Deploy To Production

Documentation status

Validation: IN PROGRESS – This how-to summarizes the current recommended path for production deployments.

Last reviewed: 2025-11-21

Community: If you run this end to end, please update this page via Contributing to docs.

What “production” means here

This guide is for taking a tested Qubinode Navigator deployment into a production-style environment:

A single-node KVM hypervisor host prepared with Qubinode Navigator.
Environment-specific deployment (Hetzner / Red Hat Demo / other bare metal) completed using the platform guides.
Optional Apache Airflow + AI Assistant for orchestration.

Prerequisites

Supported OS: RHEL 8/9/10, CentOS Stream 10, Rocky Linux, or Fedora.
SSH access with sudo/root privileges.
Sufficient CPU, RAM, and disk as per the platform guides.
DNS / domain ready if exposing public endpoints.
Review these docs first:

Step 1 – Prepare the host

Follow the Clean Install Guide to bring the host to a known-good baseline.
Verify:
- KVM / virtualization is enabled and working.
- Network interfaces and storage layout match the chosen deployment guide.

Step 2 – Choose a deployment path

Use the platform-specific guides under ../deployments/:

Hetzner: Deploying on Hetzner
Red Hat Product Demo System: Deploying on Red Hat Product Demo System
Other bare metal: Use the appropriate deployment guide for your environment.

These guides describe how to configure notouch.env, /tmp/config.yml, and other environment variables for non-interactive deployments.

Step 3 – Run the production deployment script

On the prepared host, the primary production entry point is:

git clone https://github.com/Qubinode/qubinode_navigator.git
cd qubinode_navigator

./deploy-qubinode-with-airflow.sh

Ensure your .env / notouch.env / /tmp/config.yml are configured as described in the deployment guides.
Confirm that the script completes successfully and nginx/Airflow are up.

Note for developers: other scripts such as setup_modernized.sh and deploy-qubinode.sh exist for framework and legacy flows, but end‑user production deployments should prefer deploy-qubinode-with-airflow.sh.

Step 4 – Optional: enable Airflow orchestration

If you want production-style workflow orchestration and chat-driven operations:

Read:

Then:

Enable Airflow via the documented environment variables.
Deploy the Airflow sidecar and example DAGs.
Integrate with the AI Assistant for chat-based workflow management.
To deploy the MCP services (Airflow MCP server + AI Assistant) for MCP clients (e.g. Claude Desktop), run:

cd /root/qubinode_navigator
./deploy-fastmcp-production.sh

Ensure podman-compose is installed and your MCP configuration matches the MCP documentation.

Step 5 – Verify and monitor

Use the existing production verification docs to confirm your deployment:

Production Deployment Test Results

At minimum, verify:

Hypervisor services (KVM/libvirt, Cockpit, SSH) are healthy.
Qubinode Navigator workflows complete without errors.
Optional Airflow UI and AI Assistant endpoints are reachable and responsive.

If you discover additional checks that should be included here, please propose them via a documentation contribution.

2. Execute the task

# Main command
echo "Execute task"

3. Verify results

# Verification command
echo "Verify success"

Troubleshooting

If you encounter issues:

Check condition 1
Verify setting 2