RAFIXD does it really work?

[Eric Vyncke]

I'm afraid that I've to bug you again :-) as I've no clue where to ask questions about RAFIXD.

When running it at home, it looks like Windows XP keeps the rogue RA in memory and keeps using it...

Note: as I'm running RAFIXD on Fedora rather than on *BSD, I had to modify it to use PF_PACKET rather than BPF to send the RA with lifetime=0...

Note2: running TCPDUMP on another machine indicates that the fix RA is sent from a MAC source address of all 0. Which is the expected behavior when you look at the code. 

Note3: in my set-up, the rogue RA is sent by the one and only IPv6 router of my network.

For info, here is a TCPDUMP of the fix RA:
20:28:13.395158 00:00:00:00:00:00 (oui Ethernet) > 33:33:00:00:00:01 (oui Unknown), ethertype IPv6 (0x86dd), length 70: (class 0x60, hlim 255, next-header: ICMPv6 (58), length: 16) fe80::204:27ff:fefd:5240 > ip6-allnodes: [icmp6 sum ok] ICMP6, router advertisement, length 16
        hop limit 0, Flags [none], pref medium, router lifetime 0s, reachable time 0s, retrans time 0s

In short, it seems to me that there are 2 problems with RAFIXD:
- does not work at least for Windows XP
- the source MAC address is the null address

Is the correct behavior of RAFIXD confirmed in real deployment?

Thanks in advance for any piece of information

-eric



-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.cluenet.de/pipermail/ipv6-ops/attachments/20071130/1657e3a4/attachment.htm