Re: comments on rsvp-te-06

[Lou Berger <lberger@movaz.com>]

At 02:14 PM 12/10/2001, Zhi-Wei Lin wrote:

>Hi Lou, See in-line
>
>Lou Berger wrote:
>
> >> In section 7.2.1, both step 3 are the same description.
> >
> >
> > yes, this is intentional!
> >
>
>But then what you're saying is that for ingress-initiated teardown, the
>flow is:
>(1) Path
>(2) Resv
>(3) PathTear
>
>While egress initiated it is:
>(1) Resv
>(2) PathTear

this is correct.

>This means that ingress-initiated takes 3 message passes. In OIF UNI
>1.0, a different message flow is defined for ingress-initiated, that of:
>(1) Path
>(2) PathErr (with Path_State_Removed flag set)

how nice...

>Is this latter message flow allowed, and if not, why not?

Note that the defined procedures say "SHOULD" not "MUST".

> >> Based on the
> >> flow of description, I think the first step 3 should read:
> >> "Upon receiving the Admin Status Object with the Delete (D) bit set in
> >> the Path message, the egress node sends a PathErr message (with
> >>      ^^^^              ^^^^^^              ^^^^^^^         ^^^^^
> >> Path_State_Removed flag set) upstream to remove the LSP and normal RSVP
> >> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> >> processing takes place."


> >
> >
> > The egress triggers the ingress to do the delete.
> >
>
>See above. Actually, based on the original text, the ingress triggers
>the egress which then triggers the ingress to do the delete...seems like
>extra level of message...I thought that's one reason for defining the
>Path_State_Removed flag for PathErr...
>
> >> In section 9.1, a statement under "recovery time":
> >>         A value of 0xffffffff indicates that resynchronization may
> >>           occur at a rate selected by the receiver.
> >> while in section 9.5.2, the following statement is made:
> >>         A Recovery Time value of 0xffffffff indicates that the
> >>         Recovery Period is effectively infinite.
> >> which is correct? or am I misinterpreting this?
> >
> >
> > These statements are mutually consistent.  If you'd like, I can just
> > remove the text, but I think most will find the comment useful.
> >
>
>I see the consistency now. You should keep the text.

Glad to hear we're now in sync.

Lou

>Thanks
>Zhi