Add Rancher and Kubernetes integration guides

[jasnoyaeger]

Summary

• Split the combined Rancher/Kubernetes doc into two standalone guides:
  • Kubernetes Integration — CSI Driver + Cloud Controller Manager, works with any K8s cluster on VergeOS
  • Rancher Integration — Docker Machine Driver + UI Extension, Rancher-specific provisioning
• Marks Ubuntu 24.04 as the only supported template OS (was listed as "recommended")
• Fixes "Why Not Longhorn?" section framing and CCM basic install example

Test plan

• Verify both pages render correctly on the dev server
• Confirm cross-links between the two docs work
• Check nav ordering under "Kubernetes & Containers"

🤖 Generated with Claude Code

[dzarzycki]

Review

Risk Level: 🟡 Medium — Good content structure, but several items need updating to match current implementation.

Critical Issues

1. Node driver + UI extension installation instructions are outdated

File: rancher-integration.md lines 96-130

The doc describes manual kubectl apply for the NodeDriver and a separate helm install for the UI extension. We now have a combined Helm chart (vergeos-node-driver) that installs both the NodeDriver resource and UIPlugin in one step.

Users following these instructions will get a partial install or double-install components.

Should be replaced with:

helm repo add verge-io https://verge-io.github.io/helm-charts
helm repo update
helm install vergeos-node-driver verge-io/vergeos-node-driver \
  -n cattle-system \
  --set "vergeosHosts={vergeos.example.com}"

Or via Rancher UI: add https://verge-io.github.io/helm-charts as a repository, then install VergeOS Node Driver from the Extensions page.

The "restart Rancher" instruction (line 131-133) is also unnecessary with the Helm approach.

2. UI Extension description claims features we removed

File: rancher-integration.md lines 181-186

The doc says:

• "Template VM selector (auto-populated from VergeOS API, filtered to Ubuntu 24.04 VMs with guest agent)"
• "Network selector (auto-populated from VergeOS VNets)"

These API-driven dropdowns were removed due to the Rancher proxy not supporting TLS skip for self-signed certs. They're now plain text inputs. Should read:

• Template VM name (name of the VergeOS VM to clone)
• Network name (VergeOS VNet to attach to)

3. CCM host value missing scheme

File: kubernetes-integration.md line 201

CCM install example uses --set vergeos.host=<VERGEOS_HOST> without scheme, while CSI uses --set vergeos.host=https://<VERGEOS_HOST>. We're standardizing on full URL with scheme — both should use --set vergeos.host=https://<VERGEOS_HOST>.

Recommendations

Add self-signed cert guidance

Both docs should note that self-signed certs require vergeos.verifySSL=false for CCM/CSI, and insecure=true in the cloud credential for the node driver. This will be the #1 issue in lab and many production environments.

RAM default discrepancy

rancher-integration.md line 149 — Driver options table shows RAM default as 2048, but the UI extension defaults to 4096 and the doc's own warning says 2GB will OOM with RKE2 + Calico. Should reconcile.

Mention vergeosHosts whitelist

The Helm chart's vergeosHosts value tells Rancher's proxy which VergeOS instances it's allowed to reach. Worth explaining since users will wonder what it's for.

Missing helm-charts from Repositories table

The Repositories table at the bottom of rancher-integration.md lists the driver and UI extension repos but not helm-charts — which is the actual install source.

Questions

1. The oci://ghcr.io/verge-io/ui-extension-vergeos/charts/vergeos OCI install path in the UI extension section — has this been tested? We got 403 errors from GHCR during development. The Helm repo approach (verge-io/vergeos-node-driver) is the proven path.
2. Should the Rancher doc mention cluster autoscaler setup? We validated it works end-to-end.

Security Checklist

• No hardcoded secrets or credentials
• Input validation guidance present
• Auth guidance in place (API key generation mentioned)
• TLS verification documented as configurable