Bernie Volz (volz) wrote:
But realisticly, do you expect any old client to check for these "other
configuration" parameters and if they got them, what might they do? Drop
the packet? Well, that is what the client would essentially have done
anyway since it got no addresses. So, while a poorly implemented client
might well crash, I can't really see that being likely and it would be
worth it to get that client fixed as quickly as possible since it
doesn't adhere to the "be liberal in what you receive" IETF "rule".

If we ignore DHCP client implementations which crash when they receive an unexpected option, what is the worst case of a RFC 3315 client in this case? Is it just to ignore the options in the Advertise message and proceed to send a Request message?
That wouldn't be a big deal to me.

Regarding the other issue, why would a stateless only client ever say
"Oh, I'll send Solicit instead of Information-Request"? If they did that
today, they wouldn't get very far.

The server compatibility issue isn't about today.

The issue I see if we recommend that clients (which implement both RFC 3315 and 3736) always send a Solicit (when some bit is set in the RA telling it to use DHCP), then such a client will not interoperate with currently deployed 3736 DHCP servers. My understanding is that there is some deployment of 3736 DHCP servers (to do prefix delegation and DNS configuration, etc).

Do we know that there is zero deployment of 3736 servers today?

If we need to handle those, it seems like either
 - the client needs to try a few Solicits first, and if that fails, try
   an Information-Request. (Or try them in parallel??)
 - Have some indication in the RA that only 3736 service is available on
   the network, so that the client can do the Information-Request only.

The second alternative seems simpler to me.

   Erik

--------------------------------------------------------------------
IETF IPv6 working group mailing list
ipv6@ietf.org
Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------

Reply via email to