isopen/pgisvector repository overview

⁠Description
This Docker Image⁠ extends the official PostGIS 17-3.5⁠ base with the pgvector⁠ extension compiled from source. It provides a ready-to-use PostgreSQL 17 environment featuring PostGIS for geospatial queries and pgvector for vector operations—suitable for production deployments that require both spatial and embedding-based functionality.

Key Features

1. PostgreSQL 17 + PostGIS 3.5
   • All standard PostGIS features (geography types, geometry functions, raster, etc.).
2. pgvector
   • Enables similarity searches on vector embeddings (e.g., for AI/ML use cases).
3. Multi-Stage Build
   • Keeps build tools and libraries out of the final image, reducing bloat.
4. Environment Variables
   • Configurable at runtime: POSTGRES_USER, POSTGRES_PASSWORD, and POSTGRES_DB.
5. Init Scripts
   • Any .sql or shell scripts placed in docker-entrypoint-initdb.d/ will run on first launch.
6. Health Check
   • A basic pg_isready call to ensure the container is accepting connections.

Usage

1. Pull and Run isopen/pgisvector:latest⁠

   docker pull isopen/pgisvector:latest
   docker run -d \
     --name my_pgvector_db \
     -e POSTGRES_PASSWORD=secret \
     -p 5432:5432 \
     isopen/pgisvector:latest
   

   Copy

   Replace secret with your preferred password.

2. Create Extensions
   Inside the running container (or via psql from elsewhere) to ensure everything is going well:

   CREATE EXTENSION IF NOT EXISTS postgis;
   CREATE EXTENSION IF NOT EXISTS vector;
   

   Copy

3. Check Health
   Docker automatically runs a health check (pg_isready) to verify the container is ready.

Example Docker Compose

version: '3.8'
services:
  pgvector_db:
    image: isopen/pgisvector:latest
    container_name: my_pgvector_db
    environment:
      - POSTGRES_PASSWORD=secret
    ports:
      - "5432:5432"
    healthcheck:
      test: ["CMD", "pg_isready", "-U", "postgres"]
      interval: 30s
      timeout: 5s
      retries: 3

When to Use

• Geospatial + Embeddings: If you need both advanced spatial queries (PostGIS) and vector similarity search (pgvector).
• AI/ML pipelines: Storing embeddings directly in PostgreSQL for similarity lookups.
• Analytics: Combining geospatial analytics with AI-driven queries.

Notes

• Default credentials (postgres/postgres) are for local development/testing only. Overwrite them in production.
• The final image is still based on postgis/postgis:17-3.5; a “slim” alternative may not exist at the time of writing.
• For advanced setups, you can extend this image with additional Postgres extensions or layer your own configuration files in /docker-entrypoint-initdb.d/.

License & Sources

• PostGIS/PostgreSQL License⁠
• pgvector License (MIT)⁠
• Docker Image⁠
• Github Repos⁠
• This is open