Recover from a failed update

หมายเหตุ: บทความนี้ยังเป็นภาษาอังกฤษ เนื่องจากคำแปลภาษาไทยกำลังอยู่ระหว่างจัดทำ

Purpose

Return the Docker runtime to a known-good state while preserving evidence and verified database backups.

Audience

Framework administrators and Docker operators.

Prerequisites

Access to the Compose project, app/updater logs, the previous image tag, the pre-update .emubackup, and the separately stored secret key.

Procedure

1. Record the update job error, target version, and backup path from System Maintenance or /data/update-status.json.

2. Preserve app and updater logs before recreating a container:

   docker compose logs --no-color app updater > emuframework-update.log

3. Check disk space, GHCR access, Docker socket access, updater token equality, container name, network membership, and the /data mount.

4. Verify whether the updater already restored the previous container and whether it is healthy.

5. If needed, set EMU_VERSION to the previous known-good tag, then run:

   docker compose pull
   docker compose up -d --force-recreate
   docker compose ps

6. Verify both SQLite databases and application data before retrying the update.

7. Restore the .emubackup only when data verification fails or the release notes explicitly require database rollback. Restore .emu-secret.key separately when needed.

8. Verify login, Apps, recent records, Designer customizations, SMTP, Scripts/Functions, and System Maintenance diagnostics.

Rolling back the container image does not undo database migrations. Never run the old and new images against the same volume simultaneously, never delete -wal or -shm files while a writer is active, and never use docker compose down -v during recovery.

Related topics

Framework update · Restore · Docker operations