titan-iac/services/monitoring/grafana-dashboard-sre.yaml

# services/monitoring/grafana-dashboard-sre.yaml
apiVersion: v1
kind: ConfigMap
metadata:
  name: grafana-dashboard-sre
  labels:
    grafana_dashboard: "1"
data:
  atlas-sre-overview.json: |
    {
      "annotations": {
        "list": [
          {
            "builtIn": 1,
            "datasource": {
              "type": "datasource",
              "uid": "grafana"
            },
            "enable": true,
            "hide": true,
            "iconColor": "rgba(0, 211, 255, 1)",
            "name": "Annotations & Alerts",
              "type": "dashboard"
          }
        ]
      },
      "editable": true,
      "fiscalYearStartMonth": 0,
      "graphTooltip": 0,
      "links": [],
      "panels": [
        {
          "datasource": {
            "type": "prometheus",
            "uid": "atlas-vm"
          },
          "description": "Percentage of Ready nodes.",
          "fieldConfig": {
            "defaults": {
              "color": {
                "mode": "continuous"
              },
              "mappings": [],
              "max": 100,
              "min": 0,
              "thresholds": {
                "mode": "percentage",
                "steps": [
                  {
                    "color": "red",
                    "value": null
                  },
                  {
                    "color": "green",
                    "value": 90
                  }
                ]
              },
              "unit": "percent"
            },
            "overrides": []
          },
          "gridPos": {
            "h": 7,
            "w": 6,
            "x": 0,
            "y": 0
          },
          "id": 10,
          "options": {
            "colorMode": "value",
            "graphMode": "none",
            "justifyMode": "center",
            "orientation": "auto",
            "reduceOptions": {
              "calcs": [
                "lastNotNull"
              ],
              "fields": "",
              "values": false
            }
          },
          "targets": [
            {
              "datasource": {
                "type": "prometheus",
                "uid": "atlas-vm"
              },
              "expr": "avg(kube_node_status_condition{condition=\"Ready\",status=\"true\"}) * 100",
              "refId": "A"
            }
          ],
          "title": "Ready nodes",
          "type": "stat"
        },
        {
          "datasource": {
            "type": "prometheus",
            "uid": "atlas-vm"
          },
          "gridPos": {
            "h": 7,
            "w": 6,
            "x": 6,
            "y": 0
          },
          "id": 11,
          "options": {
            "legend": {
              "displayMode": "table",
              "placement": "bottom"
            },
            "tooltip": {
              "mode": "multi"
            }
          },
          "targets": [
            {
              "datasource": {
                "type": "prometheus",
                "uid": "atlas-vm"
              },
              "expr": "sum by (node)(node_filesystem_avail_bytes{mountpoint=\"/\"})",
              "legendFormat": "{{node}}",
              "refId": "A"
            }
          ],
          "title": "Free root filesystem bytes",
          "type": "timeseries"
        },
        {
          "datasource": {
            "type": "prometheus",
            "uid": "atlas-vm"
          },
          "gridPos": {
            "h": 10,
            "w": 12,
            "x": 0,
            "y": 7
          },
          "id": 12,
          "options": {
            "legend": {
              "calcs": [],
              "displayMode": "list",
              "placement": "bottom"
            },
            "tooltip": {
              "mode": "single"
            }
          },
          "targets": [
            {
              "datasource": {
                "type": "prometheus",
                "uid": "atlas-vm"
              },
              "expr": "sum(rate(container_cpu_usage_seconds_total{namespace=\"crypto\",container!=\"\"}[5m])) by (pod)",
              "legendFormat": "{{pod}}",
              "refId": "A"
            }
          ],
          "title": "Crypto namespace CPU usage",
          "type": "timeseries"
        },
        {
          "datasource": {
            "type": "prometheus",
            "uid": "atlas-vm"
          },
          "gridPos": {
            "h": 9,
            "w": 12,
            "x": 0,
            "y": 17
          },
          "id": 13,
          "options": {
            "displayMode": "gradient",
            "orientation": "horizontal",
            "reduceOptions": {
              "calcs": [
                "lastNotNull"
              ],
              "fields": "",
              "values": false
            },
            "showUnfilled": false
          },
          "targets": [
            {
              "datasource": {
                "type": "prometheus",
                "uid": "atlas-vm"
              },
              "expr": "count(sum(kube_pod_status_phase{phase=\"Failed\"}) by (namespace))",
              "legendFormat": "",
              "refId": "A"
            }
          ],
          "title": "Namespaces with failed pods",
          "type": "bargauge"
        }
      ],
      "schemaVersion": 39,
      "style": "dark",
      "tags": [
        "atlas",
        "sre"
      ],
      "templating": {
        "list": []
      },
      "time": {
        "from": "now-12h",
        "to": "now"
      },
      "timepicker": {},
      "title": "Atlas SRE Overview",
      "uid": "atlas-sre",
      "version": 1
    }