I took a really simple example of a rust server with tcp.
use std::net::{Shutdown,TcpListener, TcpStream};
use std::thread;
use std::io::{Read,Write,Error};
fn handle_client(mut stream: TcpStream)-> Result<(), Error> {
println!("incoming connection from: {}", stream.peer_addr()?);
let mut buf = [0;512];
loop {
let bytes_read = stream.read(&mut buf)?;
if bytes_read == 0 {return Ok(())}
let tmp = format!("{}", String::from_utf8_lossy(&buf).trim());
eprintln!("getting {}",tmp);
stream.write(&buf[..bytes_read])?;
}
}
fn main() {
let listener = TcpListener::bind("0.0.0.0:8888").expect("Could not bind");
for stream in listener.incoming() {
match stream {
Err(e)=> {eprintln!("failed: {}", e)}
Ok(stream) => {
thread::spawn(move || {
handle_client(stream).unwrap_or_else(|error| eprintln!("{:?}", error));
});
}
}
}
}
Which basically takes input, spits it back at the client, and prints to it's own terminal.
I would like to be able to end this connection. Ending the connection should probably happen depending on something, but right now I want to just try to shut it down.
I tried looking thorugh the docs, and then tried adding the shutdown method
Now, I want to take some stream as input, do something with it, and then shut the channel
So I tried doing this:
fn main() {
let listener = TcpListener::bind("0.0.0.0:8888").expect("Could not bind");
for stream in listener.incoming() {
match stream {
Err(e)=> {eprintln!("failed: {}", e)}
Ok(stream) => {
thread::spawn(move || {
handle_client(stream).unwrap_or_else(|error| eprintln!("{:?}", error));
});
stream.shutdown(Shutdown::Both).expect("shutdown call failed");
}
}
}
}
But this causes an issue with the "stream" being a moved value.
So how can I shut down the channel right after receiving and doing something with the input?
(I still want to preserve this structure with the loop, since I actually want to receive many messages, and then shut down dependning on the input)
You can try to do a clone of the stream and pass that stream clone to the function you spawn. Then you can call the shutdown inside your spawned function after you have handled the client.
This way, your original stream variable would remain intact and you can have the behavior you want.
The reason you get this issue is that the spawned function moves everything it captures due to the move keyword.
fn main() {
let listener = TcpListener::bind("0.0.0.0:8888").expect("Could not bind");
for stream in listener.incoming() {
let stream_clone = stream.clone();
match stream {
Err(e)=> {eprintln!("failed: {}", e)}
Ok(stream) => {
thread::spawn(move || {
handle_client(stream_clone).unwrap_or_else(|error| eprintln!("{:?}", error));
stream_clone.shutdown(Shutdown::Both).expect("shutdown call failed");
});
}
}
}
}
Also, you would need to change your signature of handle client to fn handle_client(stream: &mut TcpStream)-> Result<(), Error> so it would not also move the cloned stream variable, but burrow as mutable instead.
Related
I want to set up socket to listen for incoming connections, and do some logic based on incoming message.
use std::io::{Read, Write};
use std::net::{TcpListener, TcpStream};
fn main() {
set_listening_server("192.168.80.180", 2048);
}
fn set_listening_server(ip: &str, port: i32) {
assert!(port > 1000 && ip.len() > 0);
println!("Function is OK!");
let addr = format!("{}:{}", ip, port);
let listener = match TcpListener::bind(&addr) {
Ok(listener) => {
println!("Listening on {}", &addr);
loop {}
}
Err(e) => {
println!("Error binding to {}: {}", &addr, e);
return;
}
};
}
The above is how I have set up the socket. At first I had the loop{} exist right after the set_listening_server call but I figured that as soon as the function finished executing that socket was dropped. Moving the loop inside the function call solves those issues, but is there perhaps a way to declare listener globally?
I want to have seperate functions to handle steps of the communication, for example if I want to have something like:
fn service_connection() {
for stream in listener.incoming() {
let stream = stream.unwrap();
println!("Connection established!");
}
}
How can I access listener if it is being declared inside of the set_listening_server?
Would calling this service_connection function inside of the loop be the correct approach?
