I've got a TCP socket which reads data. When an error occurs when reading the data, I return an undef (NULL) value. Errors can be caused by badly formatted messages or broken sockets. Can someone tell me if there is a specific function which returns the status of a socket?
There are three ways to detect whether the socket is open or closed, but neither of them are 100% full proof.
The first is to attempt a read on the socket as follows:
my $ret = recv($sockfd, $buff, 1, MSG_PEEK | MSG_NOWAIT);
If the socket has went through an orderly shutdown, i.e. the peer called shutdown for writing or called close AND the FIN packet has arrived then this call will result in a 0 length read indicating a closed socket. This also helps if your peer application crashed since the OS will close the connection and send a FIN. However, if your peer machine has crashed or your peer application has locked up this won't help you since each end of the connection maintains independent state.
A second way to detect a broken connection is by probing your peer. If you send a 0 length packet to your peer ( which is should be able to handle ) and the application has crashed then you send a second 0 length packet your application will get the SIG_PIPE signal indicating a broken pipe.
Another way to deal with this issue is to use an application level heartbeat in which the peers periodically send a heartbeat packet to each other indicating that they are alive and functioning properly.
One last option is to use the SO_KEEPALIVE socket option, although this is of limited use since it will only detect a broken socket after approximately 2 hours of inactivity.
If you really must know fairly quickly when a connection is broken, then the most reliable option is probably going to be the application level heartbeat.
doh!, the answer was obvious in retrospect, use the connected call.
$socket = IO::Socket::INET(localhost, 1000);
die "no connection" unless $socket -> connected();
$socket -> send("your face here for $20");
die "socket is dead" unless $socket -> connected();
$socket -> recv($data);
Related
I'm writing simple http server.
I want to shutdown socket after server send all data.
I considered that compare return byte of write() to socket with actuall content length, but I had read that the return value just means that data moved to send-buffer of the socket. (Im not sure and I don't know how can I check it)
If so, can I shutdown the socket just after check the bytes are same? What if the datas sended need to be retransmitted at TCP level after server send FIN flag?
The OS does not discard data you have written when you call shutdown(SHUT_WR). If the other end already shut down its end (you can tell because you received 0 bytes) then you should be able to close the socket, and the OS will keep it open until it has finished sending everything.
The FIN is treated like part of the data. It has to be retransmitted if the other end doesn't receive it, and it doesn't get processed until everything before it has been received. This is called "graceful shutdown" or "graceful close". This is unlike RST, which signals that the connection should be aborted immediately.
This question already has answers here:
How to detect when socket connection is lost?
(2 answers)
Closed 1 year ago.
I'm trying - and failing - to get a perl server to detect and get rid of connection with a client who broke the connection. Everywhere I looked, the suggested method is to use the socket's ->connected() method, but in my case it fails.
This is the server absolutely minimized:
#!/usr/bin/perl
use IO::Socket;
STDOUT->autoflush(1);
my $server = new IO::Socket::INET (
Listen => 7,
Reuse => 1,
LocalAddr => '192.168.0.29',
LocalPort => '11501',
Proto => 'tcp',
);
die "Could not create socket: $!\n" unless $server;
print "Waiting for clients\n";
while ($client = $server->accept()) {
print "Client connected\n";
do {
$client->recv($received,1024);
print $received;
select(undef, undef, undef, 0.1); # wait 0.1s before next read, not to spam the console if recv returns immediately
print ".";
} while( $client->connected() );
print "Client disconnected\n";
}
I connect to the server with Netcat, and everything works fine, the server receiving anything I send, but when I press ctrl-C to interrupt Netcat, 'recv' is no longer waiting, but $client->connected() still returns a true-like value, the main loop never returns to waiting to the next client.
(note - the above example has been absolutely minimized to show the problem, but in the complete program the socket is set to non-blocking, so I believe I can't trivially depend on recv returning an empty string. Unless I'm wrong?)
connected can't be used to reliably learn whether the peer has initiated a shutdown. It's mentioned almost word for word in the documentation:
Note that this method considers a half-open TCP socket to be "in a connected state". [...] Thus, in general, it cannot be used to reliably learn whether the peer has initiated a graceful shutdown because in most cases (see below) the local TCP state machine remains in CLOSE-WAIT until the local application calls "shutdown" in IO::Socket or close. Only at that point does this function return undef.
(Emphasis mine.)
If the other end disconnected, recv will return 0. So just check the value returned by recv.
while (1) {
my $rv = $client->recv(my $received, 64*1024);
die($!) if !defined($rv); # Error occurred when not defined.
last if $received eq ""; # EOF reached when zero.
print($received);
}
Additional bug fix: The above now calls recv before print.
Additional bug fix: Removed the useless sleep. recv will block if there's nothing received.
Performance fix: No reason to ask for just 1024 bytes. If there's any data available, it will be returned. So you might as well ask for more to cut down on the number of calls to recv and print.
Note that even with this solution, an ungraceful disconnection (power outage, network drop, etc) will go undetected. One could use a timeout or a heartbeat mechanism to solve that.
for the life of me, i can't seem to figure out how to get a standard TCP socket connection to reconnect after a disconnect, particularly in the context of an IO::Async::Loop
some basics:
#!/usr/bin/perl
use strict;
use warnings;
use Socket;
use IO::Async::Loop;
use IO::Async::Stream;
use IO::Socket;
use Time::HiRes qw(usleep);
# standard event loop
my $loop = IO::Async::Loop->new;
# notification service socket connection; we only write outgoing
my $NOTIFY = IO::Socket::INET->new(
PeerHost => $a_local_network_host,
PeerPort => $comm_port,
Proto => 'tcp',
Type => SOCK_STREAM,
ReuseAddr => 1,
Blocking => 0
) or warn("Can't connect to NOTIFY: $!\n");
setsockopt($NOTIFY, SOL_SOCKET, SO_KEEPALIVE, 1);
# main interface element via IO::Async
my $notifystream = IO::Async::Stream->new(
handle => $NOTIFY,
on_read => sub {
my ( $self, $buffref, $eof ) = #_;
# here's where we need to handle $eof if the remote goes away
if($eof) {
# i have tried several things here
usleep(200000); # give the remote service some milliseconds to start back up
# process fails if remote is not back up, so i know the timeout is 'good enough' for this test
# attempt to reconnect. have also tried recreating from scratch
$NOTIFY->connect("$a_local_network_host:$comm_port");
# this doesn't seem to have any effect
$self->configure(handle=>$NOTIFY);
}
}
);
$loop->add( $notifystream );
# kickstart the event loop
$loop->run;
### -- Meanwhile, elsewhere in the code -- ###
$notifystream->write("data for notification service\n");
in reality, there are many more things going on in the loop. i also have more sophisticated ways to test for socket closed or broken, further error handlers on the $notifystream, and a better timeout/backoff for reconnecting to the remote service, however this should show the main crux of what i'm doing.
when the remote server goes away for any reason, i'd like to attempt to reconnect to it without disrupting the rest of the system. in most cases the remote sends eof cleanly because it's intentionally rebooting (not my choice, just something i have to deal with), but i'd also like to handle other communication errors as well.
in practice, the above code acts as though it works, however the remote service no longer receives further write calls to the $notifystream. no errors are generated, the $notifystream happily takes further writes, but they are not delivered to the remote.
i have a feeling i'm doing this wrong. i'm not looking to rewrite the rest of the application's event loop, so please no 'just use AnyEvent'-type responses -- really hoping to gain a better understanding of how to reconnect/reuse/recreate the variables in use here (IO::Socket::INET and IO::Async::Stream) to compensate when a remote server is temporarily unavailable.
Any suggestions or references towards this goal are welcome. Thanks!
-=-=-=-=-
to summarize errors i have (and have not) received:
if i leave no usleep, the reconnect (or recreation) of the base socket will fail due to the remote service being unavailable.
if i attempt to recreate the socket from scratch and then 'configure' the stream, i get 'can't call method sysread on undefined' which leads me to believe the socket is not recreated correctly.
at no time do the stream's built in 'on_read_error' or 'on_write_error' handlers fire, regardless of how much i write to the socket with the current code, although if i destroy the socket entirely these will generate an error.
the socket simply seems to still be active after i know it has closed, and a reconnect does not seem to change anything. no errors are generated, but the socket is not being written to.
is there different syntax for reconnecting to an IO::Socket::INET socket? so far calls to ->connect() or rebuilding from scratch seem to be the only options for a closed connection, and neither seem to work.
You simply cannot connect an existing socket multiple times. You can only close the socket and create a new socket. This has nothing to do with IO::Socket::INET, IO::Async::Stream or even Perl but this is how the sockets API works.
In detail: The local socket actually never got disconnected, i.e. it is still configured to send data from a specific local IP address and port and to a specific address and port. Only, that the sending will no longer worked because the underlying TCP connection is broken or closed (i.e. FIN exchanged). Since there is no API to unbind and unconnect a socket the only way is to close it and create a new one which is unbound and unconnected until connect is called. This new socket then might or might not get the same file descriptor as the previous one.
I am make a application base on lwip,the applcation just send data to the server;
When my app works for some times (about 5 hours),I found that the send thread hung in send() function,and after about 30min send() return 0,and my thread run agin;
In the server side ,have make a keepalive,its time is 5min,when my app hungs,5min later the server close the sockect,but my app have not get this,still hungs in send() until 30min get 0 return; why this happen?
1: the up speed is not enough to send data,it will hungs in send?
2: maybe the server have not read data on time,and it make send buff full and hungs?
how can i avoid these peoblems in my code ? I have try to set TCP_NODELAY,SO_SNDTIMEO and select before send,but also have this problem.
send() blocks when the receiver is too far behind the sender. recv() returns zero when the peer has closed the connection, which means you must close the socket and stop reading.
I am implementing a server in which i listen for the client to connect using the accept socket call.
After the accept happens and I receive the socket, i wait for around 10-15 seconds before making the first recv/send call.
The send calls to the client fails with errno = 32 i.e broken pipe.
Since i don't control the client, i have set socket option *SO_KEEPALIVE* in the accepted socket.
const int keepAlive = 1;
acceptsock = accept(sock, (struct sockaddr*)&client_addr, &client_addr_length)
if (setsockopt( acceptsock, SOL_SOCKET, SO_KEEPALIVE, &keepAlive, sizeof(keepAlive)) < 0 )
{
print(" SO_KEEPALIVE fails");
}
Could anyone please tell what may be going wrong here and how can we prevent the client socket from closing ?
NOTE
One thing that i want to add here is that if there is no time gap or less than 5 seconds between the accept and send/recv calls, the client server communication occurs as expected.
connect(2) and send(2) are two separate system calls the client makes. The first initiates TCP three-way handshake, the second actually queues application data for transmission.
On the server side though, you can start send(2)-ing data to the connected socket immediately after successful accept(2) (i.e. don't forget to check acceptsock against -1).
After the accept happens and I receive the socket, i wait for around 10-15 seconds before making the first recv/send call.
Why? Do you mean that the client takes that long to send the data? or that you just futz around in the server for 10-15s between accept() and recv(), and if so why?
The send calls to the client fails with errno = 32 i.e broken pipe.
So the client has closed the connection.
Since I don't control the client, i have set socket option SO_KEEPALIVE in the accepted socket.
That won't stop the client closing the connection.
Could anyone please tell what may be going wrong here
The client is closing the connection.
and how can we prevent the client socket from closing ?
You can't.