Re: [Rift] AD Review of draft-ietf-rift-rift-12 (Part 3)

[Jordan Head <[e-mail]>]

Comments in line as jhead>>

On 2/10/23, 8:00 AM, "Alvaro Retana" <[e-mail] <mailto:[e-mail]>> wrote:


[External Email. Be cautious of content]




On September 13, 2022 at 8:39:48 AM, Jordan Head wrote:




...
> [major] "neighbors did not see any valid LIEs from a neighbor"
>
> The text is not precise. You're referring to what the local node has
> received (or not) -- using "neighbors did not see...from a neighbor"
> gives both the impression that the state is controlled by others, or
> that the local node is not involved (because the definition is based
> on its neighbors).
>
> I was going to make a suggestion, but the description is not clear to
> me. Reading from the FSM... In OneWay, the node hasn't seen a new
> neighbor (NewNeighbor) but it may have sent a LIE already (not
> necessary), right?
>
> Suggestion>
> OneWay: Initial state. In this state the node has not received
> a valid LIE from another node on the interface.
>
> jhead>> Fixed in new language.


I wasn't clear. The new text says:


OneWay: initial state the FSM is starting from. In this state
the neighbors did not receive any valid LIEs from a neighbor
after the state was entered.


The part that I'm having a problem with is where it says that "the
neighbors did not receive any valid LIEs from a neighbor". Which
neighbors did not receive something from which neighbor? It seems
that one of these neighbors (maybe the first one?) is the local node.
Right?


Suggestion>
OneWay: initial state the FSM is starting from. In this state
the router has not received a valid LIE from a neighbor.

    jhead>> Agreed and fixed.




...
> 1865 * MultipleNeighborsWait: occurs normally when more than two nodes
> 1866 see each other on the same link or a remote node is quickly
> 1867 reconfigured or rebooted without regressing to `OneWay` first.
> 1868 Each occurrence of the event SHOULD generate a clear, according
> 1869 notification to help operational deployments.
>
> [minor] Maybe I don't remember this from the overview -- where is it
> specified that only point-to-point links (or links with only two
> nodes) are supported?
>
> jhead>> This was in the requirements section but was removed based on previous
> review/comments, would you like us to re-introduce it?


A quick mention (in the Introduction maybe) that rift expects all the
links to be p2p, or only have two nodes would be great.

    jhead>> How about in the Terminology section? I added a bit more qualification to the "ThreeWay Adjacency" definition as well:

        RIFT tries to form a unique adjacency between two nodes over a point-to-point interface and exchange local...

    jhead>> Same for the "Neighbor" definition. And, for what it's worth it also implies the same point by stating that the neighbor structure is not shared:

        Multiple adjacencies can be formed to a remote node via parallel point-to-point interfaces but such adjacencies are *not* sharing a neighbor structure.


Juniper Business Use Only