commit 1ac93ba11bd57b73989ae78543902cf0b9fc5483 Author: nyyu Date: Fri Apr 22 20:16:39 2022 +0200 init diff --git a/.gitignore b/.gitignore new file mode 100644 index 0000000..a9babce --- /dev/null +++ b/.gitignore @@ -0,0 +1,3 @@ +.env +alertmanager.yml +data \ No newline at end of file diff --git a/README.md b/README.md new file mode 100644 index 0000000..c4d5f98 --- /dev/null +++ b/README.md @@ -0,0 +1,26 @@ +# Docker monitoring stack + +## Includes +* prometheus +* grafana +* alertmanager +* node-exporter +* blackbox-exporter +* cadvisor + +## Prerequisites +* docker +* docker compose + +## Steps + +1. Create .env file + > EXTERNAL_IP=XXX.XXX.XXX.XXX + > + > GRAFANA_PASSWORD=SECURE_PASSWORD + +2. Copy alertmanager.tmpl to alertmanager.yml + * Set telegram bot_token and chat_id + +3. Deploy the stack + > docker compose up -d \ No newline at end of file diff --git a/conf/alertmanager/alertmanager.tmpl b/conf/alertmanager/alertmanager.tmpl new file mode 100644 index 0000000..18e76d4 --- /dev/null +++ b/conf/alertmanager/alertmanager.tmpl @@ -0,0 +1,17 @@ +global: +templates: + - /etc/alertmanager/templates/*.tmpl +route: + group_wait: 10s + group_interval: 30s + repeat_interval: 30m + group_by: [alertname, instance] + receiver: telegram +receivers: + - name: telegram + telegram_configs: + - bot_token: BOT_TOKEN + chat_id: CHAT_ID + api_url: https://api.telegram.org + parse_mode: 'HTML' + message: '{{ template "telegram.custom.message" .}}' \ No newline at end of file diff --git a/conf/alertmanager/templates/telegram.tmpl b/conf/alertmanager/templates/telegram.tmpl new file mode 100644 index 0000000..67adcce --- /dev/null +++ b/conf/alertmanager/templates/telegram.tmpl @@ -0,0 +1,11 @@ +{{ define "telegram.custom.message" }} +{{ range .Alerts }} +{{ if eq .Status "firing"}}🔥{{ .Labels.alertname }}🔥{{ else }}👌{{ .Labels.alertname }}👌{{ end }} +Labels:{{ range $key, $value := .Labels }}{{ if ne $key "alertname" }} + - {{ $key }}: {{ $value }}{{ end }}{{ end }} +Annotations:{{ range $key, $value := .Annotations }} + - {{ $key }}: {{ reReplaceAll "(?s)LABELS = (.*)" "" $value }}{{ end }} +Start: {{ .StartsAt }}{{ if eq .Status "resolved"}} +Ended: {{ .EndsAt }}{{ end }} +{{ end }} +{{ end }} \ No newline at end of file diff --git a/conf/grafana/provisioning/dashboards/Node Exporter Full.json b/conf/grafana/provisioning/dashboards/Node Exporter Full.json new file mode 100644 index 0000000..55c12bb --- /dev/null +++ b/conf/grafana/provisioning/dashboards/Node Exporter Full.json @@ -0,0 +1,14116 @@ +{ + "__inputs": [ + { + "name": "DS_PROMETHEUS", + "label": "Prometheus", + "description": "", + "type": "datasource", + "pluginId": "prometheus", + "pluginName": "Prometheus" + } + ], + "__requires": [ + { + "type": "panel", + "id": "gauge", + "name": "Gauge", + "version": "" + }, + { + "type": "grafana", + "id": "grafana", + "name": "Grafana", + "version": "7.3.7" + }, + { + "type": "panel", + "id": "graph", + "name": "Graph", + "version": "" + }, + { + "type": "datasource", + "id": "prometheus", + "name": "Prometheus", + "version": "1.0.0" + }, + { + "type": "panel", + "id": "singlestat", + "name": "Singlestat", + "version": "" + } + ], + "annotations": { + "list": [ + { + "$$hashKey": "object:1058", + "builtIn": 1, + "datasource": "-- Grafana --", + "enable": true, + "hide": true, + "iconColor": "rgba(0, 211, 255, 1)", + "name": "Annotations & Alerts", + "type": "dashboard" + } + ] + }, + "editable": true, + "gnetId": 1860, + "graphTooltip": 0, + "id": null, + "iteration": 1641803921158, + "links": [ + { + "icon": "external