Re: [dhcwg] Questions about layer two relay agents in dhcpv4...

[Bharat Joshi <[e-mail]>]

Please see in line.

Thanks,
Bharat

> > I do think this is a problem. TR-101, in which I participated in
> > defining the L2RA requirements for EtherDSL, specifically requires the
> > L2RA to remove option-82 from all packets:
> >  [...]
> 
> Yes.

We (author's of the l2ra draft) also think that DHCP replies containing RAIO should never reach the client. As Pavan mentioned in his reply, we had to add that paragraph because someone pointed out that there may be Layer 2 Relay Agents which are configured not to intercept unicast DHCP replies and hence would not be able to remove RAIO.

We can simply remove this point and make sure that draft clearly articulates the need of L2 RA to intercept all unicast/broadcast replies to remove RAIO option.

> This is complicated by the fact that the l2ra draft is informational, not standards-track.   So essentially it just reports what people are doing with regards to l2ra, rather than specifying how to do l2ra.  I'm finding myself adding text to the relay agent encapsulation draft specifying how to do l2ra because of this.
> 

A little bit of history may help here. We started with a Lease Query extension draft for L2 RA and at that point, we found out that the L2 RA term itself is not defined in IETF. Few DHC WG folks encouraged us to come up with an informational draft to define L2 RA so that the other draft can be understood in a better way. We were asked to explain various L2 RA implementations that exist. We kept it as informational draft as we were just providing information on how L2 RA behaves in existing network configurations.

> Personally, I'd like to see the l2ra draft aimed at standards track, and specifying the correct way to do l2ra, rather than, as it does now, simply documenting various implementation strategies, some of which are clearly (to me!) broken.
> 

We can surely do that if every body agrees to it.

> I'm curious to know what the authors of the current draft think about this.   I'm pretty sure we've talked about this before, so I apologize for re-visiting well-trod ground, but I'm seeing the problem with new eyes now... :'}

There is no problem in re-visiting something, the only issue I feel is the time it generally takes :(

As mentioned above, we are fine with bringing out a standard track draft which aim to define the Layer 2 Relay Agent functionality. We will wait for people to respond before we do that.