From 0d61ebc8771c2f31729388dc89f3411a1cc219d3 Mon Sep 17 00:00:00 2001 From: Ian Fijolek Date: Fri, 29 Jul 2022 13:01:59 -0700 Subject: [PATCH] Add Nomad dashboard to grafana --- .../dashboards/default/nomad.json | 1105 +++++++++++++++++ 1 file changed, 1105 insertions(+) create mode 100644 nomad/metrics/grafana/provisioning/dashboards/default/nomad.json diff --git a/nomad/metrics/grafana/provisioning/dashboards/default/nomad.json b/nomad/metrics/grafana/provisioning/dashboards/default/nomad.json new file mode 100644 index 0000000..897d5f3 --- /dev/null +++ b/nomad/metrics/grafana/provisioning/dashboards/default/nomad.json @@ -0,0 +1,1105 @@ +{ + "annotations": { + "list": [ + { + "builtIn": 1, + "datasource": "-- Grafana --", + "enable": true, + "hide": true, + "iconColor": "rgba(0, 211, 255, 1)", + "name": "Annotations & Alerts", + "type": "dashboard" + } + ] + }, + "description": "Nomad built-in telemetry https://github.com/dereknex/nomad-grafana", + "editable": true, + "gnetId": 6278, + "graphTooltip": 1, + "id": 8, + "iteration": 1659118948293, + "links": [], + "panels": [ + { + "datasource": null, + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 0 + }, + "id": 24, + "title": "Allocations", + "type": "row" + }, + { + "cacheTimeout": null, + "colorBackground": false, + "colorValue": false, + "colors": [ + "#d44a3a", + "rgba(237, 129, 40, 0.89)", + "#299c46" + ], + "datasource": "Prometheus", + "fieldConfig": { + "defaults": { + "custom": {} + }, + "overrides": [] + }, + "format": "percentunit", + "gauge": { + "maxValue": 1, + "minValue": 0, + "show": true, + "thresholdLabels": false, + "thresholdMarkers": true + }, + "gridPos": { + "h": 5, + "w": 5, + "x": 0, + "y": 1 + }, + "id": 33, + "interval": null, + "links": [], + "mappingType": 1, + "mappingTypes": [ + { + "name": "value to text", + "value": 1 + }, + { + "name": "range to text", + "value": 2 + } + ], + "maxDataPoints": 100, + "nullPointMode": "connected", + "nullText": null, + "postfix": "", + "postfixFontSize": "50%", + "prefix": "", + "prefixFontSize": "50%", + "rangeMaps": [ + { + "from": "null", + "text": "N/A", + "to": "null" + } + ], + "repeat": "host", + "repeatDirection": "v", + "scopedVars": { + "host": { + "selected": true, + "text": "n2", + "value": "n2" + } + }, + "sparkline": { + "fillColor": "rgba(31, 118, 189, 0.18)", + "full": false, + "lineColor": "rgb(31, 120, 193)", + "show": true + }, + "tableColumn": "", + "targets": [ + { + "expr": "nomad_client_allocated_cpu{datacenter=~\"$datacenter\", host=~\"$host\"}/(nomad_client_unallocated_cpu{datacenter=~\"$datacenter\", host=~\"$host\"}+nomad_client_allocated_cpu{datacenter=~\"$datacenter\", host=~\"$host\"})", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "", + "refId": "A" + } + ], + "thresholds": "", + "title": "CPU", + "type": "singlestat", + "valueFontSize": "80%", + "valueMaps": [ + { + "op": "=", + "text": "N/A", + "value": "null" + } + ], + "valueName": "current" + }, + { + "cacheTimeout": null, + "colorBackground": false, + "colorValue": false, + "colors": [ + "#d44a3a", + "rgba(237, 129, 40, 0.89)", + "#299c46" + ], + "datasource": "Prometheus", + "fieldConfig": { + "defaults": { + "custom": {} + }, + "overrides": [] + }, + "format": "percentunit", + "gauge": { + "maxValue": 1, + "minValue": 0, + "show": true, + "thresholdLabels": false, + "thresholdMarkers": true + }, + "gridPos": { + "h": 5, + "w": 5, + "x": 5, + "y": 1 + }, + "id": 40, + "interval": null, + "links": [], + "mappingType": 1, + "mappingTypes": [ + { + "name": "value to text", + "value": 1 + }, + { + "name": "range to text", + "value": 2 + } + ], + "maxDataPoints": 100, + "nullPointMode": "connected", + "nullText": null, + "postfix": "", + "postfixFontSize": "50%", + "prefix": "", + "prefixFontSize": "50%", + "rangeMaps": [ + { + "from": "null", + "text": "N/A", + "to": "null" + } + ], + "repeat": "host", + "repeatDirection": "v", + "scopedVars": { + "host": { + "selected": true, + "text": "n2", + "value": "n2" + } + }, + "sparkline": { + "fillColor": "rgba(31, 118, 189, 0.18)", + "full": false, + "lineColor": "rgb(31, 120, 193)", + "show": true + }, + "tableColumn": "", + "targets": [ + { + "expr": "nomad_client_allocated_memory{datacenter=~\"$datacenter\", host=~\"$host\"}/(nomad_client_unallocated_memory{datacenter=~\"$datacenter\", host=~\"$host\"}+nomad_client_allocated_memory{datacenter=~\"$datacenter\", host=~\"$host\"})", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "", + "refId": "A" + } + ], + "thresholds": "", + "title": "Memory", + "type": "singlestat", + "valueFontSize": "80%", + "valueMaps": [ + { + "op": "=", + "text": "N/A", + "value": "null" + } + ], + "valueName": "current" + }, + { + "cacheTimeout": null, + "colorBackground": false, + "colorValue": false, + "colors": [ + "#d44a3a", + "rgba(237, 129, 40, 0.89)", + "#299c46" + ], + "datasource": "Prometheus", + "decimals": 2, + "fieldConfig": { + "defaults": { + "custom": {} + }, + "overrides": [] + }, + "format": "percentunit", + "gauge": { + "maxValue": 1, + "minValue": 0, + "show": true, + "thresholdLabels": false, + "thresholdMarkers": true + }, + "gridPos": { + "h": 5, + "w": 5, + "x": 10, + "y": 1 + }, + "id": 48, + "interval": null, + "links": [], + "mappingType": 1, + "mappingTypes": [ + { + "name": "value to text", + "value": 1 + }, + { + "name": "range to text", + "value": 2 + } + ], + "maxDataPoints": 100, + "nullPointMode": "connected", + "nullText": null, + "postfix": "", + "postfixFontSize": "50%", + "prefix": "", + "prefixFontSize": "50%", + "rangeMaps": [ + { + "from": "null", + "text": "N/A", + "to": "null" + } + ], + "repeat": "host", + "repeatDirection": "v", + "scopedVars": { + "host": { + "selected": true, + "text": "n2", + "value": "n2" + } + }, + "sparkline": { + "fillColor": "rgba(31, 118, 189, 0.18)", + "full": false, + "lineColor": "rgb(31, 120, 193)", + "show": true + }, + "tableColumn": "", + "targets": [ + { + "expr": "nomad_client_allocated_disk{datacenter=~\"$datacenter\", host=~\"$host\"}/(nomad_client_unallocated_disk{datacenter=~\"$datacenter\", host=~\"$host\"}+nomad_client_allocated_disk{datacenter=~\"$datacenter\", host=~\"$host\"})", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "", + "refId": "A" + } + ], + "thresholds": "", + "title": "Disk", + "type": "singlestat", + "valueFontSize": "80%", + "valueMaps": [ + { + "op": "=", + "text": "N/A", + "value": "null" + } + ], + "valueName": "current" + }, + { + "backgroundColor": "rgba(128,128,128,0.1)", + "colorMaps": [ + { + "color": "#CCC", + "text": "N/A" + } + ], + "crosshairColor": "#8F070C", + "datasource": "Prometheus", + "display": "timeline", + "expandFromQueryS": 0, + "extendLastValue": true, + "fieldConfig": { + "defaults": { + "custom": {} + }, + "overrides": [] + }, + "gridPos": { + "h": 5, + "w": 9, + "x": 15, + "y": 1 + }, + "highlightOnMouseover": false, + "id": 58, + "legendSortBy": "-ms", + "lineColor": "rgba(0,0,0,0.1)", + "links": [], + "metricNameColor": "#000000", + "rangeMaps": [ + { + "from": "null", + "text": "N/A", + "to": "null" + } + ], + "repeat": "host", + "repeatDirection": "v", + "rowHeight": 24, + "scopedVars": { + "host": { + "selected": true, + "text": "n2", + "value": "n2" + } + }, + "showDistinctCount": false, + "showLegend": false, + "showLegendNames": false, + "showLegendPercent": true, + "showLegendValues": false, + "showTimeAxis": true, + "showTransitionCount": false, + "targets": [ + { + "expr": "nomad_client_allocations_migrating{datacenter=~\"$datacenter\", host=~\"$host\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "Migrating", + "refId": "A" + }, + { + "expr": "nomad_client_allocations_blocked{datacenter=~\"$datacenter\", host=~\"$host\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "Blocked", + "refId": "B" + }, + { + "expr": "nomad_client_allocations_pending{datacenter=~\"$datacenter\", host=~\"$host\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "Pending", + "refId": "C" + }, + { + "expr": "nomad_client_allocations_running{datacenter=~\"$datacenter\", host=~\"$host\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "Running", + "refId": "D" + }, + { + "expr": "nomad_client_allocations_terminal{datacenter=~\"$datacenter\", host=~\"$host\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "Terminal", + "refId": "E" + } + ], + "textSize": 14, + "textSizeTime": 12, + "timeTextColor": "#d8d9da", + "title": "Summary", + "type": "natel-discrete-panel", + "units": "short", + "valueMaps": [ + { + "op": "=", + "text": "N/A", + "value": "null" + } + ], + "valueTextColor": "#000000", + "writeAllValues": false, + "writeLastValue": true, + "writeMetricNames": true + }, + { + "datasource": null, + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 6 + }, + "id": 2, + "title": "Host Status", + "type": "row" + }, + { + "cacheTimeout": null, + "colorBackground": false, + "colorValue": false, + "colors": [ + "#299c46", + "rgba(237, 129, 40, 0.89)", + "#d44a3a" + ], + "datasource": "Prometheus", + "decimals": 1, + "fieldConfig": { + "defaults": { + "custom": {} + }, + "overrides": [] + }, + "format": "dtdurations", + "gauge": { + "maxValue": 100, + "minValue": 0, + "show": false, + "thresholdLabels": false, + "thresholdMarkers": true + }, + "gridPos": { + "h": 5, + "w": 4, + "x": 0, + "y": 7 + }, + "id": 4, + "interval": null, + "links": [], + "mappingType": 1, + "mappingTypes": [ + { + "name": "value to text", + "value": 1 + }, + { + "name": "range to text", + "value": 2 + } + ], + "maxDataPoints": 100, + "nullPointMode": "connected", + "nullText": null, + "postfix": "", + "postfixFontSize": "50%", + "prefix": "", + "prefixFontSize": "50%", + "rangeMaps": [ + { + "from": "null", + "text": "N/A", + "to": "null" + } + ], + "repeat": "host", + "repeatDirection": "v", + "scopedVars": { + "host": { + "selected": true, + "text": "n2", + "value": "n2" + } + }, + "sparkline": { + "fillColor": "rgba(31, 118, 189, 0.18)", + "full": false, + "lineColor": "rgb(31, 120, 193)", + "show": false + }, + "tableColumn": "", + "targets": [ + { + "expr": "nomad_client_uptime{datacenter=~\"$datacenter\", host=~\"$host\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "", + "refId": "A" + } + ], + "thresholds": "", + "title": "Uptime($host)", + "type": "singlestat", + "valueFontSize": "50%", + "valueMaps": [ + { + "op": "=", + "text": "N/A", + "value": "null" + } + ], + "valueName": "current" + }, + { + "cacheTimeout": null, + "colorBackground": false, + "colorValue": false, + "colors": [ + "#cca300", + "#508642", + "#508642" + ], + "datasource": "Prometheus", + "fieldConfig": { + "defaults": { + "custom": {} + }, + "overrides": [] + }, + "format": "percent", + "gauge": { + "maxValue": 100, + "minValue": 0, + "show": true, + "thresholdLabels": false, + "thresholdMarkers": true + }, + "gridPos": { + "h": 5, + "w": 5, + "x": 4, + "y": 7 + }, + "id": 7, + "interval": null, + "links": [], + "mappingType": 1, + "mappingTypes": [ + { + "name": "value to text", + "value": 1 + }, + { + "name": "range to text", + "value": 2 + } + ], + "maxDataPoints": 100, + "nullPointMode": "connected", + "nullText": null, + "postfix": "", + "postfixFontSize": "50%", + "prefix": "", + "prefixFontSize": "50%", + "rangeMaps": [ + { + "from": "null", + "text": "N/A", + "to": "null" + } + ], + "repeat": "host", + "repeatDirection": "v", + "scopedVars": { + "host": { + "selected": true, + "text": "n2", + "value": "n2" + } + }, + "sparkline": { + "fillColor": "rgba(31, 118, 189, 0.18)", + "full": false, + "lineColor": "rgb(31, 120, 193)", + "show": true + }, + "tableColumn": "", + "targets": [ + { + "expr": "nomad_client_host_cpu_idle{datacenter=\"$datacenter\", host=\"$host\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "", + "refId": "A" + } + ], + "thresholds": "20,80", + "title": "CPU Idle($host)", + "type": "singlestat", + "valueFontSize": "80%", + "valueMaps": [ + { + "op": "=", + "text": "N/A", + "value": "null" + } + ], + "valueName": "current" + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "Prometheus", + "fieldConfig": { + "defaults": { + "custom": {} + }, + "overrides": [] + }, + "fill": 1, + "fillGradient": 0, + "gridPos": { + "h": 5, + "w": 5, + "x": 9, + "y": 7 + }, + "hiddenSeries": false, + "id": 11, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.6", + "pointradius": 5, + "points": false, + "renderer": "flot", + "repeat": "host", + "repeatDirection": "v", + "scopedVars": { + "host": { + "selected": true, + "text": "n2", + "value": "n2" + } + }, + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "nomad_client_host_memory_total{datacenter=~\"$datacenter\", host=~\"$host\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "Total", + "refId": "B" + }, + { + "expr": "nomad_client_host_memory_free{datacenter=~\"$datacenter\", host=~\"$host\"}", + "format": "time_series", + "instant": false, + "interval": "", + "intervalFactor": 