link", + "tags": [], + "title": "GitHub", + "type": "link", + "url": "https://github.com/rfrail3/grafana-dashboards" + }, + { + "icon": "external link", + "tags": [], + "title": "Grafana", + "type": "link", + "url": "https://grafana.com/grafana/dashboards/1860" + } + ], + "panels": [ + { + "collapsed": false, + "datasource": "${DS_PROMETHEUS}", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 0 + }, + "id": 261, + "panels": [], + "repeat": null, + "title": "Quick CPU / Mem / Disk", + "type": "row" + }, + { + "cacheTimeout": null, + "datasource": "${DS_PROMETHEUS}", + "description": "Busy state of all CPU cores together", + "fieldConfig": { + "defaults": { + "color": { + "mode": "thresholds" + }, + "custom": {}, + "mappings": [ + { + "id": 0, + "op": "=", + "text": "N/A", + "type": 1, + "value": "null" + } + ], + "max": 100, + "min": 0, + "nullValueMode": "null", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "rgba(50, 172, 45, 0.97)", + "value": null + }, + { + "color": "rgba(237, 129, 40, 0.89)", + "value": 85 + }, + { + "color": "rgba(245, 54, 54, 0.9)", + "value": 95 + } + ] + }, + "unit": "percent" + }, + "overrides": [] + }, + "gridPos": { + "h": 4, + "w": 3, + "x": 0, + "y": 1 + }, + "id": 20, + "links": [], + "options": { + "orientation": "horizontal", + "reduceOptions": { + "calcs": [ + "lastNotNull" + ], + "fields": "", + "values": false + }, + "showThresholdLabels": false, + "showThresholdMarkers": true + }, + "pluginVersion": "7.3.7", + "targets": [ + { + "expr": "(((count(count(node_cpu_seconds_total{instance=\"$node\",job=\"$job\"}) by (cpu))) - avg(sum by (mode)(rate(node_cpu_seconds_total{mode='idle',instance=\"$node\",job=\"$job\"}[$__rate_interval])))) * 100) / count(count(node_cpu_seconds_total{instance=\"$node\",job=\"$job\"}) by (cpu))", + "hide": false, + "intervalFactor": 1, + "legendFormat": "", + "refId": "A", + "step": 240 + } + ], + "title": "CPU Busy", + "type": "gauge" + }, + { + "cacheTimeout": null, + "datasource": "${DS_PROMETHEUS}", + "description": "Busy state of all CPU cores together (5 min average)", + "fieldConfig": { + "defaults": { + "color": { + "mode": "thresholds" + }, + "custom": {}, + "mappings": [ + { + "id": 0, + "op": "=", + "text": "N/A", + "type": 1, + "value": "null" + } + ], + "max": 100, + "min": 0, + "nullValueMode": "null", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "rgba(50, 172, 45, 0.97)", + "value": null + }, + { + "color": "rgba(237, 129, 40, 0.89)", + "value": 85 + }, + { + "color": "rgba(245, 54, 54, 0.9)", + "value": 95 + } + ] + }, + "unit": "percent" + }, + "overrides": [] + }, + "gridPos": { + "h": 4, + "w": 3, + "x": 3, + "y": 1 + }, + "id": 155, + "links": [], + "options": { + "orientation": "horizontal", + "reduceOptions": { + "calcs": [ + "lastNotNull" + ], + "fields": "", + "values": false + }, + "showThresholdLabels": false, + "showThresholdMarkers": true + }, + "pluginVersion": "7.3.7", + "targets": [ + { + "expr": "avg(node_load5{instance=\"$node\",job=\"$job\"}) / count(count(node_cpu_seconds_total{instance=\"$node\",job=\"$job\"}) by (cpu)) * 100", + "format": "time_series", + "hide": false, + "intervalFactor": 1, + "refId": "A", + "step": 240 + } + ], + "title": "Sys Load (5m avg)", + "type": "gauge" + }, + { + "cacheTimeout": null, + "datasource": "${DS_PROMETHEUS}", + "description": "Busy state of all CPU cores together (15 min average)", + "fieldConfig": { + "defaults": { + "color": { + "mode": "thresholds" + }, + "custom": {}, + "mappings": [ + { + "id": 0, + "op": "=", + "text": "N/A", + "type": 1, + "value": "null" + } + ], + "max": 100, + "min": 0, + "nullValueMode": "null", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "rgba(50, 