Database

The Flow-Like API uses a relational database for platform data. In this repo, the schema is expressed as Prisma schema files under:

• packages/api/prisma/schema/

CockroachDB vs PostgreSQL

The Prisma schemas declare:

• provider = "cockroachdb"

For PostgreSQL environments (local dev, many Kubernetes clusters), the repo includes a helper script:

• packages/api/scripts/make-postgres-prisma-mirror.sh

It copies the schema tree to packages/api/prisma-postgres-mirror/schema and rewrites provider to postgresql.

Applying the schema

Local development (recommended)

The Kubernetes backend ships a docker-compose.yml that starts Postgres/Redis and applies the schema.

cd apps/backend/kubernetes
cp .env.example .env
# edit .env

docker compose up -d

The db-migrate service uses the tooling in packages/api to run:

• mirror schema to PostgreSQL
• prisma db push against DATABASE_URL

Manual migration (host tools)

cd apps/backend/kubernetes
./scripts/migrate-db.sh

Manual migration (Docker)

cd apps/backend/kubernetes
./scripts/migrate-db.sh --docker

Kubernetes / Helm

The Helm chart contains a migration Job template:

• apps/backend/kubernetes/helm/templates/db-migration-job.yaml

Important:

• Today it is a hook job intended to run pre-install / pre-upgrade.
• You still need to ensure the job image has access to the Prisma schema and tooling. If you don’t ship a purpose-built migration image, you can run migrations out-of-band (CI/CD step or an admin job).

Runtime configuration

The API runtime expects:

• DATABASE_URL (required)

This is read by flow_like_api::state::State.