Skip to content

export ts --follow: a reconnect of the subscriber's own session makes the same publisher instance look like a replacement #8384

Closed as not planned

Description

@t0ms

Measured on main at bad28cefd5; nothing in the code involved has changed up to 6eff24e91b.

Repro: a publisher that never restarts; moq … export ts --delay 2s --follow 10s with --quic-idle-timeout 10s; SIGSTOP the export for 30 s, then SIGCONT.

Observed: the export reconnects (the relay logs a new session), logs broadcast ended, waiting for it to return … follow=10s, then exits 1 with another publisher instance replaced broadcast …; enable stitch to follow it as a program switch. With --stitch it returns to the live edge 2.8 s after the SIGCONT, as a flagged program switch. Without --follow, one of 256 frozen exports also ended on the replacement error. The follower compares broadcast.info().instance with self.export.instance() (follower.rs), so the instance the new session sees for the same broadcast differs from the one recorded before.

Expected: the same publisher instance, seen through a new session of the subscriber's own, is a return, not a replacement.

(Written by Claude Opus 5.5)

Activity

  1. coderabbitai commented on Nov 9, 2026

    @coderabbitai
    Contributor
    🔗 Related PRs

    #6656 - feat(mux)!: TS export follows by cause, and --linger is --follow [merged]
    #6919 - fix(mux): retain old program during stitched epoch acquisition [merged]
    #6526 - fix(cli)!: follow shared publisher instances [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

    Closing as won't fix, decided by the maintainer during moq-dev/moq#8415.

    The exit is the safe behavior. On moq-lite-06 and moq-transport, the wire carries no publisher instance, so each subscriber session mints its own random one. A minted instance can't show whether the publisher changed while the subscriber was disconnected, so the follower refuses to treat it as a return. The same applies to a publisher returning within one lite-06 session.

    • To accept the risk, use --stitch: the return becomes a flagged program switch.
    • moq-lite-07 carries the publisher's instance end to end, so once it is negotiated by default, a reconnect of your own session sees the same instance and --follow treats it as a return.

    (Written by Opus 5.5 with high)

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