Tailscale, Kubernetes & Rancher on VergeOS: A Homelab Field Report
David Vincent runs VergeOS in his homelab and wrote the playbook the hard way. Host George Crump sits down with David for a relaxed, hands-on walkthrough of identity-based network reachability with Tailscale, a reachable LoadBalancer on k3s, and getting RKE2 provisioning right with Rancher — followed by an open fireside chat.
Customers, In Their Own Words
Casual Fridays is our customer-led webinar series. Real VergeOS users present their own implementations, then sit down with host George Crump for a friendly fireside chat. It is lighter and more conversational than a standard product webinar — less pitch, more “here is exactly how I did it, including what broke.” This is the first session in the series.
What David Will Walk Through
Reachability Without Routing
Using Tailscale subnet routing to reach an isolated VergeOS management network by identity, not BGP or OSPF — a 1-vCPU VM exposes the whole subnet, nothing gets publicly exposed.
Kubernetes on VergeOS
Standing up k3s with the VergeOS CSI driver and cloud-controller-manager: two VMs, two Helm releases, and an IP pool for LoadBalancer services.
A LoadBalancer You Can Actually Reach
The SNAT companion rule nobody documents — the difference between an internal-only LoadBalancer and one that answers from your laptop, your router, and across Tailscale.
RKE2 Provisioning with Rancher
Why “the template is the integration.” Three wrong templates before a right one, and the QEMU guest-agent detail that turns “hangs forever” into “works in minutes.”
Agent-Native Operations
The thread running through it all: the vrg CLI, verifying against the API instead of the UI, and writing the playbook down so the next session starts already knowing.
Fireside Chat & Open Q&A
George Crump interviews David in a relaxed, unscripted conversation about running real infrastructure in a homelab — then opens the floor. Bring your questions and war stories.
Built for People Who Actually Run the Stack
This is a builder’s session. If you architect infrastructure, run Kubernetes, manage virtualization, or just like seeing how someone solved a real reachability problem without expanding the trust boundary, you will leave with patterns you can use the same week. Homelabbers and production engineers alike will find something here — the lessons scale both directions.
One Operating System for Your Data Center
VergeOS collapses virtualization, storage, and networking into a single operating system, replacing the stack of hypervisor, SAN, and network controllers most data centers run today. It is the platform David builds on — and the reason a homelabber can wire up Kubernetes, an identity-based overlay network, and Rancher provisioning from one control surface, scriptable through the vrg CLI and a clean REST API.
Save Your Seat
Register for the first Casual Fridays session. We’ll send the join link and the on-demand recording.