RE: [dhcwg] Address selection for Information-request/Reply messa ges

The problem comes about if we change All_DHCP_Servers to include relays to
listen for messages on this address.

Because now both the relays and the routers would forward messages, and the
relays may forward messages forwarded by routers, you could (depending on
the number of relays and routers), get a fairly large packet storm.

The DHCPv6 protocol will work fine and doesn't break. Though the load placed
on the servers and network may not be desireable. That is the point.

Bottom line, IMHO, this is a bad idea. We shouldn't have relays listening
for All_DHCP_Servers traffic. And, if we don't do that, allowing Information-Request
sent directly to All_DHCP_Servers would require site multicast to work and if
it doesn't or is disabled, these clients would fail to receive the requested
information if the server isn't local.

- Bernie

-----Original Message-----
From: Robert Elz [mailto:kre@munnari.oz.au]
Sent: Thursday, May 01, 2003 3:17 PM
To: Ralph Droms
Cc: dhcwg@ietf.org
Subject: Re: [dhcwg] Address selection for Information-request/Reply
messages 

    Date:        Thu, 01 May 2003 09:52:19 -0400
    From:        Ralph Droms <rdroms@cisco.com>
    Message-ID:  <4.3.2.7.2.20030501094136.00b8f828@funnel.cisco.com>

  | Infact, it would be important to note that a site MUST NOT both
  | implement multicast and configure relay agents to listen on
  | All_DHCP_Servers, as such a configuration would result in the
  | servers receiving multiple copies of client messages.

Surely that happens with relay agents anyway?   Any site with half a brain,
or more, configures multiple relay agents for redundancy (assuming it also
has multiple routers for redundancy of course).   If the protocol can't
handle multiple copies of messages being delivered to the servers, it badly
needs fixing (but I actually doubt it has that defect).

kre

_______________________________________________
dhcwg mailing list
dhcwg@ietf.org
https://www1.ietf.org/mailman/listinfo/dhcwg