Markdown Version | Session Recording

Session Date/Time: 25 Jan 2024 17:00

NTP

Summary

The NTP Working Group met to discuss the status of several documents and ongoing efforts. Key updates included the readiness of "Updating Registries" and "Enterprise Profile" drafts for progression. The group reached consensus to move "NTP over PTP" forward to working group last call, with one new version incorporating feedback and a request for a new IANA registry. Discussion on "NTPv5 Requirements" highlighted the need for documented consensus to progress the working group's future work, despite ongoing debate on certain aspects like algorithmic separation. Strong arguments were made for a new NTPv5 to address fundamental issues in NTPv4. Updates were also provided on Roughtime and NTS for PTP, with plans for upcoming draft revisions and potential working group adoption. The chairs announced the next virtual interim meeting and sought feedback on an in-person meeting for Brisbane.

Key Discussion Points

• Administrative Notes:

  • Draft minutes are taken in the online notes tool.
  • The December meeting minutes were not uploaded as final and will be addressed.
  • Participants were reminded to use the queue tool, be concise and respectful, and adhere to the IETF Code of Conduct.
  • No agenda bashing occurred.

• Document Review Status:

  • Updating Registries: Confirmed by Rich that all requested changes have been incorporated and the document is ready to proceed.
  • Enterprise Profile: Ready to proceed.
  • Kronos: Currently with the RFC Editor.

• NTP over PTP Working Group Last Call Results (draft-ietf-ntp-ntp-over-ptp):

  • Marek uploaded a new version incorporating most changes suggested by Martin.
  • Specific Changes: The TLV now uses an organization-specific value with a sub-type, increasing its length by eight octets. This requires a new IANA registry for the use of the Ayana organization ID for PTP messages. A formal request for this registry has been added to the document.
  • Unaddressed Comments:
    • Description of endianness for PTP header fields (common practice is to only describe exceptions to network byte order). The group found it unnecessary to explicitly state Big-endian.
    • Wording in the introduction regarding "packet encapsulation" vs. "protocol encapsulated" (Olaf's comment). The chair noted agreement with Olaf's suggestion for clearer language.
  • A sense of those present indicated consensus to move forward with this document.

• NTPv5 Requirements (draft-ietf-ntp-ntpv5-requirements):

  • James incorporated feedback from Kristoff and Ira/Era into the GitHub repository but had not yet published a new version to the data tracker, awaiting further acknowledgment.
  • The chair expressed a desire to publish this document to gather ISG feedback and help the working group achieve documented consensus, acknowledging that some members (e.g., Rich) previously suggested parking it as a draft.
  • Rich noted that if the document helps drive documented consensus, it is valuable.
  • David highlighted the importance of documenting consensus to prevent repeated arguments and raised a concern about the proposal to split algorithms from the normative specification. He stated this is a contentious subject for reference implementation developers and needs explicit consensus to avoid future slowdowns.
  • The chair acknowledged the concern but noted the historical context where separating algorithms was an original, unformalized plan for NTPv4.

• Updates on Ongoing Working Group Efforts:

  • NTPv5 Updates (Marek): No updates; awaiting agreement on the requirements document. Marek noted some opposition to the current design, with some preferring an NTPv4 extension field for improvements to maintain full compatibility.
  • Discussion on NTPv5 vs. NTPv4 Extensions:
    • David articulated strong reasons for an NTPv5, citing "landmines" in NTPv4 (e.g., reliance on transmit timestamps as unique identifiers) that lead to easily exploitable implementations (e.g., systemd-timesyncd). He plans to demonstrate these issues soon and announced two experimental NTPv5 servers on the public internet.
    • Watson and Kristoff supported the need for a new NTPv5 that is not fully compatible with NTPv4 to address fundamental problems.
  • Roughtime (draft-ietf-ntp-roughtime): David reported that he has not yet updated the draft. The plan is to add an extension for clients to signal server keys, which is crucial for key rotation. He anticipates mainly text changes, with few wire format changes remaining, and aims for a working group last call after the next version is uploaded.
  • NTS for PTP (draft-ietf-ntp-nts-for-ptp): Martin was absent, but an update was provided. He is currently implementing and writing the next version of the draft, targeting an update in 2-3 weeks. He intends to then propose moving it from an individual draft to a working group document.

• Any Other Business / Way Forward:

  • The chairs intend to continue holding virtual interim calls to maintain momentum and stay on task.
  • The current meeting time (later in the day) was adjusted to accommodate some participants. Feedback on its suitability is requested, as attendance was slightly lower than usual.
  • Future Meetings: A virtual interim is planned for February 22nd at the same time.
  • Brisbane IETF Meeting (in-person): The chairs will poll the mailing list about scheduling an in-person meeting in Brisbane. Concerns were raised about low in-person attendance potentially making it more efficient to conduct it as a virtual meeting. Karen and Rich expressed intent to attend in person; David is willing to participate virtually despite an inconvenient time zone.

Decisions and Action Items

• NTP over PTP (draft-ietf-ntp-ntp-over-ptp):
  • Decision: The working group reached consensus to move the document forward.
  • Action Item: Chairs (Karen and Deor) will send an email to the working group stating consensus.
  • Action Item: The document author (Marek) will ensure the request for a new IANA registry is formally correct.
• NTPv5 Requirements (draft-ietf-ntp-ntpv5-requirements):
  • Decision: The working group will continue to progress this document.
  • Action Item: James will immediately publish the updated version from GitHub to the data tracker.
  • Action Item: Chairs will review the new version and send a note to the mailing list, aiming for clarity on addressed vs. unaddressed discussion points before a definitive consensus call.
• Ongoing NTPv5 Efforts (Marek's draft):
  • Decision: Given the strong arguments for a new NTPv5, the working group will continue to pursue this direction based on the requirements document.
• Meeting Schedule:
  • Action Item: Chairs will send out an announcement for the next virtual interim meeting on February 22nd.
  • Action Item: Chairs will poll the mailing list to gauge interest and feasibility for an in-person meeting at IETF 119 in Brisbane.

Next Steps

• February 22nd: Next virtual interim meeting at the same time as this meeting.
• Upcoming Draft Revisions:
  • Roughtime: Next version to include client key signaling extension, aiming for WG Last Call thereafter.
  • NTS for PTP: Next version expected in 2-3 weeks, with a plan to propose WG adoption.
• Mailing List Discussion: Feedback on the current meeting time and interest in an in-person Brisbane meeting.