WIP: separate traefik for public

AGENTS.md

@@ -10,5 +10,23 @@ I have a DNS rewrite pointing *.internal to 10.0.1.250 which is traefik-internal
 
 /dogstore/ is a NFS path that's available to all nodes
 
-
 secrets are managed by sops
+
+## Load balancers
+
+Two LB implementations coexist: k3s klipper (servicelb) and MetalLB. They are
+separated by `loadBalancerClass` so they don't conflict.
+
+- **klipper** handles services with NO `loadBalancerClass`. It creates svclb
+  DaemonSet pods that bind host ports directly on every node.
+- **MetalLB** handles services with `loadBalancerClass: metallb`. Its pool has
+  `autoAssign: false`, so it only assigns IPs to services that explicitly
+  request a pool via the `metallb.io/address-pool` annotation.
+
+| Service          | loadBalancerClass | LB       | External IPs            |
+|------------------|-------------------|----------|-------------------------|
+| traefik          | (none)            | klipper  | node IPs (10.0.1.2 etc) |
+| traefik-internal | metallb           | MetalLB  | 10.0.1.250              |
+
+`loadBalancerClass` is immutable on k8s Services. Changing it requires deleting
+the Service first, then redeploying (`kubectl delete svc … && helm upgrade`).

charts/metallb/templates/pool.yaml

@@ -4,6 +4,7 @@ kind: IPAddressPool
 metadata:
   name: {{ .Values.pool.name }}
 spec:
+  autoAssign: false
   addresses:
     {{- range .Values.pool.addresses }}
     - {{ . }}

charts/metallb/values.yaml

@@ -1,4 +1,5 @@
 metallb:
+  loadBalancerClass: metallb
   controller:
     nodeSelector:
       node-role.kubernetes.io/control-plane: "true"
@@ -7,7 +8,7 @@ metallb:
         effect: NoSchedule
 
 pool:
-  enabled: false
+  enabled: true
   name: internal
   addresses:
     - "10.0.1.250-10.0.1.250"

charts/traefik-config/templates/traefik-config.yaml

@@ -5,9 +5,11 @@ metadata:
   namespace: kube-system
 spec:
   valuesContent: |-
-    service:
-      spec:
-        loadBalancerClass: io.k3s.klipper
+    nodeSelector:
+      node-role.kubernetes.io/control-plane: "true"
+    tolerations:
+      - key: node-role.kubernetes.io/control-plane
+        effect: NoSchedule
     logs:
       general:
         level: WARN
@@ -35,7 +37,12 @@ spec:
           secretKeyRef:
             name: {{ .Values.cloudflare.secretName }}
             key: {{ .Values.cloudflare.secretKey }}
-    persistence:
-      enabled: true
-      storageClass: longhorn
-      size: 128Mi
+    deployment:
+      additionalVolumes:
+        - name: acme
+          hostPath:
+            path: /dogstore/service-data/.letsencrypt
+            type: DirectoryOrCreate
+    additionalVolumeMounts:
+      - name: acme
+        mountPath: /letsencrypt

charts/traefik-config/values.yaml

@@ -9,4 +9,4 @@ cloudflare:
   secretKey: CF_DNS_API_TOKEN
 
 letsencrypt:
-  storagePath: /data/acme.json
+  storagePath: /letsencrypt/acme.json

charts/traefik-internal/templates/traefik-internal.yaml

@@ -82,6 +82,7 @@ metadata:
     metallb.io/address-pool: internal
 spec:
   type: LoadBalancer
+  loadBalancerClass: metallb
   loadBalancerIP: {{ .Values.loadBalancerIP }}
   selector:
     app: traefik-internal