How to install Dify, Ollama, and Open WebUI in the same docker compose
You may encounter these errors when integrating Ollama with Dify. This may happen if you install Dify and Ollama in separate Docker Compose setups.
httpconnectionpool(host=127.0.0.1, port=11434): max retries exceeded with url:/cpi/chat (Caused by NewConnectionError(': fail to establish a new connection:[Errno 111] Connection refused'))To avoid this error, ollama, and dify can be in the same docker component, so dify can simply access to ollama with,
http://ollama:11434Install Dify
Clone Dify
cd
git clone https://github.com/langgenius/dify.git
cd dify/docker
cp .env.example .envConfirm to run containers
docker compose up -dWait a moment to see below output.
✔ Network docker_default Created 0.2s
✔ Network docker_ssrf_proxy_network Created 0.2s
✔ Container docker-redis-1 Started 6.3s
✔ Container docker-sandbox-1 Started 6.3s
✔ Container docker-ssrf_proxy-1 Started 6.4s
✔ Container docker-weaviate-1 Started 6.3s
✔ Container docker-web-1 Started 6.3s
✔ Container docker-db-1 Started 6.2s
✔ Container docker-worker-1 Started 2.7s
✔ Container docker-api-1 Started 2.7s
✔ Container docker-nginx-1 Started 3.1sAccess the Dify web interface with http://localhost
You should see this after creating an admin account and login
Shutdown dify
docker compose downInstall Ollama together with Dify docker compose
Open “~/dify/docker/docker-compose.yaml” and add these lines at services section.
services:
...
ollama:
volumes:
- ollama:/root/.ollama
container_name: ollama
pull_policy: always
tty: true
restart: unless-stopped
image: ollama/ollama:${OLLAMA_DOCKER_TAG-latest}
- 'OLLAMA_HOST=0.0.0.0:11434'
# GPU support
deploy:
resources:
reservations:
devices:
- driver: ${OLLAMA_GPU_DRIVER-nvidia}
count: ${OLLAMA_GPU_COUNT-1}
capabilities:
- gpu
open-webui:
build:
context: .
args:
OLLAMA_BASE_URL: '/ollama'
dockerfile: Dockerfile
image: ghcr.io/open-webui/open-webui:${WEBUI_DOCKER_TAG-main}
container_name: open-webui
volumes:
- open-webui:/app/backend/data
depends_on:
- ollama
ports:
- ${OPEN_WEBUI_PORT-3000}:8080
environment:
- 'OLLAMA_BASE_URL=http://ollama:11434'
- 'WEBUI_AUTH=False'
extra_hosts:
- host.docker.internal:host-gateway
restart: unless-stoppedadd these lines at vlumes section.
volumes:
...
ollama: {}
open-webui: {}Run again,
docker compose up -dWait a moment to see below output.
[+] Running 1/1
✔ ollama Pulled 5.1s
[+] Running 15/15
✔ Network docker_ssrf_proxy_network Created 0.2s
✔ Network docker_default Created 0.2s
✔ Volume "docker_open-webui" Created 0.0s
✔ Volume "docker_ollama" Created 0.0s
✔ Container docker-ssrf_proxy-1 Started 2.6s
✔ Container docker-sandbox-1 Started 1.5s
✔ Container docker-weaviate-1 Started 2.6s
✔ Container docker-redis-1 Started 2.6s
✔ Container ollama Started 1.9s
✔ Container docker-db-1 Started 1.7s
✔ Container docker-web-1 Started 2.6s
✔ Container open-webui Started 2.8s
✔ Container docker-api-1 Started 3.3s
✔ Container docker-worker-1 Started 3.6s
✔ Container docker-nginx-1 Started 3.7sAccess the open-webui, and download a model (i.e. llava). http://localhost:3000
Integrate Ollama as a Dify’s model provider
Settings > Model Providers > Ollama,
