Skip to content

The guest's log is unreachable, so half of every two-machine bug is invisible #8

Description

@matevisky

From: mwk-ghost-pointer@main

Two client tests (3 Sep, 13 Sep) reported the same thing — dropped, unreliable — and both times the host's log showed a healthy session. That is not a contradiction: the failing half is the guest, and the guest's log lives on the guest's machine with nothing to bring it back.

v0.7.0 closes the two gaps that were on the host side (a pong watchdog for half-open sockets, and peer join/leave logging). Neither helps if the side that broke is the one we cannot read.

Today the only route is asking the guest to press Copy diagnostics and paste it, before they quit the app. That works exactly once per cooperative guest and not at all in front of a client.

Worth considering, none decided:

  • The guest's control window already has the log. It could offer a one-click send to the host over the relay — the socket is right there, and the host is the person who wants it.
  • Or the relay keeps a short ring buffer of connection events per room, which would at least answer did they join, when did they drop without touching either machine's disk.

The second changes the relay's privacy story (it currently stores nothing), so it is a spec question and belongs to the relay side, not the app. Flagging rather than deciding.

Blocks: diagnosing the residual lag from 3 Sep, which still has no identified root cause.

Activity

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

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions