The operator ruled the implementation repo's main (50c8719) the latest and accepted state. Sync the spec to it. Accepted changes are reflected: one bare-metal environment, TrueNAS bulk storage, R2 off-cluster backup, CloudNativePG, single-node VPN gateway with Cilium fail-closed, inbound WireGuard, per-app auth sidecars, Renovate, the CI gate as approver, and the pm-agent. New ADR-0017..0025; ADR-0016 corrected and accepted; R-* IDs unchanged, eight amended. Where the implementation has simply not met a requirement, the requirement stands and the shortfall is recorded in docs/16-known-gaps.md. Co-Authored-By: Claude Opus 5.5 <[email protected]>
12 KiB
01 — Architecture Overview
This document is the map. It names the chosen stack, shows how the layers fit, and points to the ADR that locks in each choice and the domain spec that details it. Do not change a technology choice here without a superseding ADR.
The stack at a glance
| Layer | Choice | Serves | ADR | Spec |
|---|---|---|---|---|
| Kubernetes distro | k3s (HA, embedded etcd) | R-PLAT-01/02 | ADR-0001 | 02 |
| CNI / network policy / egress | Cilium (+ Hubble, Gateway API, Egress Gateway, L2 LB-IPAM) | R-NET-01..07 | ADR-0002 | 03 |
| Distributed storage | Longhorn | R-STATE-01..02 | ADR-0003 | 04 |
| Bulk media | NFS from a standalone TrueNAS on the production LAN | R-STATE-04 | ADR-0018 | 04 |
| PostgreSQL | CloudNativePG (CNPG) | R-STATE-01/03, R-IDP-05 | ADR-0019 | 04 |
| GitOps engine | Flux CD | R-GITOPS-01..03 | ADR-0004 | 05 |
| VM workloads (IaaS) | KubeVirt (+ CDI) | R-PLAT-03 | ADR-0005 | 06 |
| TLS | cert-manager + Let's Encrypt (DNS-01) | R-TLS-01..03 | ADR-0006 | 08 |
| Secrets | SOPS + age (Flux-native) | R-SEC-01 | ADR-0007 | 07 |
| VPN egress | Node-level PIA/OpenVPN tunnel on a single gateway node, steered by Cilium Egress Gateway (Cilium-enforced fail-closed) | R-NET-05/06/07 | ADR-0015, ADR-0020 (provider 0013) | 03 |
| Automated updates | Renovate (images and charts), every change PR-gated | R-UPD-01..03 | ADR-0023 | 11 |
| DNS records | ExternalDNS (public via Cloudflare) + PiHole (explicit LAN answers) | R-DNS-01..03 | — | 09 |
| Ingress (north-south) | Gateway API implemented by Cilium | R-DNS-01, R-TLS-01 | ADR-0002 | 09 |
| Public web ingress | Cloudflare Tunnel (outbound-only cloudflared) | R-DNS-04/05, R-SEC-04 | ADR-0016 | 09 |
| Remote access | WireGuard on a hostNetwork DaemonSet | R-SEC-02/04 | ADR-0021 | 03 |
| Observability | kube-prometheus-stack + Loki + Grafana Alloy + Alertmanager→SMTP | R-OBS-01..04 | — | 10 |
| Backup/DR | Velero (kopia) + CNPG Barman → Cloudflare R2; no Longhorn backup target | R-STATE-03 | ADR-0018 | 04 |
| Node/infra provisioning | Imaged Debian on bare metal + Ansible (baseline, k3s bootstrap) | R-ENV-01..03 | ADR-0001, ADR-0017 | 02 |
| Control-plane endpoint | kube-vip virtual IP | R-PLAT-01 | ADR-0017 | 02 |
| Node patching & reboots | security-only unattended-upgrades + kured | R-PLAT-01 | — | 02 |
| Identity / SSO | Keycloak (OIDC + SAML) + per-app oauth2-proxy forward-auth sidecars | R-IDP-01..06 | ADR-0014, ADR-0022 | 15 |
| Command center UI | Homepage + Headlamp + Grafana panels (behind Keycloak SSO) | R-C2-01/02 | — | 12 |
| Conformance testing | Kyverno (L0 policy) + Chainsaw (L1/L2/L3 behavior + fault-injection) | R-TEST-01..06 | ADR-0010, ADR-0011 | 14 |
| Delivery & merge gate | Gitea + Gitea Actions; the CI gate is the approver | R-GITOPS-01, R-TEST-03 | ADR-0024 | 05 |
| Image builds | CI-run rootless BuildKit, published to Gitea's registry | — | — | 05 |
| AI Program Manager | pm-agent (issues to pull requests through the gate) | — | ADR-0025 | 05 |
| Hardware transcode | Intel GPU device plugin (Plex Quick Sync) | — | — | 13 |
Layer cake
┌──────────────────────────────────────────────────────────────────────────┐
│ Command & Control UI (home.keithjackson.dev root) │
│ Homepage catalog · Headlamp k8s console · Grafana dashboards · Flux status │
├──────────────────────────────────────────────────────────────────────────┤
│ Workloads │
│ PaaS: containers (Plex, PiHole, Nextcloud, Gitea, Seerr, …) │
│ IaaS: KubeVirt VMs (anything that can't be a container) │
├──────────────────────────────────────────────────────────────────────────┤
│ Platform services │
│ Flux (GitOps) · cert-manager (TLS) · ExternalDNS · Gateway API │