EDIT:
If I want to return the socket how can I do it correctly?
fn set_listening_server(ip: &str, port: i32) -> TcpListener{
assert!(port > 1000 && ip.len() > 0);
println!("Function is OK!");
let addr = format!("{}:{}", ip, port);
let listener = TcpListener::bind(&addr);
return listener;
}
In the following code, I understand why I'm not allowed to do this(I think), but I'm not sure what to do to fix the issue. I'm simply trying to perform an action based upon an incoming message on a UDPSocket. However, by sending the reference to the slice over the channel, I get a problem where the buffer doesn't live long enough. I'm hoping for some suggestions because I don't know enough about Rust to move forward.
fn main() -> std::io::Result<()> {
let (tx, rx) = mpsc::channel();
thread::spawn(move || loop {
match rx.try_recv() {
Ok(msg) => {
match msg {
"begin" => // run an operation
"end" | _ => // kill the previous operation
}
}
Err = { //Error Handling }
}
}
// start listener
let socket: UdpSocket = UdpSocket::bind("0.0.0.0:9001")?;
loop {
let mut buffer = [0; 100];
let (length, src_address) = socket.recv_from(&mut buffer)?;
println!("Received message of {} bytes from {}", length, src_address);
let cmd= str::from_utf8(&buffer[0..length]).unwrap(); // <- buffer does not live long enough
println!("Command: {}", cmd);
tx.send(cmd).expect("unable to send message to channel"); // Error goes away if I remove this.
}
}
Generally you should avoid sending non-owned values over a channel since its unlikely that a lifetime would be valid for both the sender and receiver (its possible to do, but you'd have to plan for it).
In this situation, you're trying to share pass &str across the channel but since it just references buffer which isn't guaranteed to exist whenever rx receives it, you get a borrow checking error. You would probably want to convert the &str into an owned String and pass that over the channel:
use std::net::UdpSocket;
use std::sync::mpsc;
fn main() {
let (tx, rx) = mpsc::channel();
std::thread::spawn(move || loop {
match rx.recv().as_deref() {
Ok("begin") => { /* run an operation */ }
Ok("end") => { /* kill the previous operation */ }
Ok(_) => { /* unknown */ }
Err(_) => { break; }
}
});
let socket = UdpSocket::bind("0.0.0.0:9001").unwrap();
loop {
let mut buffer = [0; 100];
let (length, src_address) = socket.recv_from(&mut buffer).unwrap();
let cmd = std::str::from_utf8(&buffer[0..length]).unwrap();
tx.send(cmd.to_owned()).unwrap();
}
}
As proposed in the comments, you can avoid allocating a string if you parse the value into a known value for an enum and send that across the channel instead:
use std::net::UdpSocket;
use std::sync::mpsc;
enum Command {
Begin,
End,
}
fn main() {
let (tx, rx) = mpsc::channel();
std::thread::spawn(move || loop {
match rx.recv() {
Ok(Command::Begin) => { /* run an operation */ }
Ok(Command::End) => { /* kill the previous operation */ }
Err(_) => { break; }
}
});
let socket = UdpSocket::bind("0.0.0.0:9001").unwrap();
loop {
let mut buffer = [0; 100];
let (length, src_address) = socket.recv_from(&mut buffer).unwrap();
let cmd = std::str::from_utf8(&buffer[0..length]).unwrap();
let cmd = match cmd {
"begin" => Command::Begin,
"end" => Command::End,
_ => panic!("unknown command")
};
tx.send(cmd).unwrap();
}
}
I ahve a really simple tcp client/server setup that I found somewhere online.
use std::net::{Shutdown,TcpListener, TcpStream};
use std::thread;
use std::io::{Read,Write,Error};
fn handle_client(mut stream: TcpStream)-> Result<(), Error> {
println!("incoming connection from: {}", stream.peer_addr()?);
let mut buf = [0;512];
loop {
let bytes_read = stream.read(&mut buf)?;
if bytes_read == 0 {return Ok(())}
let tmp = format!("{}", String::from_utf8_lossy(&buf).trim());
eprintln!("getting {}",tmp);
stream.write(&vec![1,2,3,4])?;
}
}
fn main() {
let listener = TcpListener::bind("0.0.0.0:8888").expect("Could not bind");
let mut i = 0;
for stream in listener.incoming() {
match stream {
Err(e)=> {eprintln!("failed: {}", e)}
Ok(stream) => {
thread::spawn(move || {
handle_client(stream).unwrap_or_else(|error| eprintln!("{:?}", error));
});
}
}
}
}
and the client looks like so:
use std::net::TcpStream;
use std::str;
use std::io::{self,BufRead,BufReader,Write};
fn main() {
let mut stream = TcpStream::connect("0.0.0.0:8888").expect("could not connect");
loop {
let mut input = String::new();
let mut buffer : Vec<u8> = Vec::new();
io::stdin().read_line(&mut input).expect("failed to read stdin");
stream.write(input.as_bytes()).expect("Failed to write to server");
let mut reader = BufReader::new(&stream);
reader.read_until(b'\n', &mut buffer).expect("Could not read into buffer");
println!("{}", str::from_utf8(&buffer).expect("msg: &str"))
}
println!("Hello, world!");
}
(actually vs code tells me that the code inside the loop here is unreachable, which is false)
But this basically just lets the client send a message to the server, that the server sends back.
