Add HBS monitoring panels (#950)

* Add HBS monitoring panels * Add 2 more panels
m-lab · Aug 31, 2022 · 25870a3 · 25870a3
1 parent 613b340
commit 25870a3
Showing 1 changed file with 262 additions and 6 deletions.
diff --git a/config/federation/grafana/dashboards/Locate_Service.json b/config/federation/grafana/dashboards/Locate_Service.json
@@ -21,7 +21,7 @@
   "editable": true,
   "fiscalYearStartMonth": 0,
   "graphTooltip": 0,
-  "iteration": 1659037622179,
+  "iteration": 1661804115489,
   "links": [],
   "liveNow": false,
   "panels": [
@@ -391,13 +391,249 @@
       "title": "Heartbeat Connection Request Rate (5 min)",
       "type": "timeseries"
     },
+    {
+      "datasource": {
+        "type": "prometheus",
+        "uid": "${platformdatasource}"
+      },
+      "description": "Increase in Requests to the Kubernetes API Server grouped by status.",
+      "fieldConfig": {
+        "defaults": {
+          "color": {
+            "mode": "palette-classic"
+          },
+          "custom": {
+            "axisLabel": "",
+            "axisPlacement": "auto",
+            "barAlignment": 0,
+            "drawStyle": "line",
+            "fillOpacity": 0,
+            "gradientMode": "none",
+            "hideFrom": {
+              "legend": false,
+              "tooltip": false,
+              "viz": false
+            },
+            "lineInterpolation": "linear",
+            "lineWidth": 1,
+            "pointSize": 5,
+            "scaleDistribution": {
+              "type": "linear"
+            },
+            "showPoints": "auto",
+            "spanNulls": false,
+            "stacking": {
+              "group": "A",
+              "mode": "none"
+            },
+            "thresholdsStyle": {
+              "mode": "off"
+            }
+          },
+          "mappings": [],
+          "thresholds": {
+            "mode": "absolute",
+            "steps": [
+              {
+                "color": "green",
+                "value": null
+              },
+              {
+                "color": "red",
+                "value": 80
+              }
+            ]
+          }
+        },
+        "overrides": []
+      },
+      "gridPos": {
+        "h": 8,
+        "w": 9,
+        "x": 0,
+        "y": 19
+      },
+      "id": 22,
+      "options": {
+        "legend": {
+          "calcs": [],
+          "displayMode": "list",
+          "placement": "bottom"
+        },
+        "tooltip": {
+          "mode": "single"
+        }
+      },
+      "targets": [
+        {
+          "datasource": {
+            "type": "prometheus",
+            "uid": "${platformdatasource}"
+          },
+          "exemplar": true,
+          "expr": "sum by (status) (increase(heartbeat_kubernetes_requests_total[10m]))",
+          "interval": "",
+          "legendFormat": "{{status}}",
+          "refId": "A"
+        }
+      ],
+      "title": "Kubernetes Requests",
+      "type": "timeseries"
+    },
+    {
+      "datasource": {
+        "type": "prometheus",
+        "uid": "${platformdatasource}"
+      },
+      "description": "Increase in number of port checks grouped by status.",
+      "fieldConfig": {
+        "defaults": {
+          "color": {
+            "mode": "palette-classic"
+          },
+          "custom": {
+            "axisLabel": "",
+            "axisPlacement": "auto",
+            "barAlignment": 0,
+            "drawStyle": "line",
+            "fillOpacity": 0,
+            "gradientMode": "none",
+            "hideFrom": {
+              "legend": false,
+              "tooltip": false,
+              "viz": false
+            },
+            "lineInterpolation": "linear",
+            "lineWidth": 1,
+            "pointSize": 5,
+            "scaleDistribution": {
+              "type": "linear"
+            },
+            "showPoints": "auto",
+            "spanNulls": false,
+            "stacking": {
+              "group": "A",
+              "mode": "none"
+            },
+            "thresholdsStyle": {
+              "mode": "off"
+            }
+          },
+          "mappings": [],
+          "thresholds": {
+            "mode": "absolute",
