Re: Why has RFC 4941 been designed in such a way, that it might cause address conflicts?

[Markus Hanauska <[e-mail]>]

On 2011-03-15, at 19:27 , Philip Homburg wrote:

> If you just need stable addresses, then you can also put your own random
> numbers in DHCP. 

Of course everything will be fine if you exclusively use DHCPv6. You can have a pool with addresses for well known devices, so the same device always gets the same address from the pool, while all other devices get addresses from another pool that are more or less random.

However, there are devices that support IPv6 and stateless auto-configuration, possibly even the privacy extension, but not DHCPv6. And if a device supports both, there is no way to force the device to use DHCP; if it is "nice" it may listen to the RA managed flag, but does the RFC say that it must listen to this flag and that users may not override this flag, by configuration their device to always generate stateless addresses? Not as far as I can see and the meaning of this flag is not exactly defined (this flag does not imply DHCP, it could also imply a different protocol or manual configuration only) and I saw that even deprecation of the flag has been discussed publicly.

I just thought it would be nice if DHCP, manual configuration and stateless auto-configuration can always play together nicely within the same network, even when privacy extension is being used. Without privacy extension this is pretty much the case, except for MAC address conflicts like you mentioned previously, but when privacy extension enters the game (and don't get me wrong, I think privacy extension is a good thing and it is important that such an extension exists) this is not guaranteed any longer.

Regards,
Markus