Re: DAD question

[Fernando Gont <[e-mail]>]

Hi, Fred,

On 08/10/2012 07:17 PM, Fred Baker (fred) wrote:
> Call this "making sure I'm on the same page as anyone else"…
> 
> RFC 4941 describes privacy addresses, and RFC 4291 describes an EID
> based on a MAC Address. RFC 4862 describes stateless address
> autoconfiguration, and uses RFC 4861's duplicate address detection
> mechanism.
> 
> My question is: what happens if any of them discovers that it has
> created an address that is already in use in the network?

For traditional SLAAC (i.e. EID based on the MAC address), SLAAC simply
fails.

IIRC, you get different results depending on whether DAD fails for a
link-local address, or it fails for a non-link-local address. For the
former, you may need to reboot your system. For the later, I seem to
recall that some systems "perform DAD", but do not care if DAD actually
fails.

Simplest answer: "Try it":

Run na6 (http://www.si6networks.com.ar/tools) as follows:

# ./na6 -i eth0 -j :: -c -o -e -L -vv

where:
"-i eth0": your NIC
"-j ::":   tells the tool to only respond to packets with the Src Addr
           set to the unspec addr (i.e., DAD packets)
"-c":      Set the solicited flag
"-o:"      Set the override flag (shouldn't matter in this case)
"-L":      Listen mode (wait for incoming ns'es)
"-vv":     Be very verbose



> There would appear to be two options: (1) "ah, OK, I guess I didn't
> really want to talk today"

This is what all implementations do for tradicional SLAAC (when they do
honor SLAAC --see above).


> (2) Following RFC 4941, guess again until
> one creates a unique address
> 
> Is it fair to assume that implementations do DAD and follow (2)? 

Might be fair to assume this for the temporary address, but it is
certainly not realistic to assume this for the traditional SLAAC address
(*) (**).

(*) Microsoft Windows allegedly replace the traditional SLAAC addresses
with a RFC4941-based scheme that simply does not cycle the addresses
over time -- hence "(2)" might apply to Windows.

(**) Will head in this direction for
draft-ietf-6man-stable-privacy-addresses

Cheers,
-- 
Fernando Gont
e-mail: [e-mail] || [e-mail]
PGP Fingerprint: 7809 84F5 322E 45C7 F1C9 3945 96EE A9EF D076 FFF1




-- 
Fernando Gont
SI6 Networks
e-mail: [e-mail]
PGP Fingerprint: 6666 31C6 D484 63B2 8FB1 E3C4 AE25 0D55 1D4E 7492