+            "steps": [
+              {
+                "color": "green",
+                "value": null
+              },
+              {
+                "color": "red",
+                "value": 80
+              }
+            ]
+          }
+        },
+        "overrides": []
+      },
+      "gridPos": {
+        "h": 8,
+        "w": 8,
+        "x": 9,
+        "y": 19
+      },
+      "id": 24,
+      "options": {
+        "legend": {
+          "calcs": [],
+          "displayMode": "list",
+          "placement": "bottom"
+        },
+        "tooltip": {
+          "mode": "single"
+        }
+      },
+      "targets": [
+        {
+          "datasource": {
+            "type": "prometheus",
+            "uid": "${platformdatasource}"
+          },
+          "exemplar": true,
+          "expr": "sum by(status)(increase(heartbeat_port_checks_total[10m]))",
+          "interval": "",
+          "legendFormat": "{{status}}",
+          "refId": "A"
+        }
+      ],
+      "title": "Port Checks",
+      "type": "timeseries"
+    },
+    {
+      "cards": {},
+      "color": {
+        "cardColor": "#b4ff00",
+        "colorScale": "sqrt",
+        "colorScheme": "interpolateOranges",
+        "exponent": 0.5,
+        "mode": "spectrum"
+      },
+      "dataFormat": "tsbuckets",
+      "datasource": {
+        "type": "prometheus",
+        "uid": "${platformdatasource}"
+      },
+      "gridPos": {
+        "h": 8,
+        "w": 9,
+        "x": 0,
+        "y": 27
+      },
+      "heatmap": {},
+      "hideZeroBuckets": true,
+      "highlightCards": true,
+      "id": 20,
+      "legend": {
+        "show": true
+      },
+      "reverseYBuckets": false,
+      "targets": [
+        {
+          "datasource": {
+            "type": "prometheus",
+            "uid": "${platformdatasource}"
+          },
+          "exemplar": true,
+          "expr": "sum by(le) (increase(heartbeat_kubernetes_request_time_histogram_bucket{}[5m]))",
+          "format": "heatmap",
+          "interval": "5m",
+          "legendFormat": "{{le}}",
+          "refId": "A"
+        }
+      ],
+      "title": "Kubernetes API Server Request Time",
+      "tooltip": {
+        "show": true,
+        "showHistogram": true
+      },
+      "type": "heatmap",
+      "xAxis": {
+        "show": true
+      },
+      "yAxis": {
+        "format": "short",
+        "logBase": 1,
+        "show": true
+      },
+      "yBucketBound": "auto"
+    },
     {
       "collapsed": false,
       "gridPos": {
         "h": 1,
         "w": 24,
         "x": 0,
-        "y": 19
+        "y": 35
       },
       "id": 16,
       "panels": [],
@@ -464,7 +700,7 @@
         "h": 8,
         "w": 9,
         "x": 0,
-        "y": 20
+        "y": 36
       },
       "id": 6,
       "options": {
@@ -501,12 +737,13 @@
     "list": [
       {
         "current": {
-          "selected": false,
+          "selected": true,
           "text": "Prometheus (mlab-sandbox)",
           "value": "Prometheus (mlab-sandbox)"
         },
         "hide": 0,
         "includeAll": false,
+        "label": "Prometheus Datasource",
         "multi": false,
         "name": "datasource",
         "options": [],
@@ -516,17 +753,36 @@
         "regex": "/Prometheus \\(/",
         "skipUrlSync": false,
         "type": "datasource"
+      },
+      {
+        "current": {
+          "selected": false,
+          "text": "Platform Cluster (mlab-sandbox)",
+          "value": "Platform Cluster (mlab-sandbox)"
+        },
+        "hide": 0,
+        "includeAll": false,
+        "label": "Platform Datasource",
+        "multi": false,
+        "name": "platformdatasource",
+        "options": [],
+        "query": "prometheus",
+        "queryValue": "",
+        "refresh": 1,
+        "regex": "/Platform Cluster \\(/",
+        "skipUrlSync": false,
+        "type": "datasource"
       }
     ]
   },
   "time": {
-    "from": "now-24h",
+    "from": "now-7d",
     "to": "now"
   },
   "timepicker": {},
   "timezone": "",
   "title": "Locate Service",
   "uid": "8O9tInk4k",
-  "version": 14,
+  "version": 24,
   "weekStart": ""
 }