A fast, lightweight S3-compatible object storage server written in Go. Fully developed with AI assistance using Claude Code.

• High performance - Built with Go for speed and efficient concurrency
• S3-compatible - Works with AWS CLI, SDKs, and existing S3 tools
• Single binary - No external dependencies, easy deployment
• AI-friendly - Includes CLAUDE.md for seamless AI-assisted development
• Well tested - Comprehensive test coverage using AWS SDK for Go v2

See Supported S3 APIs for the full list of implemented operations.

Download the latest binary for your platform from GitHub Releases.

# Linux (amd64)
curl -LO https://github.com/kumasuke/jog/releases/latest/download/jog_linux_amd64.tar.gz
tar xzf jog_linux_amd64.tar.gz
sudo mv jog /usr/local/bin/

# macOS (Apple Silicon)
curl -LO https://github.com/kumasuke/jog/releases/latest/download/jog_darwin_arm64.tar.gz
tar xzf jog_darwin_arm64.tar.gz
sudo mv jog /usr/local/bin/

# macOS (Intel)
curl -LO https://github.com/kumasuke/jog/releases/latest/download/jog_darwin_amd64.tar.gz
tar xzf jog_darwin_amd64.tar.gz
sudo mv jog /usr/local/bin/

docker run -p 9000:9000 ghcr.io/kumasuke/jog:latest

With persistent storage and custom credentials:

docker run -p 9000:9000 \
  -v jog-data:/data \
  -e JOG_AUTH_ACCESS_KEY=myaccesskey \
  -e JOG_AUTH_SECRET_KEY=mysecretkey \
  ghcr.io/kumasuke/jog:latest

To enable automatic metadata backup via Litestream:

docker run -p 9000:9000 \
  -v jog-data:/data \
  -e LITESTREAM_REPLICA_URL=s3://my-backup-bucket/jog \
  -e AWS_ACCESS_KEY_ID=... \
  -e AWS_SECRET_ACCESS_KEY=... \
  ghcr.io/kumasuke/jog:latest

go install github.com/kumasuke/jog/cmd/jog@latest

make build

./bin/jog server

The server starts on port 9000 by default.

Environment variables:

• JOG_SERVER_PORT - Server port (default: 9000)
• JOG_SERVER_ADDRESS - Listen address (default: 0.0.0.0)
• JOG_STORAGE_DATA_DIR - Data directory (default: ./data)
• JOG_STORAGE_METADATA_DB - Metadata DB path (default: ./data/metadata.db)
• JOG_AUTH_ACCESS_KEY - Access key (default: minioadmin)
• JOG_AUTH_SECRET_KEY - Secret key (default: minioadmin)
• JOG_LOGGING_LEVEL - Log level (default: info)
• JOG_LOGGING_FORMAT - Log format: json or console (default: json)
• JOG_LIFECYCLE_ENABLED - Run the lifecycle engine (default: true)
• JOG_LIFECYCLE_INTERVAL - Lifecycle cycle interval, Go duration (default: 1h)
• JOG_LIFECYCLE_MAX_ACTIONS - Max actions per cycle (default: 10000)
• JOG_NOTIFICATION_REGION - awsRegion stamped into notification events (default: us-east-1)
• JOG_NOTIFICATION_DELIVERY_TIMEOUT - Per-webhook delivery timeout, Go duration (default: 10s)

Webhook delivery targets (ARN → URL) are configured only via notification.targets in config.yaml; lifecycle expiration events (s3:LifecycleExpiration:Delete / :DeleteMarkerCreated) are POSTed to the matching URL when at least one target is set.

JOG executes bucket lifecycle rules with a background engine (in-server ticker, first run ~1 minute after startup). You can also drive it manually:

# Preview what a cycle would do without changing anything
jog lifecycle run --dry-run

# Run one cycle (optionally scoped to a single bucket)
jog lifecycle run --bucket my-bucket

Versions under COMPLIANCE/GOVERNANCE retention or legal hold are never deleted (fail-closed). Set JOG_LIFECYCLE_ENABLED=false to disable the engine.

# Start with Docker Compose
docker compose up -d

# Check logs
docker compose logs -f

# Stop
docker compose down

export AWS_ENDPOINT_URL=http://localhost:9000
export AWS_ACCESS_KEY_ID=minioadmin
export AWS_SECRET_ACCESS_KEY=minioadmin

# Create bucket
aws s3 mb s3://my-bucket

# Upload file
aws s3 cp file.txt s3://my-bucket/

# List objects
aws s3 ls s3://my-bucket/

# Download file
aws s3 cp s3://my-bucket/file.txt ./downloaded.txt

Benchmark JOG against MinIO, rclone, and versitygw. See benchmark/README.md for details.

For a detailed comparison with other S3-compatible servers, see S3 Alternatives Comparison.

See CLAUDE.md for development guidelines.

MIT License