172, 45, 0.97)", + "value": null + }, + { + "color": "rgba(237, 129, 40, 0.89)", + "value": 85 + }, + { + "color": "rgba(245, 54, 54, 0.9)", + "value": 95 + } + ] + }, + "unit": "percent" + }, + "overrides": [] + }, + "gridPos": { + "h": 4, + "w": 3, + "x": 6, + "y": 1 + }, + "id": 19, + "links": [], + "options": { + "orientation": "horizontal", + "reduceOptions": { + "calcs": [ + "lastNotNull" + ], + "fields": "", + "values": false + }, + "showThresholdLabels": false, + "showThresholdMarkers": true + }, + "pluginVersion": "7.3.7", + "targets": [ + { + "expr": "avg(node_load15{instance=\"$node\",job=\"$job\"}) / count(count(node_cpu_seconds_total{instance=\"$node\",job=\"$job\"}) by (cpu)) * 100", + "hide": false, + "intervalFactor": 1, + "refId": "A", + "step": 240 + } + ], + "title": "Sys Load (15m avg)", + "type": "gauge" + }, + { + "cacheTimeout": null, + "datasource": "${DS_PROMETHEUS}", + "description": "Non available RAM memory", + "fieldConfig": { + "defaults": { + "color": { + "mode": "thresholds" + }, + "custom": {}, + "decimals": 0, + "mappings": [], + "max": 100, + "min": 0, + "nullValueMode": "null", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "rgba(50, 172, 45, 0.97)", + "value": null + }, + { + "color": "rgba(237, 129, 40, 0.89)", + "value": 80 + }, + { + "color": "rgba(245, 54, 54, 0.9)", + "value": 90 + } + ] + }, + "unit": "percent" + }, + "overrides": [] + }, + "gridPos": { + "h": 4, + "w": 3, + "x": 9, + "y": 1 + }, + "hideTimeOverride": false, + "id": 16, + "links": [], + "options": { + "orientation": "horizontal", + "reduceOptions": { + "calcs": [ + "lastNotNull" + ], + "fields": "", + "values": false + }, + "showThresholdLabels": false, + "showThresholdMarkers": true + }, + "pluginVersion": "7.3.7", + "targets": [ + { + "expr": "((node_memory_MemTotal_bytes{instance=\"$node\",job=\"$job\"} - node_memory_MemFree_bytes{instance=\"$node\",job=\"$job\"}) / (node_memory_MemTotal_bytes{instance=\"$node\",job=\"$job\"} )) * 100", + "format": "time_series", + "hide": true, + "intervalFactor": 1, + "refId": "A", + "step": 240 + }, + { + "expr": "100 - ((node_memory_MemAvailable_bytes{instance=\"$node\",job=\"$job\"} * 100) / node_memory_MemTotal_bytes{instance=\"$node\",job=\"$job\"})", + "format": "time_series", + "hide": false, + "intervalFactor": 1, + "refId": "B", + "step": 240 + } + ], + "title": "RAM Used", + "type": "gauge" + }, + { + "cacheTimeout": null, + "datasource": "${DS_PROMETHEUS}", + "description": "Used Swap", + "fieldConfig": { + "defaults": { + "color": { + "mode": "thresholds" + }, + "custom": {}, + "mappings": [ + { + "id": 0, + "op": "=", + "text": "N/A", + "type": 1, + "value": "null" + } + ], + "max": 100, + "min": 0, + "nullValueMode": "null", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "rgba(50, 172, 45, 0.97)", + "value": null + }, + { + "color": "rgba(237, 129, 40, 0.89)", + "value": 10 + }, + { + "color": "rgba(245, 54, 54, 0.9)", + "value": 25 + } + ] + }, + "unit": "percent" + }, + "overrides": [] + }, + "gridPos": { + "h": 4, + "w": 3, + "x": 12, + "y": 1 + }, + "id": 21, + "links": [], + "options": { + "orientation": "horizontal", + "reduceOptions": { + "calcs": [ + "lastNotNull" + ], + "fields": "", + "values": false + }, + "showThresholdLabels": false, + "showThresholdMarkers": true + }, + "pluginVersion": "7.3.7", + "targets": [ + { + "expr": "((node_memory_SwapTotal_bytes{instance=\"$node\",job=\"$job\"} - node_memory_SwapFree_bytes{instance=\"$node\",job=\"$job\"}) / (node_memory_SwapTotal_bytes{instance=\"$node\",job=\"$job\"} )) * 100", + "intervalFactor": 1, + "refId": "A", + "step": 