Quick Start
1. Prepare prerequisites
- Unraid host with virtualization and
/dev/kvm. - Talos cluster reachable from your machine.
- GitHub App credentials for ARC.
- Tailscale tailnet available for CI.
2. Manual infrastructure apply (operator)
bash
make tf-init
make tf-plan
make tf-applyUse your local .env/TF_VAR_* process for sensitive values.
3. Bootstrap Talos and export configs
Ensure ~/.kube/config and optional ~/.talos/config are valid for this cluster.
bash
base64 -i ~/.kube/config | tr -d '\n'
base64 -i ~/.talos/config | tr -d '\n'Store these in repo secrets per Secrets Reference.
4. Run apply workflow
Dispatch .github/workflows/apply.yml with:
run_terraform=falserun_arc_install=truerun_kvm_autoscaler_apply=true
5. Validate runtime path
- Run
kvm-smokeworkflow. - Confirm dind runner pods land on KVM nodes.
- Confirm
/dev/kvmis visible in runner container.
6. Start using runner classes
- Route default jobs to container scale set.
- Route runtime-real jobs to dind/KVM scale set.