Homelab GitOps Platform
-Terraform-provisioned k3s cluster on a Dell T630, deployed and kept in sync entirely through GitOps. This page is served from it — and this line proves it: edited via git push.
+Terraform-provisioned self-hosted k3s cluster on a Home server, deployed and kept in sync entirely through GitOps. This page is served from it — and this line proves it: edited via git push.
commit __GIT_SHA__
diff --git a/docs/03-flux.md b/docs/03-flux.md
index 980b683..bc04eee 100644
--- a/docs/03-flux.md
+++ b/docs/03-flux.md
@@ -452,7 +452,16 @@ podinfo.boglabob.com CNAME boglabob.com
hello.boglabob.com CNAME boglabob.com
```
-Then in Caddy's own config, using the VM's pinned IP from stage 2:
+Then in Caddy's own config, using the VM's pinned IP from stage 2. Both
+blocks point at the exact same `` — that's not a mistake, there's
+only one VM right now, and it's simultaneously the control plane *and*
+the only place pods can run ("Why one node is enough," above). Caddy
+isn't picking between podinfo/hello-app here at all — it just forwards
+everything for both hostnames to that one node's port 80, and it's
+Traefik, once the request lands there, that actually reads the `Host:`
+header and routes to the right backend. This changes in stage 4, where
+there are real node IPs to load-balance across (`docs/Caddyfile.example`'s
+multi-node blocks):
```
podinfo.boglabob.com {