On 8/7/2010 11:48 PM, Danny Mayer wrote: > On 8/6/2010 10:30 PM, Cindy Huyser wrote: >> On Aug 6, 8:17 pm, Danny Mayer <[email protected]> wrote: >>> On 8/6/2010 12:29 PM, Cindy Huyser wrote: >>> >>> >>> >>>> Using a packet sniffer, I verified that no traffic addressed to the >>>> server (or coming from it) is going across the ethernet interface when >>>> my configuration file specifies the IPv6 server. I also verified that >>>> I saw ICMPv6 packets going across when I pinged the server, along with >>>> the neighbor solicitation and advertisement. The server is in the XP >>>> host's destination cache, and also is listed in the "neighbors" list. >>>> I'm puzzled as to why there's not even an attempt at neighbor >>>> discovery. >>> >>>> Maybe the trouble shows up in the line "findlocalinterface: kernel >>>> maps fe80::290:fbff:fe80:6aff to ::", and the subsequent use of the >>>> wildcard (or maybe not -- the address for the server is not local to >>>> the host, but the host should be able to figure out that it needs to >>>> send a neighbor discovery solicitation from it single ethernet >>>> interface!). Can anyone out there shed any light on this? >>> > > Yes, that appears to be the problem. I am not sure why it is happening > and I will have to investigate. It should never be using the wildcard > interface and the fact that it did indicates that it failed to find the > right interface and that is the interface of last resort. It should be > using the link-local interface that the system has configured. Please > file a bug report on this.
The getsockname() function is returning the wildcard address on Windows XP which is not the correct answer. I'm not sure why that is but I suspect that the function was never upgraded for IPv6 support. I'm not going to spend time testing that theory but I will try and get it fixed in some other way. Danny _______________________________________________ questions mailing list [email protected] http://lists.ntp.org/listinfo/questions
