seem to find it. I have tried googling various search terms but have hit a wall.
Any and all help would be greatly appreciated. thank you.
Dear Stackoverflowers,
I am researching networking a bit and I decided I'd like to create a small and simple networking library with Winsock. (I am using Completion Ports and Overlapped IO though)
As I researched a bit I came to the following steps for a TCP Listener(Correct me if I am wrong):
Create a Listening Socket
Bind it to a port/IP
Listen on it
When a new connection is created, give a seperate Socket for that connection.
Listener continue's to listen, the specific connection is handled as needed.
EDIT: With a 'connection' from here I mean communication between the server and distinct clients.
Though for an UDP Listener we need to make use of WSARecvFrom which returns the IP address at the lpFrom parameter. Now I was wondering the following:
Is it better to make one UDP Socket listen to incoming connections on a specific port with WSARecvFrom and create new sockets for every specific connection? Or could I just use the UDP Socket itself with WSASendTo. Would that cause any performance penalties if one UDP Socket is used for for example 1000 connections? Or would it be the same or even better then creating/duplicating seperate Sockets for each different incoming connection?
Note: If multiple sockets are needed how would you handle sockets listening on the same port or could a client accept UDP from different ports?
Hope you guys can help!
Ps. Extra tips are always welcome!
Unlike TCP, UDP is connection-less, and as such you don't need to create separate sockets for each party. One UDP socket can handle everything. Bind it to a local IP/Port and call WSARecvFrom() once, and when it reports data to your IOCP you can process the data as needed (if another thread if needed) and then call WSARecvFrom() again. Each time new data arrives, you have to look at the reported lpFrom address to know the IP/Port of the sender. And yes, you can use the same UDP socket for sending data to each sender when needed.
When I do the following steps to receive a message and send a reply, it fails.
I am using TCP. I need the program to send data from the same port it received from.
bind()
listen()
accept()
recv()
connect()//it fails to connect here using the same socket.<br>
send()
It seems you have a problem in understanding the way tcp works. There is a server and a client. The server waits for connections, and the client makes connections. Once a connection is established, the server and the client can communicate bi-directional (i.e. both can send and recive messages). Of course, their role might change, but this is the way it works. So, the server does:
bind()
listen()
accept()
recv()
send()
It is stuck at accept() until a client performs connect() on the port that the server is listening to.
As my explanation is pretty brief, I suggest you read this tutorial about linux sockets.
My application connects as a client across an ethernet to a server process.
As the server is well known and will not change, UDP and TCP are both setup using
socket();
setsockopt(SO_REUSEADDR);
bind();
connect();
The connection protocol includes heartbeats sent both ways.
When I detect an error with the connection e.g. hearbeat timeout, I need to reset the connection.
Is it sufficient just to connect() to the NULL address and then re-connect() after a short pause, or should I close the socket and then reinitialise from scratch?
thanks
After a socket error you have to discard the one in hand and restart the setup with a new socket.
Winsock documentation, for example:
When a connection between sockets is
broken, the sockets should be
discarded and recreated. When a
problem develops on a connected
socket, the application must discard
and recreate the needed sockets in
order to return to a stable point.
You have to close(2) the socket and re-do everything again. Why do you bind(2) on the client?
I am writing simple client-server program.
Client send some messages to server using UDP or TCP. Server must be able to support both UDP and TCP.
If client, sends message using UDP, sequence of method calls in client is socket(),bind(),sendto(),recvfrom(),close() and that in server is socket(),bind(),sendto(),recvfrom(),close().
If it uses TCP,
sequence of call in server is
socket(),bind(),listen(),accept(),send(),recv(),close().
and that in client is
socket(),bind(),connect(),send(),recv(),close()
In my program, user/client is given choice in the start to select what he want to use UDP or TCP. So, my main problem is how can I know or differentiate in the server side, if the client is sending message using TCP or UDP. If it uses TCP, I must call listen(),accept(),send(),recv()
and if it uses UDP, I don't call listen(),accept() but call sendto() and recvfrom().
So, how can I differentiate/know this in the beginning so that I can make appropriate function calls.
Thanks.
Before the packet reaches you, you don't know whether it's UDP or TCP.
So you want to bind to both UDP and TCP sockets if you expect requests both ways.
Once you did, you just know which way it came by the socket you received the packet through.
When you create the socket, you pass a type - SOCK_STREAM (TCP) or SOCK_DGRAM (UDP)
So the two kinds of traffic will be on two different sockets.
Just as Henry Troup pointed out, an IP socket is defined as
(transport, interface, port).
(UDP, 127.0.0.1, 80) is not the same IP socket as (TCP, 127.0.0.1, 80) , thus you can safely bind to both of them and listen for incoming traffic.
just let the TCP socket listen on port X, and do the UDP connections through port Y