refactor(k8s): Reorganize into 5-layer structure with production kustomizations

2026-08-18 15:08:01 -07:00
parent 563f720d09
commit 0f30d77288
214 changed files with 405 additions and 885 deletions
@@ -0,0 +1,55 @@
+# k8s/monitoring/dashboards/control-plane-logs.yaml
+# Surfaces controller/control-plane logs that are already in Loki today
+# (Promtail scrapes every namespace with no filter) — this dashboard is the
+# "make it visible" piece, not new log collection.
+apiVersion: v1
+kind: ConfigMap
+metadata:
+  name: control-plane-logs-dashboard
+  namespace: logging
+  labels:
+    grafana_dashboard: "1"
+data:
+  control-plane-logs.json: |
+    {
+      "title": "Cluster Control Plane & Controllers (Logs)",
+      "uid": "control-plane-logs",
+      "schemaVersion": 39,
+      "timezone": "browser",
+      "time": { "from": "now-1h", "to": "now" },
+      "refresh": "30s",
+      "panels": [
+        {
+          "id": 1,
+          "title": "Error rate by namespace",
+          "type": "timeseries",
+          "gridPos": { "h": 6, "w": 24, "x": 0, "y": 0 },
+          "datasource": { "type": "loki", "uid": "loki" },
+          "targets": [
+            {
+              "expr": "sum by (namespace) (count_over_time({namespace=~\"kube-system|cert-manager|ingress-nginx|longhorn-system\"} |= \"error\" [5m]))"
+            }
+          ]
+        },
+        {
+          "id": 2,
+          "title": "Control plane (kube-apiserver, controller-manager, scheduler)",
+          "type": "logs",
+          "gridPos": { "h": 10, "w": 24, "x": 0, "y": 6 },
+          "datasource": { "type": "loki", "uid": "loki" },
+          "targets": [
+            { "expr": "{namespace=\"kube-system\"}" }
+          ]
+        },
+        {
+          "id": 3,
+          "title": "Cluster add-ons (cert-manager, ingress-nginx, longhorn)",
+          "type": "logs",
+          "gridPos": { "h": 10, "w": 24, "x": 0, "y": 16 },
+          "datasource": { "type": "loki", "uid": "loki" },
+          "targets": [
+            { "expr": "{namespace=~\"cert-manager|ingress-nginx|longhorn-system\"}" }
+          ]
+        }
+      ]
+    }