lab1 workers on both PVE nodes: per-node template + MD, balanced CA

PVE is now the 2-node cluster homelab (pv1-1 + pv1-2), no shared storage: - template 101 copied to pv1-2 as 9101; MDs lab1-workers-pv1-1 (max 6) and lab1-workers-pv1-2 (max 12), PMTs pinned with allowedNodes; pool .31-.50; lab1-workers renamed for symmetry (decision #18, #7 superseded) - CA: balancing-label node.cluster.x-k8s.io/pool (+ pool label, capacity labels with os/arch, maxPods on the MDs), scan-interval 1m, v 2, default QPS (#19). Scale test: 16 pods -> 4 + 4 across both nodes. - runbook, image-builder copy procedure, fixes tables, future task to report three CA clusterapi bugs upstream, lab facts, changelog, execution log Plan: docs/plans/2026-09-29-2315-pv1-2-workers.md Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-09-30 11:07:17 +02:00
parent 5e5b6c506a
commit 6ef55778ff
14 changed files with 476 additions and 50 deletions
@@ -20,6 +20,17 @@
 #   7. MachineDeployment lab1-workers: autoscaler min/max + capacity annotations (runbook Step 5)
 #   8. MachineDeployment lab1-workers: rollout maxSurge 0 / maxUnavailable 1 (memory-tight host)
 #   9. ProxmoxCluster lab1: schedulerHints.memoryAdjustment 160 (CAPMOX memory overcommit, decision #17)
+#  10. Second worker group on PVE node pv1-2 (decision #18): MachineDeployment lab1-workers-pv1-2 +
+#      ProxmoxMachineTemplate lab1-worker-2g-pv1-2 (template copy 9101, allowedNodes [pv1-2]).
+#      local-lvm is node-local, so every node needs its own template VMID and its own MD.
+#  11. ProxmoxCluster lab1: node IP pool .31-.39 -> .31-.50 (20 addresses). Max sizes 6 (pv1-1) +
+#      12 (pv1-2) + 1 CP = 19; one address stays free for the KCP rollout surge (CAPMOX pools also
+#      hold a freed address for 60 s before reuse).
+#  12. Both worker MDs: label node.cluster.x-k8s.io/pool=workers (template + capacity labels
+#      annotation) for CA --balancing-label; capacity key maximum-pods -> maxPods.
+#  13. Symmetric per-node names (owner's request): MachineDeployment lab1-workers -> lab1-workers-pv1-1,
+#      ProxmoxMachineTemplate lab1-worker-2g -> lab1-worker-2g-pv1-1 (+ explicit allowedNodes [pv1-1]).
+#      A rename is a new object: the old MD and PMT are deleted by hand once the old MD is at 0.
 apiVersion: cluster.x-k8s.io/v1beta2
 kind: Cluster
 metadata:
@@ -59,7 +70,7 @@ spec:
  - 1.1.1.1
  ipv4Config:
    addresses:
-    - 192.168.4.31-192.168.4.39
+    - 192.168.4.31-192.168.4.50
    gateway: 192.168.4.8
    prefix: 24
  # CAPMOX reserves the configured memory of every non-template VM, stopped ones included, against
@@ -235,15 +246,18 @@ kind: MachineDeployment
 metadata:
  annotations:
    cluster.x-k8s.io/cluster-api-autoscaler-node-group-min-size: "0"
-    cluster.x-k8s.io/cluster-api-autoscaler-node-group-max-size: "5"
-    # scale-from-zero: CAPMOX publishes no status.capacity; must match ProxmoxMachineTemplate/lab1-worker-2g
+    cluster.x-k8s.io/cluster-api-autoscaler-node-group-max-size: "6"
+    # scale-from-zero: CAPMOX publishes no status.capacity; must match ProxmoxMachineTemplate/lab1-worker-2g-pv1-1
    capacity.cluster-autoscaler.kubernetes.io/cpu: "4"                # numSockets 1 × numCores 4
    capacity.cluster-autoscaler.kubernetes.io/memory: "2Gi"           # memoryMiB 2048
    capacity.cluster-autoscaler.kubernetes.io/ephemeral-disk: "40Gi"  # bootVolume sizeGb 40
-    capacity.cluster-autoscaler.kubernetes.io/maximum-pods: "110"
+    capacity.cluster-autoscaler.kubernetes.io/maxPods: "110"          # CA key; "maximum-pods" is ignored
+    # Labels for the from-zero template node. CA 1.35.2 reads MD labels from a path v1beta2 doesn't
+    # have, and CAPMOX publishes no nodeInfo, so without this os/arch are "" on the template.
+    capacity.cluster-autoscaler.kubernetes.io/labels: "node.cluster.x-k8s.io/pool=workers,kubernetes.io/os=linux,kubernetes.io/arch=amd64"
  labels:
    cluster.x-k8s.io/cluster-name: lab1
-  name: lab1-workers
+  name: lab1-workers-pv1-1
  namespace: default
 spec:
  clusterName: lab1
@@ -263,6 +277,8 @@ spec:
      labels:
        cluster.x-k8s.io/cluster-name: lab1
        node-role.kubernetes.io/node: ""
+        # CA --balancing-label; node.cluster.x-k8s.io/* is synced to the Node by CAPI
+        node.cluster.x-k8s.io/pool: workers
    spec:
      bootstrap:
        configRef:
@@ -273,17 +289,21 @@ spec:
      infrastructureRef:
        apiGroup: infrastructure.cluster.x-k8s.io
        kind: ProxmoxMachineTemplate
-        name: lab1-worker-2g
+        name: lab1-worker-2g-pv1-1
      version: v1.35.9
 ---
 apiVersion: infrastructure.cluster.x-k8s.io/v1alpha2
 kind: ProxmoxMachineTemplate
 metadata:
-  name: lab1-worker-2g
+  name: lab1-worker-2g-pv1-1
  namespace: default
 spec:
  template:
    spec:
+      # Pinned like the pv1-2 template, independent of ProxmoxCluster.allowedNodes: template 101
+      # lives on pv1-1's local-lvm.
+      allowedNodes:
+      - pv1-1
      disks:
        bootVolume:
          disk: scsi0
@@ -301,6 +321,85 @@ spec:
      sourceNode: pv1-1
      templateID: 101
 ---
+# Worker group on pv1-2. Keep in sync with lab1-workers-pv1-1 / lab1-worker-2g-pv1-1 except node,
+# template and max-size: CA --balancing-label skips resource checks, so capacity must stay identical.
+apiVersion: cluster.x-k8s.io/v1beta2
+kind: MachineDeployment
+metadata:
+  annotations:
+    cluster.x-k8s.io/cluster-api-autoscaler-node-group-min-size: "0"
+    cluster.x-k8s.io/cluster-api-autoscaler-node-group-max-size: "12"
+    # scale-from-zero: CAPMOX publishes no status.capacity; must match ProxmoxMachineTemplate/lab1-worker-2g-pv1-2
+    capacity.cluster-autoscaler.kubernetes.io/cpu: "4"                # numSockets 1 × numCores 4
+    capacity.cluster-autoscaler.kubernetes.io/memory: "2Gi"           # memoryMiB 2048
+    capacity.cluster-autoscaler.kubernetes.io/ephemeral-disk: "40Gi"  # bootVolume sizeGb 40
+    capacity.cluster-autoscaler.kubernetes.io/maxPods: "110"          # CA key; "maximum-pods" is ignored
+    # Labels for the from-zero template node. CA 1.35.2 reads MD labels from a path v1beta2 doesn't
+    # have, and CAPMOX publishes no nodeInfo, so without this os/arch are "" on the template.
+    capacity.cluster-autoscaler.kubernetes.io/labels: "node.cluster.x-k8s.io/pool=workers,kubernetes.io/os=linux,kubernetes.io/arch=amd64"
+  labels:
+    cluster.x-k8s.io/cluster-name: lab1
+  name: lab1-workers-pv1-2
+  namespace: default
+spec:
+  clusterName: lab1
+  rollout:
+    strategy:
+      type: RollingUpdate
+      rollingUpdate:
+        maxSurge: 0
+        maxUnavailable: 1
+  selector:
+    matchLabels: null
+  template:
+    metadata:
+      labels:
+        cluster.x-k8s.io/cluster-name: lab1
+        node-role.kubernetes.io/node: ""
+        # CA --balancing-label; node.cluster.x-k8s.io/* is synced to the Node by CAPI
+        node.cluster.x-k8s.io/pool: workers
+    spec:
+      bootstrap:
+        configRef:
+          apiGroup: bootstrap.cluster.x-k8s.io
+          kind: KubeadmConfigTemplate
+          name: lab1-worker
+      clusterName: lab1
+      infrastructureRef:
+        apiGroup: infrastructure.cluster.x-k8s.io
+        kind: ProxmoxMachineTemplate
+        name: lab1-worker-2g-pv1-2
+      version: v1.35.9
+---
+apiVersion: infrastructure.cluster.x-k8s.io/v1alpha2
+kind: ProxmoxMachineTemplate
+metadata:
+  name: lab1-worker-2g-pv1-2
+  namespace: default
+spec:
+  template:
+    spec:
+      # Overrides ProxmoxCluster.allowedNodes ([pv1-1], kept for the control plane): template 9101
+      # lives on pv1-2's local-lvm, and PVE refuses cross-node clones from local storage.
+      allowedNodes:
+      - pv1-2
+      disks:
+        bootVolume:
+          disk: scsi0
+          sizeGb: 40
+      full: true
+      memoryMiB: 2048
+      network:
+        networkDevices:
+        - bridge: vmbr0
+          model: virtio
+          name: net0
+          vlan: 4
+      numCores: 4
+      numSockets: 1
+      sourceNode: pv1-2
+      templateID: 9101
+---
 apiVersion: bootstrap.cluster.x-k8s.io/v1beta2
 kind: KubeadmConfigTemplate
 metadata: