Thursday, 27 July 2023, 13:00-15:00 UTC-7
Room: Continental 5
Chairs: Bob Hinden, Jen Linkova, Ole Trøan
Minute taker: Greg Mirsky
Jabber Scribe: TBD
Jabber Room: 6man@jabber.ietf.org
Meetecho: https://meetings.conf.meetecho.com/ietf117/?session=30486
Thursday, 27 July 2023, 13:00-15:00 PDT
Participants: 110
Agenda: dense and interesting drafts, WG and individual drafts.
Nothing published since IETF-117. One draft submitted to IESG. Two drafts are in WG LC. A good interactions with the SPRING WG, and now can move forward with SRv6 compressed SIDs.
Drafts from other WGs should, preferably, reach certain level of maturity and consensus of the anchoring WG before being brought to 6man for a discussion and comments.
Bob Hinden presenting on behalf of Brian Carpenter.
Erik Kline, AD giving a summary of IESG. All issues addressed. Will try to close discusses and continue to engage with W3C et al. Unclear what the outcome will be.
Comments from Joel Jaeggli supporting pushing this forward.
David Lamparter supported the work on addressing the IESG comments and discusses.
Erik Kline: Clarifying that there is no ABNF for interface names. (Notable line: "poop-emoji 0 is an interface name")
The goal is provide a clear definition and concise architecture, separating a subnet from a broadcast domain construct. EVPN is an example of NBMA media. Another example is WiFi that may be presented as NBMA with broadcast emulation. Pascal Thubert presented updates since IETF-116. An open invitation to a continued discussion of the terminology; is the document grew to be too large; other ideas/proposals. David Lamparter asked about the multicast in NBMA. Pascal outlined mechanisms described in the draft. Lorenzo Colitti thinks that the document is too complex and too long, and questioned why ND is excluded. Lorenzo pointed that if ND is unreliable then perhaps that should be addressed and fixed. That might help to simplify the solution proposed for NBMA. Erik Nordmark has noted that several terms are still under-defined. Also, it appears that the proposed architecture replaces the existing architecture. Perhaps there a space and opportunity to clarify the scope of the problem being addressed. Erik Kline
Xiao Min presented the updates since IETF-116. Among those, aligned with the hbh-option draft. David Lampar pointed to a known security issue when a reply message is significantly larger then the query. David suggested to add that to the Security Considerations. Jen suggested on the proper track. It appears like Standard is appropriate for the draft. Let's take it to the list.
Gorry Fairhurst presented updates since 06 through 08 versions addressing WG LC comments. Tom Herbert asked about the conditions related to a HbH option that may result in a packet being dropped. The recommendation is to forward a packet with the unprocessed HbH option, unless the security requires otherwise. Ron Bonica brought up a question of handling unsupported, un-processable HbH options. Suresh's question will go to the List discussion too. Tom Herbert noted that it appears that none of HbH options be processed by any transit nodes. Lorenzo Colitti suggested that a node processes options that it is capable, understands before forwarding. Also, first apply security filter, then process extension headers. Mikael Abrahamson asked about the intended audience of the document. Timothy Winters will review and comment on the list. Jen, summarizing the discussion, likely another WG LC after the authors and reviewers converge on the issues brought in the meeting.
The draft reflects on the operational experience while transitioning from IPv4 to IPv6 and a period of running dual-stack environment. Most of the lessons learned are underemphasized need for configurable controls. Also, reflected on the on-list discussion (~300 messages that prompted the draft). Lorenzo Colitti wondered about the example analyzed. Also, supported progressing the draft, hopefully, will be reflected in Linux. Jen, no Chair hat, pointed to the normative language used in the draft. Kyle Rose supported the proposed changes to the default policy table. Timothy Winters shared his experience, making MUST is a challenge, perhaps SHOULD is practical. Ted Lemon encouraged moving fast with incremental updates, new drafts. Dave Thaler in support of 'yes' Q1 and 'no' to Q2 and Q3 (see the Discussion slide). Erik Kline suggested request IoT review. Continue the discussion on the list.
Lorenzo Colitti refreshed the problem in the scope of the draft. IPv4 device usually has multiple addresses. A good progress since IETF-116. A related DHCP PD per-device draft in WG LC at v6ops. Requested WG adoption. Erik Nordmark supported the work. Wonder about the dynamic of the process. Lorenzo noted that it is for a failing nodes. Erik suggested clarification of the failure conditions. Erik Vincke supported the draft. A=1 - old devices (Jen clarified). Pascal Thubert supported the draft.
Discussion of how the mechanism would work relating to the A flag in PIO and why the flag is proposed in PIO instead of in RA.
No action. Continue discussion on list.
Describing a mechanism that is similar to ICMP echo, that collects information along the path and returns it back to the originator of the message. Where the loopback reply contains the information learnt on the forward path / in the request. This behaves similarly to how ICMP error messages work.
Authors believe the document is ready for WG adoption.
Greg suggests to use IOAM loopback flag, something that would make it more general.
Discussion ensues.
Greg: Summary, I don't see the benefit of this proposal.
Chairs adjourned the meeting, the 120 minute meeting slot was used. The talks in the "Time Allows" section should be discussed on the mailing list.