Fix: Open WebUI Connection Refused, OpenAI Network Problem & Ollama Not Found

When opening Open WebUI, you encounter red toast notifications stating “Connection Refused”, “OpenAI Network Problem”, or the model selector dropdown shows “No models available” / “Ollama Not Found”.

---

Quick Fix

Inside your docker-compose.yml, change OLLAMA_BASE_URL from localhost or 127.0.0.1 to the Docker service name (http://ollama:11434), and set OLLAMA_HOST=0.0.0.0 on the Ollama service:

services:
  ollama:
    image: ollama/ollama:latest
    container_name: ollama
    environment:
      - OLLAMA_HOST=0.0.0.0

  open-webui:
    image: ghcr.io/open-webui/open-webui:main
    container_name: open-webui
    environment:
      - OLLAMA_BASE_URL=http://ollama:11434
      - OPENAI_API_BASE_URL=http://ollama:11434/v1

---

Symptoms & Common Error Messages

1. “Connection Refused” Notification: Open WebUI UI loads, but generating a chat response immediately fails.
2. “OpenAI Network Problem”: Appears when Open WebUI attempts to connect to an OpenAI-compatible endpoint (like vLLM or Ollama /v1) running on the same server.
3. Empty Model Selector: Dropdown says “No models available” despite models existing in /root/.ollama.
4. Container Log Trace:

   ConnectionError: HTTPConnectionPool(host='127.0.0.1', port=11434): Max retries exceeded with url: /api/tags (Caused by NewConnectionError('<urllib3.connection.HTTPConnection object>: Failed to establish a new connection: [Errno 111] Connection refused'))

---

Root Causes

1. localhost Ambiguity in Containers: localhost inside open-webui refers to its own container sandbox, not the host machine or adjacent containers.
2. Ollama Default Binding: By default, standalone Ollama binds to 127.0.0.1, which refuses incoming packets from the Docker bridge gateway. Setting OLLAMA_HOST=0.0.0.0 is required.
3. Network Isolation: The two containers were launched in separate Compose files without a shared network bridge.

---

Complete Verified docker-compose.yml

version: '3.8'

services:
  ollama:
    image: ollama/ollama:latest
    container_name: ollama
    restart: unless-stopped
    ports:
      - "11434:11434"
    volumes:
      - ollama_models:/root/.ollama
    environment:
      - OLLAMA_HOST=0.0.0.0
      - OLLAMA_KEEP_ALIVE=24h
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: all
              capabilities: [gpu]
    networks:
      - ai_net

  open-webui:
    image: ghcr.io/open-webui/open-webui:main
    container_name: open-webui
    restart: unless-stopped
    ports:
      - "3000:8080"
    environment:
      - OLLAMA_BASE_URL=http://ollama:11434
      - WEBUI_SECRET_KEY=generate_a_random_secret_string_here
    volumes:
      - openwebui_data:/app/backend/data
    depends_on:
      - ollama
    networks:
      - ai_net

volumes:
  ollama_models:
  openwebui_data:

networks:
  ai_net:
    driver: bridge

---

Diagnostic Verification Steps

# 1. Test Ollama API response from the host
curl http://localhost:11434/api/tags

# 2. Test connectivity from INSIDE the Open WebUI container to Ollama
docker exec -it open-webui curl -I http://ollama:11434/api/tags

# 3. Pull a test model if none are installed
docker exec -it ollama ollama run llama3.2:3b "Hello"

---

Related Guides & Blueprints

• Blueprint: Private AI Stack: Ollama, Open WebUI & OPNsense
• Tunneling: Fix: Cloudflare Tunnel 502 Bad Gateway with Open WebUI
• GPU Detection: Fix: Ollama GPU Not Detected in Docker