You have the inherent problem that a TCP connection is terminated by an explitic "FIN" message from the terminating party, followed by a FIN-ACK and ACK.

Refer to here.

If your client (or server) just powers off without closing the connection properly, that's it. The connection hangs up and no data will be transfered if it is not explicitly handled.

If you must know whether the client is still there, you must use a heartbeat meachanism. That is, after some period time without a response has passed, query the connection party with a "heartbeat" message and see if it responds after a given maximum time. The other way around would be to make the client send a periodic "Keep Alive" message which holds up the connection as long as it is sent repeadetly.

Refer to here and here.

You have the inherent problem that a TCP connection is terminated by an explicit "FIN" message from the terminating party, followed by a FIN-ACK and ACK.

Refer to here.

If your client (or server) just powers off without closing the connection properly, that's it. The connection hangs up and no data will be transfered if it is not explicitly handled.

If you must know whether the client is still there, you must use a heartbeat meachanism. That is, after some period time without a response has passed, query the connection party with a "heartbeat" message and see if it responds after a given maximum time. The other way around would be to make the client send a periodic "Keep Alive" message which holds up the connection as long as it is sent repeadetly.

Refer to here and here.

You have the inherent problem that a TCP connection is terminated by an explitic "FIN" message from the terminating party, followed by a FIN-ACK and FIN.

Refer to here.

If your client (or server) just powers off without closing the connection properly, that's it. The connection hangs up and no data will be transfered if it is not explicitly handled.

If you must know whether the client is still there, you must use a heartbeat meachanism. That is, after some period time without a response has passed, query the connection party with a "heartbeat" message and see if it responds after a given maximum time. The other way around would be to make the client send a periodic "Keep Alive" message which holds up the connection as long as it is sent repeadetly.

Refer to here and here.

You have the inherent problem that a TCP connection is terminated by an explitic "FIN" message from the terminating party, followed by a FIN-ACK and ACK.

Refer to here.

If your client (or server) just powers off without closing the connection properly, that's it. The connection hangs up and no data will be transfered if it is not explicitly handled.

If you must know whether the client is still there, you must use a heartbeat meachanism. That is, after some period time without a response has passed, query the connection party with a "heartbeat" message and see if it responds after a given maximum time. The other way around would be to make the client send a periodic "Keep Alive" message which holds up the connection as long as it is sent repeadetly.

Refer to here and here.