Hi... On Wed, Feb 20, 2008 at 3:44 PM, Hayim <hayim@xxxxxxxxxxxx> wrote: > Don't understand. > > This trace is from an HTTP server, where a file is being sent from the > server to a client. Ah sorry, I thought you were showing the trace from your own raw socket program. > After the "GET" command the client does not send anything else, so why > should the ack sequences be different? Well, to the best I know , same ACK seq happens when: 1. during 3 way handshake 2. retransmission, thus ack-ing same packet(s). So, in normal tcp operations, usually (at least AFAIK) seq number should grow higher during data communication > The server sends packets as much as the TCP-window allows. > But here, a packet was sent twice, with no apparent reason. TTL (or any other flag that means "time out) is too low? > Also, note my first mail, that packets sent by this server are not sent > through the TCP-stack (using 'send' on a TCP socket), but instead, they > are sent on a raw socket, where the TCP is implemented by my user space > application. I see...that includes that HTTP session, where your tcp/ip stack acts as the server or client? > I can most positively say that the user space application called sendto > only once for each packet, whereas one packet was sent twice. This is something I can't answer right now.... probably because I am not so keen about the whole jungle of TCP/IP regards, Mulyadi. -- To unsubscribe from this list: send an email with "unsubscribe kernelnewbies" to ecartis@xxxxxxxxxxxx Please read the FAQ at http://kernelnewbies.org/FAQ
