Re: [dhcwg] RFC3315 and Hop Limit

On Oct 21, 2014, at 5:53 PM 10/21/14, Michael Richardson <mcr+ietf@sandelman.ca> wrote:

> 
> I could not find any advice in RFC3315 as to what to set the IPv6
> HOP Limit to for multicast'ed DHCPv6 messages.
> 
> I would have thought that the sender should set it to 255, and the
> sender should validate that it is 255, to be sure that it's onlink.
> This is what mDNS/Bonjour does, but maybe 3315 predates that thinking.

I don't think the 255 hop count limit from RFC 6762 applies.  RFC 6762 mentions the use of the 255 hop count limit, but only for backward compatibility.  RFC 6762 does not mandate the use of the 255 hop count limit as a mandatory requirement of mDNS.  I read RFC 6762 to specify that mDNS depends on the network infrastructure for limiting the distance of over which datagrams with a link-scope multicast address are forwarded.

DHCPv6 uses a link-scope multicast address for All_DHCP_Relay_Agents and a site-scope multicast address for All_DHCP_Servers.  In my opinion, it is fine for DHCPv6 to trust the network infrastructure to limit the forwarding of datagrams sent to both of those multicast addresses.

- Ralph

> Should 3315bis say something different?
> (Clearly, it has to be backwards compatible for awhile)
> 
> --
> Michael Richardson <mcr+IETF@sandelman.ca>, Sandelman Software Works
> -= IPv6 IoT consulting =-
> 
> 
> 
> _______________________________________________
> dhcwg mailing list
> dhcwg@ietf.org
> https://www.ietf.org/mailman/listinfo/dhcwg