Skip to content

Self-hosted platform: no command clears the local install record after its database is gone #428

Open

Description

@Brooooooklyn

Problem

If the platform PostgreSQL database is deleted or can no longer be reached, the local
installation record stays, and no Void command can clear it. The installation name stays
blocked on this machine.

database deleted / unreachable
        │
        ▼
void platform uninstall <id>  ──▶ needs the database lifecycle lock first ──▶ fails
void platform install --resume ──▶ reads the saved checkpoint from the database ──▶ fails
void platform install (same name) ──▶ "Installation <id> already uses this name.
                                       Resume it with void platform install --resume" ──▶ loop

A user can still pick a new name, so this is a cleanup dead end, not a full block. The Cloudflare
resources of that installation also stay, and uninstall cannot list them.

Repro (code reading; base 23ddc6e)

  1. Install a platform (or stop an install after the database claim).
  2. Delete the PostgreSQL database at the provider.
  3. Run void platform uninstall <id>: it stops at the database lock.
  4. Run void platform install --name <same-name>: refused, points to --resume.
  5. Run void platform install --resume --name <id>: it stops at the database read.

Cause

  • Every lifecycle command, uninstall included, takes the database lifecycle lock before any change:
    packages/void/src/cli/platform-management.ts:591-606 (acquirePlatformDatabaseLifecycleLock).
    uninstall --plan reads project Zero Trust state and projects from the database:
    platform-management.ts:311-323.
  • A non-draft resume reads the authoritative manifest from the database first:
    packages/void/src/cli/platform-lifecycle.ts:476-486 (readPlatformDatabaseLifecycleManifest).
  • The name check blocks any record that is not uninstalled:
    packages/void/src/cli/platform-registry.ts:256-268.
  • docs/guide/platform/installation/uninstall.md:92 says uninstall stops when the database cannot
    be reached, with no next step. Deleting ~/.void/platforms/<account>/<id>.json is only mentioned
    for after a finished uninstall (uninstall.md:38).

Options

  1. Docs only: say that when the database is gone for good, the user can delete
    ~/.void/platforms/<account>/<id>.json (and the .recovery.enc / .secrets.enc next to it),
    then remove the Cloudflare resources by hand (the list in "Remove retained resources").
    Smallest; no code risk.
  2. A void platform forget <id> command that removes only the local record after a typed
    confirmation, prints the recorded Cloudflare resources first, and refuses while the database
    is reachable. Needs CLI help, docs and interactive hints.
  3. Let uninstall --purge-resources run without the database when the user passes an explicit
    flag (for example --database-gone). Largest: the database lock and the Zero Trust guard exist
    to stop races with the API, so this needs a careful design of what is still safe to delete.

Recommendation: option 1 now; option 2 if users hit it more than once.

🤖 Generated with Claude Code

https://claude.ai/code/session_0123r5HnanmW2VUrZ7VWDufH

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

    bugSomething isn't working

    Type

    No type

    Fields

    Priority

    None yet

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions