Docker Compose Deployment

Deploy DB-GPT with MySQL using Docker Compose — a production-ready setup with persistent storage.

Prerequisites

• Docker and Docker Compose installed
• An API key from a supported provider

Quick start

The root docker-compose.yml deploys DB-GPT with a MySQL database and SiliconFlow as the default LLM provider.

Step 1 — Set your API key

SiliconFlow

export SILICONFLOW_API_KEY="your-siliconflow-api-key"

Get your key at SiliconFlow.

AI/ML API

export AIMLAPI_API_KEY="your-aiml-api-key"

Get your key at AI/ML API.

Step 2 — Start the services

SiliconFlow

SILICONFLOW_API_KEY=${SILICONFLOW_API_KEY} docker compose up -d

AI/ML API

AIMLAPI_API_KEY=${AIMLAPI_API_KEY} docker compose up -d

You should see output like:

[+] Running 3/3
 ✔ Network dbgptnet              Created   0.0s
 ✔ Container db-gpt-db-1         Started   0.2s
 ✔ Container db-gpt-webserver-1  Started   0.2s

Step 3 — Open the Web UI

Visit http://localhost:5670 in your browser.

First start may take a moment

The webserver waits for MySQL to finish initializing. If it fails on first start, it will auto-restart. Check logs with docker logs db-gpt-webserver-1 -f.

What gets deployed

The default docker-compose.yml creates:

Service	Image	Port	Purpose
db	mysql/mysql-server	3306	MySQL database for metadata
webserver	eosphorosai/dbgpt-openai:latest	5670	DB-GPT application server

Common operations

View logs

# Webserver logs
docker logs db-gpt-webserver-1 -f

# Database logs
docker logs db-gpt-db-1 -f

Stop services

docker compose down

Restart services

docker compose restart

Reset everything (including data)

docker compose down -v

warning

The -v flag removes all volumes, including the MySQL database. All data will be lost.

Customization

Use a different config file

Mount your own TOML config and override the startup command:

webserver:
  image: eosphorosai/dbgpt-openai:latest
  command: dbgpt start webserver --config /app/configs/your-config.toml
  volumes:
    - ./your-config.toml:/app/configs/your-config.toml

Mount local models

For GPU deployments, mount a local models directory:

webserver:
  volumes:
    - /data/models:/app/models
  deploy:
    resources:
      reservations:
        devices:
          - driver: nvidia
            capabilities: [gpu]

Other Compose examples

DB-GPT ships with additional Compose files for specific scenarios:

File	Use case
docker-compose.yml	Default proxy deployment with MySQL
docker/compose_examples/cluster-docker-compose.yml	Multi-worker cluster with GPU
docker/compose_examples/ha-cluster-docker-compose.yml	High-availability cluster
docker/compose_examples/dbgpt-oceanbase-docker-compose.yml	OceanBase database backend

Example:

docker compose -f docker/compose_examples/cluster-docker-compose.yml up -d

Next steps

Topic	Link
Cluster deployment	Cluster
Docker (single container)	Docker
Source code deployment	Source Code