Runs, logs, schedules, and accounts are stored here. This is Piply's own state, not the databases your pipelines read from.
Relative paths resolve against the config folder. In Docker, point this at a mounted volume — a path inside the container is erased on redeploy.
The schema is created automatically. Run one Piply instance per database.
Piply connects before saving, so a wrong value is caught now rather than at the next restart. The choice is written to .env as PIPLY_DATABASE, and an admin can change it later under Settings.