Re: [Roll] [roll] #110: trickle-mcast: do applications receive all multicast, or just MPL encapsulated ones

[Robert Cragie <[e-mail]>]

Hi Jonathan,

Comments inline.

Robert

On 01/11/2012 6:51 PM, Jonathan Hui (johui) wrote:
> HI Robert,
>
> I'm not sure I follow you.  If we require MPL forwarders to drop packets sent to FF0X::MPL and do not contain an MPL Option, how do we expect one to encapsulate it for forwarding within the MPL domain?
<RCC>I wasn't suggesting that. If you look at the definitions I 
suggested, any original non-MPL multicast packet is always encapsulated, 
not dropped, irrespective of its destination address (unless link local, 
which is never likely to happen).</RCC>
>
> I suggested in another email that encapsulation be required unless the IPv6 Destination Address be equal to FF0X::MPL.  If it is anything else, then encapsulation is required.
<RCC>I was saying that the FF0X::MPL address should also have an 
accompanying MPL option otherwise if a MPL forwarder encountered an 
original non-MPL multicast packet with destination address FF0X::MPL, it 
would (probably) end up encapsulating it.</RCC>
>
> Also, since Peter would like to use unicast prefix-based multicast addresses, there is a one-to-one mapping between the MPL domain and the (address scope, group identifier) tuple.
<RCC>That seems to make sense but I would like to see some more details 
on that proposal</RCC>
>
> --
> Jonathan Hui
>
> On Nov 1, 2012, at 11:21 AM, Robert Cragie <[e-mail]> wrote:
>
>> We could say that address has to be subscribed to by all MPL forwarders and SHOULD (MUST?) be used with a MPL Option. If the language in the draft were tightened up a bit regarding packet types, this would mean that if for some reason a packet to FF0X::MPL was originated without a MPL Option, it would be harmlessly (but wastefully) encapsulated and the chances are anything beyond the MPL domain would not be subscribed so would be ignored for that reason. Of course, if used with a MPL Option, no encapsulation is needed, there is a one-to-one mapping between the MPL domain and the address scope and it would not be forwarded on interfaces which don't understand MPL.
>>
>> Robert
>>
>>
>> On 01/11/2012 4:08 PM, Jonathan Hui (johui) wrote:
>>> The FF0X::MPL multicast address serves to identify a group of MPL forwarders within an MPL domain.
>>>
>>> I think if any device sends a packet to FF0X::MPL, it should be delivered to and processed by all devices subscribed to that address.
>>>
>>> Thoughts from others?
>>>
>>> --
>>> Jonathan Hui
>>>
>>> On Nov 1, 2012, at 7:47 AM, roll issue tracker <[e-mail]> wrote:
>>>
>>>> #110: trickle-mcast: do applications receive all multicast, or just MPL
>>>> encapsulated ones
>>>>
>>>> There's still the case that multicasts from non-MPL nodes are received by
>>>> MPL-forwarder-nodes
>>>> (presumably). Should there be guidelines here? Example: if an application
>>>> on my MPL-forwarder node
>>>> joins multicast group FF05::1234, will the application then receive IP
>>>> multicasts sent
>>>> with destination FF05::1234, or will it only receive those IP multicasts
>>>> to FF05::1234
>>>> that were delivered encapsulated in an FF0X::MPL packet ?
>>>>
>>>> That decision could be out of scope; but on the other hand may lead to
>>>> different
>>>> implementers making different choices here.
>>>>
>>>> -- 
>>>> -----------------------------+---------------------------------------------
>>>> Reporter:  mcr@…            |      Owner:  draft-ietf-roll-trickle-mcast@…
>>>>      Type:  enhancement      |     Status:  new
>>>> Priority:  major            |  Milestone:
>>>> Component:  trickle-mcast    |    Version:
>>>> Severity:  In WG Last Call  |   Keywords:
>>>> -----------------------------+---------------------------------------------
>>>>
>>>> Ticket URL: <https://svn.tools.ietf.org/wg/roll/trac/ticket/110>
>>>> roll <http://tools.ietf.org/wg/roll/>
>>>>
>>> _______________________________________________
>>> Roll mailing list
>>> [e-mail]
>>> https://www.ietf.org/mailman/listinfo/roll
>>>
>>
>> _______________________________________________
>> Roll mailing list
>> [e-mail]
>> https://www.ietf.org/mailman/listinfo/roll
>