1, + "legendFormat": "Free", + "refId": "A" + }, + { + "expr": "nomad_client_host_memory_used{datacenter=~\"$datacenter\", host=~\"$host\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "Used", + "refId": "C" + }, + { + "expr": "nomad_client_host_memory_available{datacenter=~\"$datacenter\", host=~\"$host\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "Available", + "refId": "D" + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Memory ($host)", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "decbytes", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": true + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "Prometheus", + "fieldConfig": { + "defaults": { + "custom": {} + }, + "overrides": [] + }, + "fill": 1, + "fillGradient": 0, + "gridPos": { + "h": 5, + "w": 5, + "x": 14, + "y": 7 + }, + "hiddenSeries": false, + "id": 13, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.6", + "pointradius": 5, + "points": false, + "renderer": "flot", + "repeat": "host", + "repeatDirection": "v", + "scopedVars": { + "host": { + "selected": true, + "text": "n2", + "value": "n2" + } + }, + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "max(nomad_client_host_disk_size{datacenter=~\"$datacenter\", host=~\"$host\"}) by (disk)", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "Total - {{disk}}", + "refId": "B" + }, + { + "expr": "avg(nomad_client_host_disk_available{datacenter=~\"$datacenter\", host=~\"$host\"}) by (disk)", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "Available - {{disk}}", + "refId": "A" + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Disk Usage ($host)", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "decimals": 0, + "format": "decbytes", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "format": "decbytes", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "Prometheus", + "fieldConfig": { + "defaults": { + "custom": {} + }, + "overrides": [] + }, + "fill": 1, + "fillGradient": 0, + "gridPos": { + "h": 5, + "w": 5, + "x": 19, + "y": 7 + }, + "hiddenSeries": false, + "id": 15, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.6", + "pointradius": 5, + "points": false, + "renderer": "flot", + "repeat": "host", + "repeatDirection": "v", + "scopedVars": { + "host": { + "selected": true, + "text": "n2", + "value": "n2" + } + }, + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "avg(nomad_client_host_disk_inodes_percent{datacenter=~\"$datacenter\", host=~\"$host\"}) by (disk)", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "{{disk}}", + "refId": "A" + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Disk Inodes ($host)", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "decimals": null, + "format": "percent", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": true + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + } + ], + "refresh": "1m", + "schemaVersion": 26, + "style": "dark", + "tags": [ + "nomad", + "host", + "cluster" + ], + "templating": { + "list": [ + { + "allValue": null, + "current": { + "selected": false, + "text": "dc1", + "value": "dc1" + }, + "datasource": "Prometheus", + "definition": "", + "error": null, + "hide": 0, + "includeAll": false, + "label": "DC", + "multi": false, + "name": "datacenter", + "options": [], + "query": "label_values(nomad_client_uptime, datacenter)", + "refresh": 1, + "regex": "", + "skipUrlSync": false, + "sort": 0, + "tagValuesQuery": "", + "tags": [], + "tagsQuery": "", + "type": "query", + "useTags": false + }, + { + "allValue": null, + "current": { + "selected": true, + "tags": [], + "text": [ + "n2" + ], + "value": [ + "n2" + ] + }, + "datasource": "Prometheus", + "definition": "", + "error": null, + "hide": 0, + "includeAll": true, + "label": "Host", + "multi": true, + "name": "host", + "options": [], + "query": "label_values(nomad_client_uptime{datacenter=~\"$datacenter\"}, host)", + "refresh": 1, + "regex": "", + "skipUrlSync": false, + "sort": 0, + "tagValuesQuery": "", + "tags": [], + "tagsQuery": "", + "type": "query", + "useTags": false + } + ] + }, + "time": { + "from": "now-15m", + "to": "now" + }, + "timepicker": { + "refresh_intervals": [ + "5s", + "10s", + "30s", + "1m", + "5m", + "15m", + "30m", + "1h", + "2h", + "1d" + ], + "time_options": [ + "5m", + "15m", + "1h", + "6h", + "12h", + "24h", + "2d", + "7d", + "30d" + ] + }, + "timezone": "", + "title": "Nomad Cluster", + "uid": "CiP3mZVik", + "version": 5 +}