240 + } + ], + "title": "SWAP Used", + "type": "gauge" + }, + { + "cacheTimeout": null, + "datasource": "${DS_PROMETHEUS}", + "description": "Used Root FS", + "fieldConfig": { + "defaults": { + "color": { + "mode": "thresholds" + }, + "custom": {}, + "mappings": [ + { + "id": 0, + "op": "=", + "text": "N/A", + "type": 1, + "value": "null" + } + ], + "max": 100, + "min": 0, + "nullValueMode": "null", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "rgba(50, 172, 45, 0.97)", + "value": null + }, + { + "color": "rgba(237, 129, 40, 0.89)", + "value": 80 + }, + { + "color": "rgba(245, 54, 54, 0.9)", + "value": 90 + } + ] + }, + "unit": "percent" + }, + "overrides": [] + }, + "gridPos": { + "h": 4, + "w": 3, + "x": 15, + "y": 1 + }, + "id": 154, + "links": [], + "options": { + "orientation": "horizontal", + "reduceOptions": { + "calcs": [ + "lastNotNull" + ], + "fields": "", + "values": false + }, + "showThresholdLabels": false, + "showThresholdMarkers": true + }, + "pluginVersion": "7.3.7", + "targets": [ + { + "expr": "100 - ((node_filesystem_avail_bytes{instance=\"$node\",job=\"$job\",mountpoint=\"/\",fstype!=\"rootfs\"} * 100) / node_filesystem_size_bytes{instance=\"$node\",job=\"$job\",mountpoint=\"/\",fstype!=\"rootfs\"})", + "format": "time_series", + "intervalFactor": 1, + "refId": "A", + "step": 240 + } + ], + "title": "Root FS Used", + "type": "gauge" + }, + { + "cacheTimeout": null, + "colorBackground": false, + "colorValue": false, + "colors": [ + "rgba(245, 54, 54, 0.9)", + "rgba(237, 129, 40, 0.89)", + "rgba(50, 172, 45, 0.97)" + ], + "datasource": "${DS_PROMETHEUS}", + "description": "Total number of CPU cores", + "fieldConfig": { + "defaults": { + "custom": {} + }, + "overrides": [] + }, + "format": "short", + "gauge": { + "maxValue": 100, + "minValue": 0, + "show": false, + "thresholdLabels": false, + "thresholdMarkers": true + }, + "gridPos": { + "h": 2, + "w": 2, + "x": 18, + "y": 1 + }, + "id": 14, + "interval": null, + "links": [], + "mappingType": 1, + "mappingTypes": [ + { + "name": "value to text", + "value": 1 + }, + { + "name": "range to text", + "value": 2 + } + ], + "maxDataPoints": 100, + "maxPerRow": 6, + "nullPointMode": "null", + "nullText": null, + "postfix": "", + "postfixFontSize": "50%", + "prefix": "", + "prefixFontSize": "50%", + "rangeMaps": [ + { + "from": "null", + "text": "N/A", + "to": "null" + } + ], + "sparkline": { + "fillColor": "rgba(31, 118, 189, 0.18)", + "full": false, + "lineColor": "rgb(31, 120, 193)", + "show": false + }, + "tableColumn": "", + "targets": [ + { + "expr": "count(count(node_cpu_seconds_total{instance=\"$node\",job=\"$job\"}) by (cpu))", + "interval": "", + "intervalFactor": 1, + "legendFormat": "", + "refId": "A", + "step": 240 + } + ], + "thresholds": "", + "title": "CPU Cores", + "type": "singlestat", + "valueFontSize": "50%", + "valueMaps": [ + { + "op": "=", + "text": "N/A", + "value": "null" + } + ], + "valueName": "current" + }, + { + "cacheTimeout": null, + "colorBackground": false, + "colorValue": false, + "colors": [ + "rgba(245, 54, 54, 0.9)", + "rgba(237, 129, 40, 0.89)", + "rgba(50, 172, 45, 0.97)" + ], + "datasource": "${DS_PROMETHEUS}", + "decimals": 1, + "description": "System uptime", + "fieldConfig": { + "defaults": { + "custom": {} + }, + "overrides": [] + }, + "format": "s", + "gauge": { + "maxValue": 100, + "minValue": 0, + "show": false, + "thresholdLabels": false, + "thresholdMarkers": true + }, + "gridPos": { + "h": 2, + "w": 4, + "x": 20, + "y": 1 + }, + "hideTimeOverride": true, + "id": 15, + "interval": null, + "links": [], + "mappingType": 1, + "mappingTypes": [ + { + "$$hashKey": "object:1094", + "name": "value to text", + "value": 1 + }, + { + "$$hashKey": "object:1095", + "name": "range to text", + "value": 2 + } + ], + "maxDataPoints": 100, + "nullPointMode": "null", + "nullText": null, + "postfix": "s", + "postfixFontSize": "50%", + "prefix": "", + "prefixFontSize": "50%", + "rangeMaps": [ + { + "from": "null", + "text": "N/A", + "to": "null" + } + ], + "sparkline": { + "fillColor": "rgba(31, 118, 189, 0.18)", + "full": false, + "lineColor": "rgb(31, 120, 193)", + "show": false + }, + "tableColumn": "", + "targets": [ + { + "expr": "node_time_seconds{instance=\"$node\",job=\"$job\"} - node_boot_time_seconds{instance=\"$node\",job=\"$job\"}", + "intervalFactor": 1, + "refId": "A", + "step": 240 + } + ], + "thresholds": "", + "title": "Uptime", + "type": "singlestat", + "valueFontSize": "50%", + "valueMaps": [ + { + "$$hashKey": "object:1097", + "op": "=", + "text": "N/A", + "value": "null" + } + ], + "valueName": "current" + }, + { + "cacheTimeout": null, + "colorBackground": false, + "colorValue": false, + "colors": [ + "rgba(50, 172, 45, 0.97)", + "rgba(237, 129, 40, 0.89)", + "rgba(245, 54, 54, 0.9)" + ], + "datasource": "${DS_PROMETHEUS}", + "decimals": 0, + "description": "Total RootFS", + "fieldConfig": { + "defaults": { + "custom": {} + }, + "overrides": [] + }, + "format": "bytes", + "gauge": { + "maxValue": 100, + "minValue": 0, + "show": false, + "thresholdLabels": false, + "thresholdMarkers": true + }, + "gridPos": { + "h": 2, + "w": 2, + "x": 18, + "y": 3 + }, + "id": 23, + "interval": null, + "links": [], + "mappingType": 1, + "mappingTypes": [ + { + "name": "value to text", + "value": 1 + }, + { + "name": "range to text", + "value": 2 + } + ], + "maxDataPoints": 100, + "maxPerRow": 6, + "nullPointMode": "null", + "nullText": null, + "postfix": "", + "postfixFontSize": "50%", + "prefix": "", + "prefixFontSize": "50%", + "rangeMaps": [ + { + "from": "null", + "text": "N/A", + "to": "null" + } + ], + "sparkline": { + "fillColor": "rgba(31, 118, 189, 0.18)", + "full": false, + "lineColor": "rgb(31, 120, 193)", + "show": false + }, + "tableColumn": "", + "targets": [ + { + "expr": "node_filesystem_size_bytes{instance=\"$node\",job=\"$job\",mountpoint=\"/\",fstype!=\"rootfs\"}", + "format": "time_series", + "hide": false, + "intervalFactor": 1, + "refId": "A", + "step": 240 + } + ], + "thresholds": "70,90", + "title": "RootFS Total", + "type": "singlestat", + "valueFontSize": "50%", + "valueMaps": [ + { + "op": "=", + "text": "N/A", + "value": "null" + } + ], + "valueName": "current" + }, + { + "cacheTimeout": null, + "colorBackground": false, + "colorValue": false, + "colors": [ + "rgba(245, 54, 54, 0.9)", + "rgba(237, 129, 40, 0.89)", + "rgba(50, 172, 45, 0.97)" + ], + "datasource": "${DS_PROMETHEUS}", + "decimals": 0, + "description": "Total RAM", + "fieldConfig": { + "defaults": { + "custom": {} + }, + "overrides": [] + }, + "format": "bytes", + "gauge": { + "maxValue": 100, + "minValue": 0, + "show": false, + "thresholdLabels": false, + "thresholdMarkers": true + }, + "gridPos": { + "h": 2, + "w": 2, + "x": 20, + "y": 3 + }, + "id": 75, + "interval": null, + "links": [], + "mappingType": 1, + "mappingTypes": [ + { + "name": "value to text", + "value": 1 + }, + { + "name": "range to text", + "value": 2 + } + ], + "maxDataPoints": 100, + "maxPerRow": 6, + "nullPointMode": "null", + "nullText": null, + "postfix": "", + "postfixFontSize": "70%", + "prefix": "", + "prefixFontSize": "50%", + "rangeMaps": [ + { + "from": "null", + "text": "N/A", + "to": "null" + } + ], + "sparkline": { + "fillColor": "rgba(31, 118, 