Projects & databases

Projects group related databases — most teams create one project per application or per environment. Databases are individual Postgres instances.

Creating a database

curl -X POST https://api.astrabase.app/api/v1/organizations/$ORG_ID/projects/$PROJECT_ID/databases \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "production",
    "region_code": "eu-west",
    "isolation_level": "shared",
    "cpu_limit": 1,
    "ram_limit_mb": 1024,
    "storage_limit_mb": 512
  }'

This returns 202 Accepted with a job_id — provisioning is asynchronous. List available regions with GET /regions.

Isolation levels

• shared — runs on a shared Postgres cluster. Faster to provision, lower cost, the default.
• dedicated — an isolated Postgres instance of its own.

Connecting

curl https://api.astrabase.app/api/v1/organizations/$ORG_ID/projects/$PROJECT_ID/databases/$DATABASE_ID/connection \
  -H "Authorization: Bearer $TOKEN"

Returns host, port, database name, credentials, SSL mode, and a ready-to-use connection_string.

Roles & credentials

Each database can have multiple Postgres roles beyond the default one, each with a scope:

• app — normal read/write access.
• readonly — SELECT only.
• admin — full privileges, including DDL.

curl -X POST https://api.astrabase.app/api/v1/organizations/$ORG_ID/projects/$PROJECT_ID/databases/$DATABASE_ID/roles \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"name": "analytics_ro", "scope": "readonly"}'

The password is returned once, at creation — it isn't stored anywhere retrievable, so save it immediately. Rotate a role's password with POST /roles/{role_id}/rotate.

Extensions

List, install, and remove Postgres extensions per database:

curl https://api.astrabase.app/api/v1/organizations/$ORG_ID/projects/$PROJECT_ID/databases/$DATABASE_ID/extensions -H "Authorization: Bearer $TOKEN"

Resizing, suspending, deleting

# Resize (cpu/ram/storage)
curl -X POST .../databases/$DATABASE_ID/resize -d '{"cpu_limit": 2, "ram_limit_mb": 2048, "storage_limit_mb": 20480}'

# Suspend / resume
curl -X POST .../databases/$DATABASE_ID/suspend
curl -X POST .../databases/$DATABASE_ID/resume

# Delete
curl -X DELETE .../databases/$DATABASE_ID

All state-changing operations (create, resize, delete, backup, restore) go through the same job system — check status at GET /jobs/{job_id}, or use an SDK's waitForJob/wait_for_job helper to block until it settles.