On Mon, 2006-09-25 at 14:29 +0200, Janusz U. wrote: > isn't the TCP keep alive responsible for that? I have similar problem with > PPP. Link failed in middle of physical connection and lwIP does not detect > the situation. All other systems have timeouts!
TCP keep-alives are a "MAY" rather than "MUST" or even "SHOULD" part of the TCP specification: http://www.freesoft.org/CIE/RFC/1122/114.htm but lwIP does include support for them, both as a sender and receiver of such probes. However, to be useful, the probes must be enabled on that connection - I think the default is "off", but haven't checked, and the connection must be idle for a certain time before the probe is sent, and this is a long time by default - something like an hour or two. The times can be configured with socket options if you're using the sockets API. They're not designed for detecting and responding rapidly to a sudden physical layer death, more for cleaning up connections that have been idle and no response received to the probes for a long time. Kieran _______________________________________________ lwip-users mailing list [email protected] http://lists.nongnu.org/mailman/listinfo/lwip-users
