fix: derive hashrate from the counter, not the gauge

miner_hash_rate spikes to impossible values after a restart (1771 MH/s
on a 4090 that peaks near 145) because it is computed per job over a
tiny elapsed time. Hashrate panels now use rate(miner_hashes_total).

Adds node and miner restart annotations so version changes are visible
on every graph, a node version stat, a hash-rate-per-watt panel, and
puts discovered peers on its own axis so connected is legible.

asset/grafana/quantus.json

@@ -9,7 +9,7 @@
   ],
   "timezone": "browser",
   "schemaVersion": 39,
-  "version": 6,
+  "version": 7,
   "refresh": "30s",
   "time": {
     "from": "now-6h",
@@ -575,7 +575,34 @@
             "mode": "palette-classic"
           }
         },
-        "overrides": []
+        "overrides": [
+          {
+            "matcher": {
+              "id": "byName",
+              "options": "discovered"
+            },
+            "properties": [
+              {
+                "id": "custom.axisPlacement",
+                "value": "right"
+              },
+              {
+                "id": "custom.fillOpacity",
+                "value": 0
+              },
+              {
+                "id": "custom.lineStyle",
+                "value": {
+                  "dash": [
+                    6,
+                    6
+                  ],
+                  "fill": "dash"
+                }
+              }
+            ]
+          }
+        ]
       },
       "options": {
         "legend": {
@@ -587,7 +614,8 @@
           "mode": "multi",
           "sort": "desc"
         }
-      }
+      },
+      "description": "Discovered is an order of magnitude larger than connected, so it is on its own axis \u2014 otherwise the connected line reads as flat zero."
     },
     {
       "type": "timeseries",
@@ -948,7 +976,7 @@
     },
     {
       "type": "timeseries",
-      "title": "Hash rate by engine",
+      "title": "Hash rate (counter-derived)",
       "datasource": {
         "type": "prometheus",
         "uid": "prometheus"
@@ -966,8 +994,8 @@
             "uid": "prometheus"
           },
           "editorMode": "code",
-          "expr": "miner_hash_rate{job=\"quantus-miner\"}",
-          "legendFormat": "total",
+          "expr": "sum by (instance) (rate(miner_hashes_total{job=\"quantus-miner\"}[5m]))",
+          "legendFormat": "{{instance}}",
           "range": true,
           "refId": "A"
         },
@@ -977,21 +1005,10 @@
             "uid": "prometheus"
           },
           "editorMode": "code",
-          "expr": "miner_gpu_hash_rate{job=\"quantus-miner\"}",
-          "legendFormat": "gpu",
+          "expr": "sum(rate(miner_hashes_total{job=\"quantus-miner\"}[5m]))",
+          "legendFormat": "fleet",
           "range": true,
           "refId": "B"
-        },
-        {
-          "datasource": {
-            "type": "prometheus",
-            "uid": "prometheus"
-          },
-          "editorMode": "code",
-          "expr": "miner_cpu_hash_rate{job=\"quantus-miner\"}",
-          "legendFormat": "cpu",
-          "range": true,
-          "refId": "C"
         }
       ],
       "fieldConfig": {
@@ -1019,11 +1036,11 @@
           "sort": "desc"
         }
       },
-      "description": "CPU is deliberately 0 (--cpu-workers 0): measured at ~1% of GPU throughput for a couple hundred watts."
+      "description": "Derived from miner_hashes_total, NOT the miner_hash_rate gauge. That gauge spikes to physically impossible values after a restart \u2014 1771 MH/s was observed on a 4090 that peaks near 145 \u2014 because it is computed per job over a tiny elapsed time. A counter derivative cannot do that."
     },
     {
       "type": "timeseries",
-      "title": "Hashes computed (rate)",
+      "title": "Hash rate per watt",
       "datasource": {
         "type": "prometheus",
         "uid": "prometheus"
@@ -1041,7 +1058,7 @@
             "uid": "prometheus"
           },
           "editorMode": "code",
-          "expr": "rate(miner_hashes_total{job=\"quantus-miner\"}[5m])",
+          "expr": "sum by (instance) (rate(miner_hashes_total{job=\"quantus-miner\"}[5m])) / on() group_left sum(nvidia_gpu_power_watts)",
           "legendFormat": "{{instance}}",
           "range": true,
           "refId": "A"
@@ -1057,7 +1074,7 @@
           "color": {
             "mode": "palette-classic"
           },
-          "unit": "Hz"
+          "unit": "none"
         },
         "overrides": []
       },
@@ -1072,7 +1089,7 @@
           "sort": "desc"
         }
       },
-      "description": "Counter derivative \u2014 the honest check that work is being done, independent of the reported gauge."
+      "description": "Efficiency: hashes per second per watt of GPU draw. This is the number a power-limit change should move, and it is comparable across cards where raw hashrate is not."
     },
     {
       "type": "row",
@@ -2450,6 +2467,99 @@
           "sort": "desc"
         }
       }
+    },
+    {
+      "type": "stat",
+      "title": "Node version",
+      "datasource": {
+        "type": "prometheus",
+        "uid": "prometheus"
+      },
+      "gridPos": {
+        "h": 4,
+        "w": 8,
+        "x": 0,
+        "y": 112
+      },
+      "targets": [
+        {
+          "datasource": {
+            "type": "prometheus",
+            "uid": "prometheus"
+          },
+          "editorMode": "code",
+          "expr": "substrate_build_info",
+          "legendFormat": "{{version}}",
+          "range": true,
+          "refId": "A"
+        }
+      ],
+      "description": "Read together with the yellow restart annotations: they mark where this changed.",
+      "fieldConfig": {
+        "defaults": {
+          "color": {
+            "mode": "fixed",
+            "fixedColor": "text"
+          }
+        },
+        "overrides": []
+      },
+      "options": {
+        "colorMode": "none",
+        "graphMode": "none",
+        "textMode": "name",
+        "reduceOptions": {
+          "calcs": [
+            "lastNotNull"
+          ],
+          "fields": "",
+          "values": false
+        }
+      }
     }
-  ]
+  ],
+  "annotations": {
+    "list": [
+      {
+        "builtIn": 1,
+        "datasource": {
+          "type": "grafana",
+          "uid": "-- Grafana --"
+        },
+        "enable": true,
+        "hide": true,
+        "iconColor": "rgba(0, 211, 255, 1)",
+        "name": "Annotations & Alerts",
+        "type": "dashboard"
+      },
+      {
+        "datasource": {
+          "type": "prometheus",
+          "uid": "prometheus"
+        },
+        "enable": true,
+        "hide": false,
+        "iconColor": "yellow",
+        "name": "node restart",
+        "expr": "changes(substrate_process_start_time_seconds[2m]) > 0",
+        "step": "1m",
+        "titleFormat": "node restart",
+        "textFormat": "a version or config change \u2014 treat data either side as separate populations"
+      },
+      {
+        "datasource": {
+          "type": "prometheus",
+          "uid": "prometheus"
+        },
+        "enable": true,
+        "hide": false,
+        "iconColor": "orange",
+        "name": "miner restart",
+        "expr": "resets(miner_hashes_total[2m]) > 0",
+        "step": "1m",
+        "titleFormat": "miner restart",
+        "textFormat": "counter reset \u2014 hashrate readings near this line are unreliable"
+      }
+    ]
+  }
 }