**Session Date/Time:** 27 Jun 2022 14:00

# [ROLL](../wg/roll.html)

## Summary
The ROLL Working Group held an interim meeting to review the status of active drafts, discuss proposed changes, and plan next steps. Key discussions included updates on `AODV-RPL` (draft-ietf-roll-aodv-rpl), `DAO Projection` (draft-ietf-roll-ri-rpl-rs), and `Root Node Failure Detector` (RNFD, draft-ietf-roll-rnfd). Significant progress was reported on `AODV-RPL` with a request to return to Working Group Last Call. `DAO Projection` is currently in Working Group Last Call and seeking further reviews. The new work item `RNFD` sparked a detailed technical discussion about its design principles and potential edge cases, particularly regarding the detection of a live root. The chairs encouraged continued asynchronous discussion on the mailing list and announced an upcoming interim meeting to further these topics.

## Key Discussion Points

*   **Working Group Status:**
    *   `6LoSA` is in progress with AAD evaluation, addressing open issues.
    *   `AODV-RPL` and `DAO Projection` were slated for discussion, with `DAO Projection` currently in WG Last Call.
    *   `Enrollment Priority` has open issues on GitHub and the mailing list needing resolution.
    *   `MOPEX` and `Capabilities` drafts have expired and require attention as future WG topics.
    *   `DAG-Knowledge` is targeted for WG adoption.
    *   `RNFD` (Fast Border Router Crash Detection) was presented as new work.
    *   `DODAG modifications` and `APR Multicast` drafts are inactive and require a decision on their future.
    *   Milestones for several drafts (DAO Projection, Enrollment Priority, MOPEX, Capabilities, RNFD, DODAG modifications, J4FBL, Multicast) extend into November, when the WG is scheduled for rechartering or closure.

*   **IETF 114 Planning:**
    *   A one-hour ROLL WG meeting is scheduled for Thursday.
    *   A joint meeting with MANET, BABEL, and ROLL is planned for Friday to discuss multicast handling across working groups and potential joint work.
    *   Several key participants (Pascal, Conrad) will not be able to attend IETF 114 in person or remotely due to conflicting schedules or funding.

*   **AODV-RPL (draft-ietf-roll-aodv-rpl) Update:**
    *   Presented by Charles Perkins. Draft-14 was recently submitted to address comments from earlier reviews.
    *   **Core Protocol:** AODV-RPL uses RPL options for Route Request (RREQ) and Route Reply (RREP) to create two DODAGs (rooted at originating and target nodes, respectively), supporting both asymmetric and symmetric routes.
    *   **Key Change: New Multicast Group:** A significant change involves requesting a new multicast group for AODV-RPL nodes. This is necessary because the base RPL specification forwards unrecognized options rather than dropping them, which would interfere with AODV-RPL's intended operation.
    *   **Instance IDs:** AODV-RPL uses local instance IDs, which are paired with the IP address of the root node for unique identification. The "Delta Field" mechanism was introduced to handle cases where a target node might re-use an instance ID.
    *   **Draft-14 Resolutions:** Clarifications were added regarding scenarios for AODV-RPL's use, hop-by-hop routes, DODAG rejoin mechanisms, source routing definitions, and the role of border routers (Figure 4). Guidance was included for selecting the 'L' field value and evaluating link asymmetry.
    *   The presenter believes all issues have been resolved, and the document is ready for Working Group Last Call.

*   **Projected DAO (Root Initiated Routing State - draft-ietf-roll-ri-rpl-rs) Update:**
    *   Presented by Pascal Thubert. This draft enables the root to control the establishment of Traffic Engineering (TE) paths within a RPL domain.
    *   **Key Concepts:**
        *   **Track:** A non-storing, dotted-line multipath DODAG, acting as a skeleton where segments (straight paths between dots) are installed using storing mode. This allows for richer topology information at the root.
        *   **VR Information Option:** Introduced as a multi-hop equivalent of the Transit Information Option for installing segments.
    *   **Reviews:** The document underwent significant reorganization for readability based on early reviews. Subsequent technical reviews led to clarifications, including the nature of a track as an overlay and the function of the VR Information Option.
    *   **Status:** The document is currently in Working Group Last Call.
    *   **Requested Feedback:** Reviews are welcomed, particularly regarding missing status codes or flows. Conrad offered to provide a review by late summer, which the chairs indicated would still be welcome even if the document proceeds to ASG submission before then.

*   **Root Node Failure Detector (RNFD - draft-ietf-roll-rnfd) Update:**
    *   Presented by Conrad Wawrzyniak. This draft proposes an algorithm for faster detection of Border Router (BR) crashes in LLNs.
    *   **Problem Statement:** BRs are critical; their failures lead to chaos in current RPL implementations due to slow, reactive link failure detection and inconsistent network knowledge.
    *   **Proposed Solution (RNFD):** An algorithm working in parallel to RPL. "Sentinels" (select root neighbors) monitor their links to the root and exchange this status via a new RNFD Option in link-local RPL messages. A consensus among sentinels determines if the root has crashed.
    *   **Benefits:** Measured latency reduction by an order of magnitude, and traffic reduction by a factor of four.
    *   **Status:** Adopted by the WG in March. The WG concluded the topic is important but the solution might require modification.
    *   **Discussion:**
        *   **Experimental Status:** Michael suggested publishing RNFD as an experimental RFC to reduce friction, given its novel approach.
        *   **Edge Cases & Robustness:** Pascal raised concerns about scenarios where first-hop nodes cannot communicate or where the algorithm might prematurely declare a root dead while it is still operational, leading to unintended DODAG collapse or reconstruction. Conrad clarified that RNFD is probabilistic, and in cases of significant link loss (e.g., 99%), a DODAG reconstruction is desirable. The root, if alive, would be aware of the situation and could issue a new DODAG version.

## Decisions and Action Items

*   **AODV-RPL:** The chairs will initiate a new Working Group Last Call for `draft-ietf-roll-aodv-rpl-14`, assuming Charlie Perkins resolves outstanding GitHub issues.
*   **DAO Projection:** Participants are strongly encouraged to review `draft-ietf-roll-ri-rpl-rs` and provide feedback during its current Working Group Last Call.
*   **RNFD:** Further technical discussion on the mailing list is needed to refine the `draft-ietf-roll-rnfd` design, particularly concerning edge cases and its interaction with core RPL mechanisms.
*   **Multicast:** The joint meeting with MANET and BABEL at IETF 114 will explore common interests and potential joint work on multicast handling.
*   **6Lo Multicast:** The `6Lo Multicast` document will undergo a joint Working Group Last Call with ROLL.

## Next Steps

*   **AODV-RPL:** Proceed with Working Group Last Call for draft-14.
*   **DAO Projection:** Continue to gather reviews during Working Group Last Call.
*   **RNFD:** Engage in detailed technical discussion on the mailing list to address the raised design concerns and determine the path forward (e.g., experimental RFC, further modifications).
*   **IETF 114:** Participate in the ROLL WG session and the joint multicast meeting, if able.
*   **Future Interim:** An interim meeting is planned for late August or early September to continue advancing technical discussions, particularly for RNFD and other drafts.