│ Keycloak (SSO/IdP) + oauth2-proxy (forward-auth) │
│ kube-prometheus-stack + Loki + Alloy (observability) │
│ Velero + CNPG Barman (backup) · Renovate (updates) · SOPS (secrets) │
│ cloudflared (public ingress) · CNPG (PostgreSQL) · kured (node reboots) │
├──────────────────────────────────────────────────────────────────────────┤
│ Storage: Longhorn (replicated block) + TrueNAS NFS bulk tier (ADR-0018) │
├──────────────────────────────────────────────────────────────────────────┤
│ Network: Cilium (default-deny NetworkPolicy · Egress Gateway · Hubble) │
│ VPN egress (node tunnel + Cilium egress GW) · WireGuard inbound │
├──────────────────────────────────────────────────────────────────────────┤
│ Kubernetes: k3s (3 nodes, HA control plane, embedded etcd) │
├──────────────────────────────────────────────────────────────────────────┤
│ Nodes: 3× mini-PC bare metal, local disks (imaged Debian + Ansible) │
├──────────────────────────────────────────────────────────────────────────┤
│ Edge: SOHO router — single WAN, NAT; forwards only WireGuard + Plex │
│ Public web ingress: Cloudflare Tunnel (outbound-only cloudflared, ADR-0016)│
└──────────────────────────────────────────────────────────────────────────┘
Two clearly-separated virtualization layers (do not conflate)
This platform has two kinds of VM, and confusing them is the most common design error:
-
Infrastructure VMs = the cluster nodes themselves. These are the bare-metal mini-PCs, imaged and then baselined and joined by Ansible; "VM" is historical shorthand for this layer, and it remains the place a node would be virtualized. See
02-environments-and-provisioning.md. -
Application VMs = IaaS workloads that run inside the cluster via KubeVirt. These are managed by the same Kubernetes API/YAML/GitOps as containers. They satisfy R-PLAT-03 ("not everything fits in a container"). See
06-vm-workloads-kubevirt.md.
Layer 1 is how the platform exists. Layer 2 is a feature the platform offers.
How a service flows through the platform (the golden path)
- Operator adds/edits a ServiceDefinition in the git catalog (see
13-service-catalog.md) declaring: kind (container|vm), image/source, storage, egress class (internet|vpn), exposure (internal|public), dependencies (explicit allow rules), update policy, and observability expectations. - Flux reconciles the change into the cluster.
- Longhorn provisions replicated persistent volumes, so state is durable; large media attaches from the NFS bulk tier.
- Cilium applies default-deny plus the explicitly declared allow rules; assigns the correct egress path (local internet or VPN, with kill-switch).
- A
publicservice gets a proxied ExternalDNS record at<service>.home.keithjackson.devand arrives through the Cloudflare Tunnel. Aninternalservice has no public record and is answered on the LAN by PiHole. Gateway API routes north-south traffic for both; cert-manager issues a trusted TLS cert via ACME DNS-01. - kube-prometheus-stack + Loki scrape metrics/logs; Alertmanager emails on issues; the C2 UI lists and links the service.
- Renovate keeps the service on the latest stable version per policy, through machine-authored changes that pass the merge gate.
- Velero and CNPG Barman, writing off-cluster, protect the data; destroy/rebuild restores state.
Every numbered step maps to a domain spec and a workstream in roadmap/workstreams.md.
Design principles
- Declarative everything. If it can't be expressed as YAML reconciled from git, question it.
- One control plane. Containers and VMs share the Kubernetes surface; operators learn one model.
- Secure by default. Deny-by-default network, least-privilege identity, encrypted secrets, trusted TLS everywhere.
- Durable by default. No workload is "pet" hardware; state lives on replicated storage with tested restore.
- Hands-off operations. Updates, cert renewal, DNS, and drift correction are automated.
- Promotable. Configuration is one base plus an environment-scoped overlay.
- Provable. Every requirement is verified by an automated, reproducible, behavior-based conformance test (14, ADR-0010); "done" is a green, requirement-tagged suite, not a manual demo.
- Spec-fixed. The specification is human-owned and immutable to agents (R-TEST-06); agents move the implementation and its tests toward the spec, never the spec toward the implementation.