I would actually like to send vectors back and forth, which actually seems like a simpler task.
So, I change the line in the server that writes to the stream to this:
stream.write(&vec![1,2,3,4])?;
and now, in the client that prints the message it gets from the serve to:
println!("{:?}", &buffer)
But when I do this, nothing happens on the clientside, nothing is printed, and it seems like the loop is just stuck somewhere.
I gugess it has something to do with this line:
reader.read_until(b'\n', &mut buffer).expect("Could not read into buffer");
and I read some other way of reading in. I tried using a function from tcpstream called "read_vectored", but I can't use it at all for the stream object I have.
I have four modules. The client is sending messages and the server is receiving messages. Once the server receives the message, it tries to send the message to the MPSC channel. I put the receiver in the other .rs file where I intend to receive the message.
I am not getting any message on the receiver side.
Maybe an infinite loop on the server side creates a problem, but is there a way to make this channel communication working?
client.rs
use std::io::prelude::*;
use std::os::unix::net::UnixDatagram;
use std::path::Path;
use std::sync::mpsc;
pub fn tcp_datagram_client() {
pub static FILE_PATH: &'static str = "/tmp/datagram.sock";
let socket = UnixDatagram::unbound().unwrap();
match socket.connect(FILE_PATH) {
Ok(socket) => socket,
Err(e) => {
println!("Couldn't connect: {:?}", e);
return;
}
};
println!("TCP client Connected to TCP Server {:?}", socket);
loop {
socket
.send(b"Hello from client to server")
.expect("recv function failed");
}
}
fn main() {
tcp_datagram_client();
}
server.rs
use std::os::unix::net::UnixDatagram;
use std::path::Path;
use std::str::from_utf8;
use std::sync::mpsc::Sender;
fn unlink_socket(path: impl AsRef<Path>) {
let path = path.as_ref();
if path.exists() {
if let Err(e) = std::fs::remove_file(path) {
eprintln!("Couldn't remove the file: {:?}", e);
}
}
}
static FILE_PATH: &'static str = "/tmp/datagram.sock";
pub fn tcp_datagram_server(tx: Sender<String>) {
unlink_socket(FILE_PATH);
let socket = match UnixDatagram::bind(FILE_PATH) {
Ok(socket) => socket,
Err(e) => {
eprintln!("Couldn't bind: {:?}", e);
return;
}
};
let mut buf = vec![0; 1024];
println!("Waiting for client to connect...");
loop {
let received_bytes = socket.recv(&mut buf).expect("recv function failed");
println!("Received {:?}", received_bytes);
let received_message = from_utf8(&buf).expect("utf-8 convert failed");
tx.send(received_message.to_string());
}
}
message_receiver.rs
use crate::server;
use std::sync::mpsc;
pub fn handle_messages() {
let (tx, rx) = mpsc::channel();
server::tcp_datagram_server(tx);
let message_from_tcp_server = rx.recv().unwrap();
println!("{:?}", message_from_tcp_server);
}
main.rs
mod server;
mod message_receiver;
fn main() {
message_receiver::handle_messages();
}
Once the TCP client is connected:
TCP client Connected to TCP Server UnixDatagram { fd: 3, local: (unnamed), peer: "/tmp/datagram.sock" (pathname) }
I receive no messages on the channel receiver end:
Waiting for client to connect...
Maybe an infinite loop on the server side creates a problem
Yes, quite literally, your server code does an infinite loop to handle continuously messages from the client(s). So the call to tcp_datagram_server never returns.
but is there a way to make this channel communication working?
