Re: WG Adoption request for draft-mirsky-bfd-mpls-demand

["Reshad Rahman (rrahman)" <[e-mail]>]

Hi Greg,

Please see inline.

From: Rtg-bfd <[e-mail]> on behalf of Greg Mirsky <[e-mail]>
Date: Monday, February 25, 2019 at 1:40 PM
To: Jeffrey Haas <[e-mail]>
Cc: "Reshad Rahman (rrahman)" <[e-mail]>, "[e-mail]" <[e-mail]>
Subject: Re: WG Adoption request for draft-mirsky-bfd-mpls-demand

Hi Jeff,
now with GIM6>>.

Regards,
Greg

On Mon, Feb 25, 2019 at 9:26 AM Jeffrey Haas <[e-mail]<mailto:[e-mail]>> wrote:
On Mon, Feb 25, 2019 at 09:10:16AM -0800, Greg Mirsky wrote:
> Hi Jeff,
> please find my answers in-line tagged GIM4>>.
>
> Regards,
> Greg
>
> On Sun, Feb 24, 2019 at 7:24 PM Jeffrey Haas <[e-mail]<mailto:[e-mail]>> wrote:
>
> > Greg,
> >
> > On Sun, Feb 24, 2019 at 09:53:20AM -0800, Greg Mirsky wrote:
> > > Hi Jeff,
> > > I'm glad that you feel that our discussion is helpful. I want to point
> > that
> > > the use of the Poll sequence to communicate to the remote BFD system in
> > the
> > > Concatenated Paths section is to relay the failure detected in the
> > > downstream segment of the multi-segment OAM domain. Also, section 6.8.17
> > > does not specify how the upstream BFD system responds to the situation:
> > >    Note that if the BFD session subsequently fails, the diagnostic code
> > will
> > >    be overwritten with a code detailing the cause of the failure.  It is
> > >    up to the interworking agent to perform the above procedure again,
> > >    once the BFD session reaches Up state, if the propagation of the
> > >    concatenated path failure is to resume.
> >
> > Correct.  That is up to the upstream to determine its behavior.
> >
> > > And so far we were discussing RFC 5880 though the scope of the draft is
> > on
> > > the use of Demand mode over MPLS LSP.
> >
> > MPLS does not magically change the behavior of demand mode specified in the
> > core RFC.
> >
> GIM4>> The draft defines how the head-end LER reacts to receiving the BFD
> control message with Diag set to Control Detection Time Expired and the
> Poll flag set:
>    Reception of such BFD control packet by the ingress
>    LER indicates that the monitored LSP has a failure and sending BFD
>    control packet with the Final flag set to acknowledge failure
>    indication is likely to fail.  Instead, the ingress LER transmits the
>    BFD Control packet to the egress LER over the IP network with:
>
>    o  destination IP address MUST be set to the destination IP address
>       of the LSP Ping Echo request message [RFC8029];
>
>    o  destination UDP port set to 4784 [RFC5883];
>
>    o  Final (F) flag in BFD control packet MUST be set;
>
>    o  Demand (D) flag in BFD control packet MUST be cleared.
>
>    The ingress LER changes the state of the BFD session to Down and
>    changes rate of BFD Control packets transmission to one packet per
>    second.  The ingress LER in Down mode changes to Asynchronous mode
>    until the BFD session comes to Up state once again.  Then the ingress
>    LER switches to the Demand mode.
>
>
> > > And the draft does describe how the
> > > BFD system acts after it receives the control message with Diag field set
> > > to Control Detection Time Expired, a.k.a. RDI, and the Poll flag set. In
> > > that, I consider, the draft is complimentary to RFC 5884 whose scope is
> > on
> > > the Asynchronous mode only.
> >
> > I continue to have problems understanding how the text in your draft is
> > intended to be different than 6.8.4 of RFC 5880.  Simply saying "we're
> > allowed to use demand mode" can't be it?
> >
> GIM4>> Section 6.8.4 does not specify that if the BFD system is in Demand
> mode and the bfd.LocalDiag is set to 1 (Control Detection Time Expired) the
> Poll sequence MAY, SHOULD or MUST be used to notify the remote BFD system.

