Skip to content

moq-transport: a late joiner gets no catalog through a draft-18 relay, since MAX_CACHE_DURATION 0 (#7729) #8465

Closed

Description

@t0ms

Since #7729 the catalog track advertises a zero max age, which on the moq-transport wire is MAX_CACHE_DURATION 0. #7729 means the zero to bound superseded groups only ("a late joiner still reads the newest snapshot"), but draft 18 §12.3 applies it to every object, the newest included:

If present, the relay MUST NOT start forwarding any individual Object received through this subscription or fetch after the specified number of milliseconds has elapsed since the beginning of the Object was received. … Once Objects have expired from cache, their state becomes unknown, and a relay that handles a downstream request that includes those Objects re-requests them.

So a conformant relay serves a late SUBSCRIBE no catalog object, and only a downstream FETCH would make it re-request one. At cd560bde5e the subscriber sends no joining FETCH for the catalog, and waits for a next catalog group that import ts never writes.

Reproduced on loopback (macOS arm64; moqx v0.3.7's release binary, threads: 1, use_relay_thread, use_local_forwarders; a 140 s synthetic TS from moq-interop-runner's make-fixture.sh through tsp -P regulate into moq import ts; moq export ts --delay 1s, draft 18, cd560bde5e throughout):

Relay Joiner after the first subscriber Joiner's output in 10 s
moqx v0.3.7 +5 s 0 bytes
moqx v0.3.7 +90 s 0 bytes
moq-relay cd560bde5e +90 s 2.2 MB

Each failing joiner logs SUBSCRIBE_OK for catalog.json (largest group 5 or 6, max_cache_duration: Some(0ns)), then nothing until it ends; no FETCH.

Across the internet, through two moqx v0.3.7 deployments, with a live 25 Mb/s TS: joiners at +60, +120 and +180 s exported 0 bytes on both. Crossing the builds at +90 s, a f2fca6fd2 publisher (no MAX_CACHE_DURATION on its catalog) with cd560bde5e subscribers delivered 25.5 MB in 10 s, and a cd560bde5e publisher with f2fca6fd2 subscribers 29.8 MB. There, unlike on loopback, joiners within about a minute of the first subscriber received the catalog (eight of eight).

Expected: a late joiner reads the newest catalog through any draft-18 relay. Either the moq-transport encoding omits MAX_CACHE_DURATION for a track whose max age bounds superseded groups only, or the subscriber FETCHes the largest group when its publisher declared a zero duration.

The commit before #7729, d66c892833, at both ends: a joiner at +180 s through the same deployment delivered 25.5 MB in 10 s.

(Written by Claude Opus 5.5)

Activity

  1. coderabbitai commented on Nov 10, 2026

    @coderabbitai
    Contributor
    🔗 Related PRs

    #7089 - feat(js): join moq-transport tracks at the current group [merged]


    📝 Issue Planner

    Check the box below or use the @coderabbitai plan command to generate an implementation plan and prompts that you can use with your favorite coding assistant.

    • Create Plan

    🧪 Issue enrichment is currently in open beta.

    You can configure auto-planning by selecting labels in the issue_enrichment configuration.

    To disable automatic issue enrichment, add the following to your .coderabbit.yaml:

    issue_enrichment:
      auto_enrich:
        enabled: false

    💬 Have feedback or questions? Drop into our discord!

  2. added
    questBeing tracked/planned in a quest. See `quest/`
    on Nov 10, 2026
  3. kixelated commented on Nov 10, 2026

    @kixelated
    Collaborator

    Decision: moq-transport stops sending our max age as MAX_CACHE_DURATION, in both Rust and JS. A received one still decodes. Our max age expires only superseded groups, while draft 18 §12.3 expires every object, the newest included, so no value can express it. Omitting the property fixes late joiners behind third-party relays too, unlike a joining FETCH on our side. A publisher's sealed MAX_CACHE_DURATION is still forwarded, now as advisory. This is m0, because #7729 isn't released yet and the next cut would ship the regression.

    Tracked by quest/m0/ietf-max-age.md, planned in #8467. Anyone who wants to donate tokens can run /quest-start quest/m0/ietf-max-age to take it on.

    (Written by Claude Opus 5.5 with high effort)

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    questBeing tracked/planned in a quest. See `quest/`

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions