This guide covers adding an NVIDIA Orin Nano as a worker node to an existing RKE2 cluster managed by Rancher, with full GPU support enabled.
By default, running the Rancher custom cluster script will not enable GPU support in containerd. This guide provides the necessary steps to configure the NVIDIA runtime for RKE2/containerd.
| #!/bin/bash | |
| # This script pulls a Helm chart from an OCI registry or uses a local Chart.yaml file, | |
| # extracts container images defined in the `helm.sh/images` annotation of the Chart.yaml, | |
| # and generates a Hauler Images manifest. | |
| OCI_LOCATION="" | |
| OUTPUT_FILE="" | |
| LOCAL_CHART="" | |
| SOURCE_REGISTRY="" |
| # Define some variables | |
| NAMESPACE ?= default | |
| BACKUP_NAME ?= backup-$(shell date +%Y%m%d%H%M%S) | |
| RESTORE_NAME ?= restore-$(shell date +%Y%m%d%H%M%S) | |
| SCHEDULE_NAME ?= daily-backup | |
| BUCKET ?= velero-backup-bucket | |
| POLL_INTERVAL ?= 5 # Check backup status every 5 seconds | |
| # Help message | |
| .PHONY: help |
| #!/bin/bash | |
| # Script to delete a Proxmox VM and all associated hard disks. | |
| # Usage: | |
| # ./delete_vm_with_disks.sh <vmid> - Deletes the specified VM after user confirmation. | |
| # ./delete_vm_with_disks.sh <vmid> --force - Force deletes the specified VM without confirmation. | |
| # | |
| # Arguments: | |
| # <vmid> : The ID of the Proxmox VM to delete. | |
| # --force : Optional argument to skip the confirmation prompt. |
| #!/bin/bash | |
| # Check if an argument was provided | |
| if [[ -z "$1" ]]; then | |
| echo "Usage: $0 <node number (1-4)>" | |
| exit 1 | |
| fi | |
| # Map node number to serial device | |
| case "$1" in |
| --- | |
| - name: Setup Let's Encrypt with Cloudflare DNS on PiKVM | |
| hosts: pikvm | |
| become: yes | |
| vars: | |
| cloudflare_email: "user@example.com" | |
| cloudflare_api_key: "your-cloudflare-api-key" | |
| domain: "pikvm.example.com" | |
| auth_file_path: "/var/lib/kvmd/pst/data/certbot/runroot/.cloudflare.auth" |
| #!/bin/bash | |
| # Standard disclaimer | |
| echo "This script is provided as-is without any warranty. Use it at your own risk. Do not use this script for any illegal activities." | |
| # Default values | |
| DEFAULT_NAME="synology-arc-test" | |
| DEFAULT_MEMORY="4096" | |
| DEFAULT_CORES="4" | |
| DEFAULT_VM_ID="119" |
Once you have serial access to the device, reboot the host and boot to either USB or PXE and install Harvester per usual.
Because the small Atom-based witness node operates in a headless capacity, we need to serial into the node. Here's how you do it.
| #!/bin/bash | |
| ## https://forum.proxmox.com/threads/console-failed-to-connect-to-server-host-key-verification-failed.78957/ | |
| # Usage: ./proxmox_webvnc_fix.sh <shorthostname> | |
| # Notes: may need to run twice for the fix to take. | |
| set -e | |
| DOMAIN="lark.lol" | |
| if [ $# -eq 0 ]; then | |
| echo "No hostname provided. Exiting..." |
| prep: | |
| helm repo add nvidia https://helm.ngc.nvidia.com/nvidia \ | |
| && helm repo update | |
| install: | |
| helm install --wait nvidiagpu \ | |
| -n gpu-operator --create-namespace \ | |
| --set toolkit.env[0].name=CONTAINERD_CONFIG \ | |
| --set toolkit.env[0].value=/var/lib/rancher/k3s/agent/etc/containerd/config.toml \ | |
| --set toolkit.env[1].name=CONTAINERD_SOCKET \ |