I shall paste this one last time:

:   If Demand mode is active on either or both systems, a Poll Sequence
:   MUST be initiated whenever the contents of the next BFD Control
:   packet to be sent would be different than the contents of the
:   previous packet, with the exception of the Poll (P) and Final (F)
:   bits.  This ensures that parameter changes are transmitted to the
:   remote system and that the remote system acknowledges these changes.
GIM6>> RFC 5880 uses the term "parameter" in relation to the timers, and most are in section 6.8.3. The Diag field is defined not a parameter of a BFD session but as:
      A diagnostic code specifying the local system's reason for the
      last change in session state.
<RR> Use of the term “parameter” is indeed not very clear, elsewhere in 5880 “timing parameters” and “timer parameters” are used. But what is clear is “whenever the contents of the next BFD Control packet to be sent would be different than the contents of the previous packet, with the exception of the Poll (P) and Final (F) bits”, the next packet would be different in this case because of state/diag change. So as per my previous reply, my interpretation is that this is indeed covered by 5880.


If you've changed diag, you've changed the contents.  If you are running in
demand mode, you will send a poll.
GIM6>> If the interpretation of RFC 5880 is as you're suggesting, then draft-ietf-bfd-multipoint must be updated to the fact that when a MultipointTail detects that Control Detection Time Expired it MUST initiate the Poll sequence to the MultipointHead. And draft-ietf-bfd-multipoint-active-tail seems unnecessary as the same functionality ensured by the previously mentioned update.
<RR> I don’t have all the history on these 2 drafts but this text from draft-ietf-bfd-multipoint mentions no return path, so this would explain why tail does not do poll sequence.
   As multipoint transmissions are inherently unidirectional, this
   mechanism purports only to verify this unidirectional connectivity.
   Although this seems in conflict with the "Bidirectional" in BFD, the
   protocol is capable of supporting this use case.  Use of BFD in
   Demand mode enables a tail monitor availability of a multipoint path
   even without the existence of some kind of a return path to the head.
   As an option, if a return path from a tail to the head exists, the
   tail may notify the head of the lack of multipoint connectivity.
   Details of tail notification to the head are outside the scope of
   this document and are discussed in
   [I-D.ietf-bfd-multipoint-active-tail<https://tools.ietf.org/html/draft-ietf-bfd-multipoint-18#ref-I-D.ietf-bfd-multipoint-active-tail>].



If you're also saying that the ingress is NOT receiving a Down state change
from the egress as part of this and that the ingress moves to down just
because the Diag changes, that at least is clear, and is worth further
discussion.
GIM6>> Thank you for the suggestion. The draft states:
   The ingress LER changes the state of the BFD session to Down and
   changes rate of BFD Control packets transmission to one packet per
   second.  The ingress LER in Down mode changes to Asynchronous mode
   until the BFD session comes to Up state once again.
Can we discuss that?
<RR> Isn’t the paragraph covered in 6.6 of 5880?
   Note that
   the Demand bit MUST NOT be set unless both systems perceive the
   session to be Up (the local system thinks the session is Up, and the
   remote system last reported Up state in the State (Sta) field of the
   BFD Control packet).

Regards,
Reshad (no hat).


> > It will help clear this conversation if you simply state your changes in
> > behavior vs. 5880 and 5884.
> >
> GIM4>> I've never stated or hinted that the draft is to update RFC 5884.
> The scope of RFC 5884 is the use of BFD in the Asynchronous mode over MPLS
> LSPs. As stated in section 6 RFC 5884:
>
> BFD demand mode is outside the scope of this specification.

You seem to be confused about how this boilerplate text is used.

If there are no changes to procedure, existing procedure applies - it is
simply not discussed in this document.

If there is changes to procedure (what we are trying to determine), then
further discussion is warranted.
GIM6>> I don't consider the switch to the Demand mode as "change to procedure" defined in RFC 5884 because the Demand mode is explicitly outside the scope of the document. True, the initialization of a BFD session follows the same steps as defined in RFC 5884. But that all changes:
  Once the BFD session is in Up state the ingress LER
   that supports this specification MUST switch to the Demand mode by
   setting Demand (D) bit in its Control packet and initiating a Poll
   Sequence.  If the egress LER supports this specification it MUST
   respond with the Final (F) bit set in its BFD Control packet sent to
   the ingress LER and ceases further transmission of periodic BFD
   control packets to the ingress LER.
I haven't viewed these steps as "change to procedure" of RFC 5884 as they lead to the state that is outside the scope for RFC 5884.

> > A reminder that we don't vote.  C.f. RFC 7282, section 6..
> >
> GIM4>> I'm confused to differentiate when you raise the objection as the
> individual contributor and evaluate them and the consensus as the WG chair.

Chairs are not prohibited from offering technical feedback.  If you remain
confused on this issue, I suggest you discuss this with Martin at the
upcoming IETF.

-- Jeff