01
- Rescue
- A binary that will not start, a backup that will not restore, or a migration that keeps failing. We find the piece that is wrong and get it serving traffic again.
02
- Backup and restore drills
- Scheduled backups to S3 or similar, with restore rehearsed into a fresh binary so an outage is a replay rather than an experiment.
03
- Migration reconciliation
- Hand-built collections and generated migrations lined up, with a clean migration history you can replay in CI.
04
- Hosting hardening
- Reverse proxy, TLS, WebSocket timeouts, file-upload limits and process supervision set up so the single binary is not also the single point of failure for silly reasons.
05
- Move to Postgres
- When the SQLite ceiling becomes real, we plan the migration onto a Postgres-backed stack without losing auth records or stored files.
06
- Managed Tech
- Backups watched, version upgrades tested on a copy of pb_data, and an engineer who picks up when the admin UI will not load.