Skip to content

GCP_DEPLOYMENT.md

Latest commit

 

History

History
298 lines (213 loc) · 6.24 KB

GCP_DEPLOYMENT.md

File metadata and controls

298 lines (213 loc) · 6.24 KB
Raw

Deploying to Google Cloud Platform with Docker

This guide walks you through deploying the dev-workflow MCP server to Google Cloud Compute Engine using Docker and PostgreSQL.

Prerequisites

  • Google Cloud Platform account with a Compute Engine instance running
  • SSH access to your GCP instance
  • Git installed locally and on GCP instance
  • Basic familiarity with Docker and SSH

Architecture Overview

The deployment consists of two Docker containers:

  • MCP Server: Node.js application running the dev-workflow MCP server
  • PostgreSQL: Database for persistent workflow state storage

Communication between your local MCP client and the cloud server happens via SSH tunneling.

Step 1: Prepare Your GCP Instance

1.1 SSH into Your Instance

ssh your-username@YOUR_GCP_IP

1.2 Run the Setup Script

From your local machine, copy the setup script to your GCP instance:

scp scripts/setup-gcp-instance.sh your-username@YOUR_GCP_IP:~/
ssh your-username@YOUR_GCP_IP 'bash ~/setup-gcp-instance.sh'

This script will:

  • Install Docker and Docker Compose
  • Install Git
  • Create the application directory

Note: You may need to log out and back in for Docker group permissions to take effect.

Step 2: Deploy the Application

2.1 Clone the Repository

On your GCP instance:

cd ~/dev-workflow-mcp-server
git clone https://github.com/programinglive/dev-workflow-mcp-server.git .

2.2 Configure Environment Variables

Create a .env file from the production template:

cp .env.production .env
nano .env

Update the following values:

  • POSTGRES_PASSWORD: Set a strong, unique password
  • DEV_WORKFLOW_DB_URL: Update with the same password
  • DEV_WORKFLOW_USER_ID: (Optional) Set your username

Example .env:

DEV_WORKFLOW_DB_TYPE=postgres
DEV_WORKFLOW_DB_URL=postgres://devworkflow:MySecureP@ssw0rd@postgres:5432/devworkflow
POSTGRES_PASSWORD=MySecureP@ssw0rd
DEV_WORKFLOW_USER_ID=dhika
NODE_ENV=production

2.3 Start the Containers

docker-compose up -d

2.4 Verify Deployment

Check that both containers are running:

docker-compose ps

You should see:

NAME                      STATUS
dev-workflow-mcp          Up
dev-workflow-postgres     Up (healthy)

View logs:

docker-compose logs -f mcp-server

Step 3: Configure Local MCP Client

Update your local MCP client configuration to connect via SSH.

Windows (Claude Desktop)

Edit C:\Users\DHIKA\.gemini\antigravity\mcp_config.json:

{
  "mcpServers": {
    "dev-workflow": {
      "command": "ssh",
      "args": [
        "-i", "C:\\Users\\DHIKA\\.ssh\\id_rsa",
        "your-username@34.50.121.142",
        "docker", "exec", "-i", "dev-workflow-mcp", "node", "/app/index.js"
      ]
    }
  }
}

Linux/Mac

Edit your MCP config file:

{
  "mcpServers": {
    "dev-workflow": {
      "command": "ssh",
      "args": [
        "-i", "/home/user/.ssh/id_rsa",
        "your-username@34.50.121.142",
        "docker", "exec", "-i", "dev-workflow-mcp", "node", "/app/index.js"
      ]
    }
  }
}

Important: Replace:

  • your-username with your actual GCP username
  • 34.50.121.142 with your actual GCP external IP
  • SSH key path with your actual key location

Step 4: Test the Connection

4.1 Test SSH Connection

From your local machine:

ssh your-username@34.50.121.142 'docker exec -i dev-workflow-mcp node /app/index.js call get_workflow_status'

You should see the workflow status output.

4.2 Test via MCP Client

Restart your MCP client (e.g., Claude Desktop) and try using a dev-workflow command like start_task.

Updating the Deployment

Automated Deployment

Use the deployment script from your local machine:

# Set environment variables
export GCP_USER=your-username
export GCP_HOST=34.50.121.142
export GCP_SSH_KEY=~/.ssh/id_rsa

# Run deployment
bash scripts/deploy-gcp.sh

Manual Update

On your GCP instance:

cd ~/dev-workflow-mcp-server
git pull origin main
docker-compose down
docker-compose build --no-cache
docker-compose up -d

Database Management

Backup Database

docker exec dev-workflow-postgres pg_dump -U devworkflow devworkflow > backup.sql

Restore Database

cat backup.sql | docker exec -i dev-workflow-postgres psql -U devworkflow devworkflow

Access PostgreSQL Console

docker exec -it dev-workflow-postgres psql -U devworkflow devworkflow

Common commands:

  • \dt - List tables
  • \d table_name - Describe table
  • SELECT * FROM workflow_state; - View workflow data

Troubleshooting

Container Won't Start

Check logs:

docker-compose logs mcp-server
docker-compose logs postgres

Database Connection Issues

Verify PostgreSQL is healthy:

docker-compose ps
docker exec dev-workflow-postgres pg_isready -U devworkflow

SSH Connection Fails

Test basic SSH:

ssh -v your-username@YOUR_GCP_IP

Check SSH key permissions:

chmod 600 ~/.ssh/id_rsa

MCP Client Can't Connect

  1. Verify SSH works: ssh your-username@YOUR_GCP_IP 'echo test'
  2. Test Docker exec: ssh your-username@YOUR_GCP_IP 'docker ps'
  3. Check MCP server: ssh your-username@YOUR_GCP_IP 'docker logs dev-workflow-mcp'

Security Best Practices

  1. Change Default Passwords: Always use strong, unique passwords in .env
  2. Firewall Configuration: Only expose necessary ports (SSH port 22)
  3. SSH Key Authentication: Disable password authentication, use SSH keys only
  4. Regular Updates: Keep Docker images and system packages updated
  5. Backup Regularly: Schedule automated database backups

Performance Optimization

Monitor Resource Usage

docker stats

Increase PostgreSQL Performance

Edit docker-compose.yml to add PostgreSQL tuning:

postgres:
  environment:
    POSTGRES_SHARED_BUFFERS: 256MB
    POSTGRES_EFFECTIVE_CACHE_SIZE: 1GB

Uninstalling

To completely remove the deployment:

cd ~/dev-workflow-mcp-server
docker-compose down -v  # -v removes volumes (deletes data!)
cd ~
rm -rf dev-workflow-mcp-server

Support

For issues or questions: