DB Operations (Manual)
Scripts to run database initialization and migrations manually during your release process. Default wait timeout is 30s with 2s interval.
Required env (typical):
- DB_HOST, DB_PORT, DB_NAME, DB_USER, DB_PASS
- PROJECT_DIR (defaults /app)
- FLASK_APP (defaults to ${PROJECT_DIR}/scripts/run.py)
- COMPONENT_NAME (defaults to eveai_app)
- PYTHONPATH (defaults to ${PROJECT_DIR})
Order (standard: DB already exists):
- 00-env-check.sh
- 01-wait-for-db.sh (WAIT_FOR_DB_TIMEOUT=30 WAIT_FOR_DB_INTERVAL=2)
- 02-db-bootstrap-ext.sh (creates pgvector extension if missing)
- 03-db-migrate-public.sh
- 04-db-migrate-tenant.sh (single run for all tenants)
- 06-verify-minimal.sh
Optional, one-off for a new environment where DB does not exist yet:
- 90-initial-cluster-db-create.sh (run before step 3)
Optional seed/init data (one-off per env, idempotent):
- 05-seed-or-init-data.sh
Notes:
- Scripts are idempotent where applicable.
- Do not echo secrets. Password is read from DB_PASS via PGPASSWORD env.
- Ensure you run these inside the app image/container or any environment with psql, pg_isready, flask and the project code present.
Podman quickstart (manual testing)
Below are two simple ways to run these scripts in a Podman-based setup. Choose the approach that best matches your workflow.
Prerequisites:
- A reachable PostgreSQL 16 instance and credentials (DB_HOST, DB_PORT, DB_NAME, DB_USER, DB_PASS)
- Your application code available inside the container at /app (default PROJECT_DIR)
- Tools available in the container: psql, pg_isready, flask, python
- docker/Dockerfile.base already includes psql; use it to build a tools image if needed
Environment variables (typical):
- DB_HOST, DB_PORT, DB_NAME, DB_USER, DB_PASS
- PROJECT_DIR=/app (default)
- FLASK_APP=/app/scripts/run.py (default)
- COMPONENT_NAME=eveai_app (default)
- PYTHONPATH="/app" (default)
Order to run (standard):
- scripts/dbops/00-env-check.sh
- scripts/dbops/01-wait-for-db.sh (WAIT_FOR_DB_TIMEOUT=30 WAIT_FOR_DB_INTERVAL=2)
- scripts/dbops/02-db-bootstrap-ext.sh
- scripts/dbops/03-db-migrate-public.sh
- scripts/dbops/04-db-migrate-tenant.sh
- scripts/dbops/06-verify-minimal.sh
Optional (one-time for a new environment where the DB does not yet exist):
- scripts/dbops/90-initial-cluster-db-create.sh (run before step 3)
Optional seed/init (one-time per environment):
- scripts/dbops/05-seed-or-init-data.sh
Approach A — Exec into an existing app container
- Start your stack (e.g., via Podman Compose) so the app container and DB are running.
- Find the app container name: podman ps
- Exec into the container shell: podman exec -it bash
- Export DB variables inside the container (adjust values):
export DB_HOST=127.0.0.1
export DB_PORT=5432
export DB_NAME=eveai
export DB_USER=postgres
export DB_PASS=yourpassword
Optional (defaults exist)
export PROJECT_DIR=/app export FLASK_APP=/app/scripts/run.py export COMPONENT_NAME=eveai_app export PYTHONPATH="/app" - Run the scripts in order, for example:
scripts/dbops/00-env-check.sh &&
scripts/dbops/01-wait-for-db.sh &&
scripts/dbops/02-db-bootstrap-ext.sh &&
scripts/dbops/03-db-migrate-public.sh &&
scripts/dbops/04-db-migrate-tenant.sh &&
scripts/dbops/06-verify-minimal.sh - Optionally seed/init once per environment: scripts/dbops/05-seed-or-init-data.sh
Notes:
- If your DB runs in a separate container, DB_HOST should be that container's network hostname (often the service name from compose). If using host networking, 127.0.0.1 may work.
Approach B — Use a one-off tools container based on Dockerfile.base This runs the scripts without depending on a running app container. It mounts the repo at /app and uses the base image with psql installed.
-
Build the tools image (from repo root): podman build -f docker/Dockerfile.base -t eveai/tools:local .
-
Run a one-off container with your env vars and the repo mounted: podman run --rm -it
--name eveai-dbops
--network host
-e DB_HOST=127.0.0.1
-e DB_PORT=5432
-e DB_NAME=eveai
-e DB_USER=postgres
-e DB_PASS=yourpassword
-e PROJECT_DIR=/app
-e FLASK_APP=/app/scripts/run.py
-e COMPONENT_NAME=eveai_app
-e PYTHONPATH="/app"
-e WAIT_FOR_DB_TIMEOUT=30
-e WAIT_FOR_DB_INTERVAL=2
-v "$(pwd)":/app
-w /app
eveai/tools:local
bash -lc 'scripts/dbops/00-env-check.sh && scripts/dbops/01-wait-for-db.sh && scripts/dbops/02-db-bootstrap-ext.sh && scripts/dbops/03-db-migrate-public.sh && scripts/dbops/04-db-migrate-tenant.sh && scripts/dbops/06-verify-minimal.sh' -
Optional seed/init once per environment: podman run --rm -it --network host
-e DB_HOST -e DB_PORT -e DB_NAME -e DB_USER -e DB_PASS
-e PROJECT_DIR -e FLASK_APP -e COMPONENT_NAME -e PYTHONPATH
-v "$(pwd)":/app -w /app
eveai/tools:local
bash -lc 'scripts/dbops/05-seed-or-init-data.sh'
Notes:
- --network host simplifies connectivity on Linux. If you cannot use it, use a user-defined Podman network and set DB_HOST to the DB container hostname.
- Do not echo secrets; scripts read DB_PASS via the PGPASSWORD env variable automatically.
- If flask is missing at runtime, ensure your tools image has Python deps installed or run inside your actual app image/container that already has them.
Troubleshooting
- Connection refused: verify DB_HOST/PORT and network mode; try --network host.
- Authentication failed: check DB_USER/DB_PASS and that the user has rights on DB_NAME.
- pgvector missing even after step 3: ensure your DB user has rights to CREATE EXTENSION in DB_NAME.
- Flask command not found: run within the app image or install project dependencies in the tools image before running.