Skip to content

Update readiness reads the host lock JSON record as evidence #437

Open

Description

@SashaMIT

update_controller.rs prove_ready (on develop, around line 2323) reads {data}/host-process.lock and requires its JSON pid, generation and role (gateway or gateway-public) to match the claimed Home generation. Any process that runs as the same user can rewrite that record, so it is not proof of who holds the lock.

Today this check is one of several: prove_ready also requires the private gateway coordinates, the health version, the private attach secret accepted by the listener, and the served Home bytes. A forged record therefore cannot make an unready Home pass; it can only make a ready Home fail.

Desired behavior: the update controller proves that the Home listener holds the gateway host lock from the flock, as /healthz does after #394 (host_lock::holds_gateway_host_lock, answered by the process that owns the socket). For example, require GET /healthz on the reported Home origin to return OK, and keep the generation binding in the private coordinates. The fixture in update_controller/tests.rs (around line 1513) then serves /healthz instead of writing the record.

Found while fixing the subordinate lane in #394, which removes the record from the other ownership decisions (runtime_control.rs subordinate lane, host_lock.rs active_host_process).

Activity

  1. added this to the After 0.8.0 milestone on Oct 21, 2026
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

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions