railway-pgduckdb-pgvectorscale

A Docker container combining pg_duckdb and pgvectorscale extensions for PostgreSQL 17 on Railway.

Features

This container includes:

• PostgreSQL 17 - Latest major version
• pg_duckdb (v1.0.0) - DuckDB integration for analytical queries
• pgvector (v0.8.1) - Vector similarity search
• pgvectorscale (v0.8.0) - Scalable vector search with DiskANN

Quick Start

Deploy on Railway

Local Docker Build

docker build -t pgduckdb-pgvectorscale .
docker run -d \
  -p 5432:5432 \
  -e POSTGRES_PASSWORD=yourpassword \
  pgduckdb-pgvectorscale

Environment Variables

PostgreSQL Variables

• POSTGRES_PASSWORD - Database password (required)
• POSTGRES_USER - Database user (default: postgres)
• POSTGRES_DB - Database name (default: postgres)

S3 Configuration (Optional)

Configure S3 access for DuckDB:

• S3_ACCESS_KEY_ID - AWS access key ID
• S3_SECRET_ACCESS_KEY - AWS secret access key
• S3_REGION - AWS region
• S3_SESSION_TOKEN - AWS session token (optional)
• S3_ENDPOINT - Custom S3 endpoint (optional)
• S3_URL_STYLE - S3 URL style: 'path' or 'vhost' (default: path)

DuckDB Extensions (Optional)

• DUCKDB_EXTENSIONS - Comma-separated list of DuckDB extensions to install

Example:

DUCKDB_EXTENSIONS=httpfs,parquet,json

Usage Examples

Using pg_duckdb

-- Create a table and query with DuckDB
CREATE TABLE events (
    id SERIAL PRIMARY KEY,
    event_time TIMESTAMP,
    user_id INTEGER,
    event_type TEXT
);

-- DuckDB will automatically handle analytical queries
SELECT
    DATE_TRUNC('day', event_time) as day,
    COUNT(*) as event_count
FROM events
GROUP BY day
ORDER BY day;

-- Query S3 data directly (if S3 is configured)
SELECT * FROM read_parquet('s3://bucket/data.parquet');

Using pgvectorscale

-- Create a table with vector column
CREATE TABLE documents (
    id SERIAL PRIMARY KEY,
    content TEXT,
    embedding VECTOR(1536)
);

-- Create a DiskANN index for scalable vector search
CREATE INDEX ON documents USING diskann (embedding);

-- Perform similarity search
SELECT id, content
FROM documents
ORDER BY embedding <-> '[0.1, 0.2, ...]'::vector
LIMIT 10;

Documentation

• pg_duckdb Docs
• pgvector Docs
• pgvectorscale Docs
• Railway Docs

License

MIT License - see LICENSE file for details.

Contributing

Contributions welcome! Please open an issue or submit a PR.