[Pce] 回复: Re: WG Last Call for draft-ietf-pce-flexible-grid

[Zhenghaomian <[e-mail]>]

Hi Andrew,

Thanks for taking the time to review and sending us your comments. We missed this one in the previous update and just addressed them as follow:

1. On the S Flag text, you are right that the original reference to RFC 5440 was too vague. Bidirectionality is indicated by the B (Bi-directional) flag in the RP Object, so the text now points explicitly to Section 7.4.1 of RFC 5440. For clarity, the update would be:

   S (Symmetry, 1 bit): This flag is only meaningful when the request
   is for a bidirectional LSP, i.e., when the B (Bi-directional) flag
   is set in the RP Object (see Section 7.4.1 of [RFC5440]).

2. On the error handling point, we agree that "insufficient memory" is not RSA-specific and overlaps with the existing PCEP overload mechanism in RFC 5440 Section 7.15. Let’s remove the RSA-specific error case, keep "Error-Type TBD7" scoped to RSA-specific conditions, and clarify that generic resource exhaustion at the PCE is handled via the overload mechanism rather than this "Error-Type".

These changes are reflected in the new (-14) version which was just posted.

Best wishes,
Haomian

发件人: Andrew Stone (Nokia) <andrew.stone=[e-mail]>
发送时间: 2025年5月8日 3:27
收件人: [e-mail]; [e-mail]
主题: [Pce] Re: WG Last Call for draft-ietf-pce-flexible-grid

Hi authors, PCE WG

I did a read through of the document – keep in mind I'm absolutely not an optical person and have very limited knowledge in optics therefore much of the content and use case flew over my head so I can't judge on its accuracy/correctness/usefulness 😊  With that said, the overall document structure is well written and the PCEP protocol extension encoding itself and some expectations on PCE are clear.

Two questions/comments:

Section 4.1 mentions S Flag for Symmetry " This flag is only meaningful when the request is for a bidirectional LSP " and references RFC5440 - but RFC5440 doesn't actually use the terms symmetry/symmetrical, or bidirectional. Therefore, I assume this is trying to reference Synchronized Path Requests / SVEC. If so, it might be worth explicitly mentioning see section 7.13 in RFC5440?

Regarding Error-Type TBD7:1 -> insufficient memory - is this related to the memory of the PCE itself?  This feels like more of a generic PCEP kind of error rather than anything to do with flexi grid-based computations? I guess I'm wondering what value this is to RSA requests, and, also whether a PCE would implement or be capable of responding to PCReq  with errors if it has insufficient memory. At that point it should go into overload state instead which implicitly signals to the PCC to stop sending PcReq?  i.e is this useful? and if so, should it be scoped within RSA error codes? Or should the implementation just go to overload and drop incoming PcReq which achieves the same?

Thanks
Andrew

From: [e-mail]<mailto:[e-mail]> <[e-mail]<mailto:[e-mail]>>
Date: Wednesday, April 23, 2025 at 5:02 AM
To: [e-mail]<mailto:[e-mail]> <[e-mail]<mailto:[e-mail]>>
Subject: [Pce] WG Last Call for draft-ietf-pce-flexible-grid
Hi all,

This email starts a 2-week WG last call for
draft-ietf-pce-flexible-grid-11 [1]. Please review and share your
comments using the PCE mailing list. Even a short feedback will be
appreciated.

This WGLC will end on Wednesday 7th May 2025.

Thanks,

Dhruv & Julien

--
[1] https://datatracker.ietf.org/doc/draft-ietf-pce-flexible-grid/