Of course, it seems you are simply missing a second thread for your message_receiver. Wrapping your tcp_datagram_server(tx) in std::thread::spawn should do it. You could also add a loop to keep processing requests to match the one in tcp_datagram_server:
pub fn handle_messages() {
let (tx, rx) = mpsc::channel();
std::thread::spawn(|| tcp_datagram_server(tx));
loop {
let message_from_tcp_server = rx.recv().unwrap();
println!("{}", message_from_tcp_server);
}
}
I'm new to Rust and I'm trying to configure a simple tcp socket server which will listen to connections and will reply with the same message that received.
The thing is, this works as I want except when connecting with multiple clients.. The first client that connects will send and receive the messages but if a second client connects, the first one keeps working but the second never receives messages, in fact the message never gets in the code that will handle it. And if I disconnect the first socket, the server will start spamming forever that received a message from the first socket with the same content than the last message it sent.
I am pretty sure I did something wrong in my code but I can't find it
This is my server struct:
use std::collections::HashMap;
use std::io::Read;
use std::io::Write;
use std::net::Shutdown;
use std::net::TcpListener;
use std::net::TcpStream;
use std::str;
use std::sync::{Arc, RwLock};
use threadpool::ThreadPool;
#[derive(Clone, Debug)]
pub struct Server {
id: Arc<RwLock<u32>>,
connections: Arc<RwLock<HashMap<u32, TcpStream>>>,
url: String,
thread_pool: ThreadPool
}
impl Server {
pub fn new(url: String) -> Server {
let server = Server {
id: Arc::new(RwLock::new(0)),
connections: Arc::new(RwLock::new(HashMap::new())),
url,
thread_pool: ThreadPool::new(10)
};
server
}
pub fn start(&self) {
let listener = TcpListener::bind(&self.url).expect("Could not start the server");
println!("Server started succesfully");
for stream in listener.incoming() {
match stream {
Ok(stream) => {
let mut self_clone = self.clone();
self.thread_pool.execute(move || {
self_clone.on_client_connect(stream.try_clone().unwrap());
});
}
Err(error) => eprintln!("Error when tried to use stream. Error = {:?}", error),
}
}
}
fn on_client_connect(&mut self, stream: TcpStream) {
println!("Client connected from {}", stream.local_addr().unwrap());
let mut id = self.id.write().unwrap();
{
*id += 1;
}
self.connections
.write()
.unwrap()
.insert(*id, stream.try_clone().unwrap());
let mut stream = stream.try_clone().unwrap();
let mut buffer = [0; 1024];
while match stream.read(&mut buffer) {
Ok(size) => {
println!(
"Message received from {} - {}",
id,
str::from_utf8(&buffer).unwrap()
);
stream.write_all(&buffer[0..size]).unwrap();
true
}
Err(error) => {
println!(
"Error when reading message from socket. Error = {:?}",
error
);
stream.shutdown(Shutdown::Both).unwrap();
false
}
} { }
}
}
And in my main.rs I'm just calling the connect function and the server starts working
In this piece of code in your on_client_connect function, you're aquiring a read lock for self.id:
let mut id = self.id.write().unwrap();
{
*id += 1;
}
However, the id variable, which holds the lock, is not released until it drops at the end of the function. This means that all other clients will wait for this lock to be released, which won't happen until the function currently holding the lock has completed (which happens when that client disconnects).
You can solve this by rewriting the above code to only keep the lock while incrementing, and then storing the ID value in a variable:
let id: u32 = {
let mut id_lock = self.id.write.unwrap();
*id_lock += 1;
*id_lock
// id_lock is dropped at the end of this block, so the lock is released
};
Even better, you can use AtomicU32, which is still thread-safe yet does not require locking at all:
use std::sync::atomic::{AtomicU32, Ordering};
struct {
id: Arc<AtomicU32>,
// ...
}
// Fetch previous value, then increment `self.id` by one, in a thread-safe and lock-free manner
let id: u32 = self.id.fetch_add(1, Ordering::Relaxed);
Also, when the connection is closed your code goes into an infinite loop because you're not handling the case where stream.read() returns Ok(0), which indicates that the connection was closed:
while match stream.read(&mut buffer) {
Ok(0) => false, // handle connection closed...
Ok(size) => { /* ... */ }
Err(err) => { /* ... */ }
} {}