Ollama and Open-WebUI Docker Setup

This repository provides the necessary configuration files to deploy two Docker services, Ollama and Open-WebUI, using Docker Compose and Jenkins for automated management. These services are deployed on the shark-wrangler node, making use of NVIDIA GPU acceleration for Ollama.

Prerequisites

To get started, ensure you have the following installed and configured:

• Docker and Docker Compose
• NVIDIA GPU drivers (compatible with the Ollama container)
• Jenkins with access to the shark-wrangler node

Services Overview

Ollama

• Image: ollama/ollama:latest
• Container Name: ollama
• Ports: 11434:11434
• Environment Variables:
  • OLLAMA_LLM_LIBRARY=cuda_v12 for GPU acceleration.
• Volumes: Mounts data at /root/.ollama to persist Ollama's state.
• Networks: Connected to shared_net.

Open-WebUI

• Image: ghcr.io/open-webui/open-webui:dev
• Container Name: open-webui
• Ports: 3000:8080
• Environment Variables:
  • OAuth Configuration: Enables login and OAuth features.
  • OpenAI API Key: Configured to enable integrations.
  • OLLAMA_BASE_URL: Points to the running Ollama instance for interoperability.
• Volumes: Mounts data at /app/backend/data to persist WebUI state.
• Networks: Connected to shared_net.

Docker Compose Configuration

The Docker Compose configuration deploys the following:

• ollama and open-webui services with defined volumes and ports.
• shared_net network for container communication.

The relevant section in docker-compose.yaml:

docker-compose.yaml
services:
  ollama:
    image: ollama/ollama:latest
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              capabilities: [gpu]
              count: all
    ports:
      - "11434:11434"
    environment:
      - OLLAMA_LLM_LIBRARY=cuda_v12
    volumes:
      - ollama:/root/.ollama
    restart: always
    container_name: ollama
    networks:
      - shared_net

  open_webui:
    image: ghcr.io/open-webui/open-webui:dev
    ports:
      - "3000:8080"
    volumes:
      - open-webui:/app/backend/data
    restart: always
    container_name: open-webui
    networks:
      - shared_net
    environment:
      - ENV=dev
      - ENABLE_LOGIN_FORM=true
      - ENABLE_OAUTH_SIGNUP=true
      - OAUTH_USERNAME_CLAIM=name
      - OAUTH_MERGE_ACCOUNTS_BY_EMAIL=true
      - OAUTH_PROVIDER_NAME=git.adamoutler.com permissions
      - WEBUI_URL=https://ai.hackedyour.info
      - OPENID_PROVIDER_URL=https://git.adamoutler.com/.well-known/openid-configuration
      - OPENID_REDIRECT_URI=https://ai.hackedyour.info/oauth/oidc/callback
      - OLLAMA_BASE_URL=http://ollama:11434
      - OPENAI_API_KEY=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
      - OAUTH_CLIENT_ID=a7bb86aa-fea4-4c0d-a07f-3e0ca4428159
      - OAUTH_CLIENT_SECRET=gto_v5va4hznsqck4myz33fve7md2tk7nntgpwoskvki6mxiwg3ux6vq

volumes:
  ollama:
  open-webui:

networks:
  shared_net:

Jenkinsfile for CI/CD

The Jenkinsfile automates the deployment of Ollama and Open-WebUI services using Docker Compose:

Pipeline Stages

1. Check NVIDIA Driver Version:

   • Ensures NVIDIA drivers are available and compatible.

2. Check Ollama and Open-WebUI Versions (Before Deploy):

   • Retrieves the current version of the Ollama and Open-WebUI containers.

3. Deploy with Docker Compose:

   • Pulls the latest images and recreates the containers using Docker Compose.

4. Check Ollama and Open-WebUI Versions (After Deploy):

   • Ensures that the services are running and updated to the latest version after deployment.

The relevant Jenkinsfile snippet:

pipeline {
    agent {
        node { label 'shark-wrangler' }
    }
    environment {
        DOCKER_HOST = 'unix:///var/run/docker.sock'
    }
    stages {
        stage('Check NVIDIA Driver Version') {
            steps {
                script {
                    catchError(buildResult: 'SUCCESS', stageResult: 'UNSTABLE') {
                        sh 'nvidia-smi --query-gpu=driver_version --format=csv,noheader'
                    }
                }
            }
        }
        stage('Check Ollama and Open-WebUI Versions (Before Deploy)') {
            steps {
                script {
                    catchError(buildResult: 'SUCCESS', stageResult: 'UNSTABLE') {
                        echo 'Checking Ollama version before deploy:'
                        sh 'docker exec -i ollama ollama -v || echo "Ollama check failed"'

                        echo 'Checking Open-WebUI version before deploy:'
                        sh 'docker exec -i open-webui jq -r .version /app/package.json || echo "Open-WebUI check failed"'
                    }
                }
            }
        }
        stage('Deploy with Docker Compose') {
            steps {
                script {
                    sh '''
                      docker pull ollama/ollama
                      docker pull ghcr.io/open-webui/open-webui:main
                      docker compose up -d  --force-recreate
                      '''
                }
            }
        }
        stage('Check Ollama and Open-WebUI Versions (After Deploy)') {
            steps {
                script {
                    catchError(buildResult: 'SUCCESS', stageResult: 'UNSTABLE') {
                        echo 'Checking Ollama version after deploy:'
                        sh 'docker exec -i ollama ollama -v || echo "Ollama check failed"'

                        echo 'Checking Open-WebUI version after deploy:'
                        sh 'docker exec -i open-webui jq -r .version /app/package.json || echo "Open-WebUI check failed"'
                    }
                }
            }
        }
    }
    post {
        always {
            echo 'Pipeline finished.'
        }
    }
}

Usage

To deploy the services, simply use Docker Compose:

docker compose up -d

For automated deployments, you can use Jenkins with the provided Jenkinsfile to ensure the latest versions are deployed and tested.

Notes

• Ensure all environment variables are correctly set, particularly the OAuth credentials and OpenAI API key.
• Update Docker images regularly to maintain security and access new features.

License

This project is open-source and licensed under the MIT License.