189, 0.18)", + "full": false, + "lineColor": "rgb(31, 120, 193)", + "show": false + }, + "tableColumn": "", + "targets": [ + { + "expr": "node_memory_MemTotal_bytes{instance=\"$node\",job=\"$job\"}", + "intervalFactor": 1, + "refId": "A", + "step": 240 + } + ], + "thresholds": "", + "title": "RAM Total", + "type": "singlestat", + "valueFontSize": "50%", + "valueMaps": [ + { + "op": "=", + "text": "N/A", + "value": "null" + } + ], + "valueName": "current" + }, + { + "cacheTimeout": null, + "colorBackground": false, + "colorValue": false, + "colors": [ + "rgba(245, 54, 54, 0.9)", + "rgba(237, 129, 40, 0.89)", + "rgba(50, 172, 45, 0.97)" + ], + "datasource": "${DS_PROMETHEUS}", + "decimals": 0, + "description": "Total SWAP", + "fieldConfig": { + "defaults": { + "custom": {} + }, + "overrides": [] + }, + "format": "bytes", + "gauge": { + "maxValue": 100, + "minValue": 0, + "show": false, + "thresholdLabels": false, + "thresholdMarkers": true + }, + "gridPos": { + "h": 2, + "w": 2, + "x": 22, + "y": 3 + }, + "id": 18, + "interval": null, + "links": [], + "mappingType": 1, + "mappingTypes": [ + { + "name": "value to text", + "value": 1 + }, + { + "name": "range to text", + "value": 2 + } + ], + "maxDataPoints": 100, + "maxPerRow": 6, + "nullPointMode": "null", + "nullText": null, + "postfix": "", + "postfixFontSize": "70%", + "prefix": "", + "prefixFontSize": "50%", + "rangeMaps": [ + { + "from": "null", + "text": "N/A", + "to": "null" + } + ], + "sparkline": { + "fillColor": "rgba(31, 118, 189, 0.18)", + "full": false, + "lineColor": "rgb(31, 120, 193)", + "show": false + }, + "tableColumn": "", + "targets": [ + { + "expr": "node_memory_SwapTotal_bytes{instance=\"$node\",job=\"$job\"}", + "intervalFactor": 1, + "refId": "A", + "step": 240 + } + ], + "thresholds": "", + "title": "SWAP Total", + "type": "singlestat", + "valueFontSize": "50%", + "valueMaps": [ + { + "op": "=", + "text": "N/A", + "value": "null" + } + ], + "valueName": "current" + }, + { + "collapsed": false, + "datasource": "${DS_PROMETHEUS}", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 5 + }, + "id": 263, + "panels": [], + "repeat": null, + "title": "Basic CPU / Mem / Net / Disk", + "type": "row" + }, + { + "aliasColors": { + "Busy": "#EAB839", + "Busy Iowait": "#890F02", + "Busy other": "#1F78C1", + "Idle": "#052B51", + "Idle - Waiting for something to happen": "#052B51", + "guest": "#9AC48A", + "idle": "#052B51", + "iowait": "#EAB839", + "irq": "#BF1B00", + "nice": "#C15C17", + "softirq": "#E24D42", + "steal": "#FCE2DE", + "system": "#508642", + "user": "#5195CE" + }, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "decimals": 2, + "description": "Basic CPU info", + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 4, + "fillGradient": 0, + "gridPos": { + "h": 7, + "w": 12, + "x": 0, + "y": 6 + }, + "hiddenSeries": false, + "id": 77, + "legend": { + "alignAsTable": false, + "avg": false, + "current": false, + "max": false, + "min": false, + "rightSide": false, + "show": true, + "sideWidth": 250, + "sort": null, + "sortDesc": null, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": true, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "alias": "Busy Iowait", + "color": "#890F02" + }, + { + "alias": "Idle", + "color": "#7EB26D" + }, + { + "alias": "Busy System", + "color": "#EAB839" + }, + { + "alias": "Busy User", + "color": "#0A437C" + }, + { + "alias": "Busy Other", + "color": "#6D1F62" + } + ], + "spaceLength": 10, + "stack": true, + "steppedLine": false, + "targets": [ + { + "expr": "sum by (instance)(rate(node_cpu_seconds_total{mode=\"system\",instance=\"$node\",job=\"$job\"}[$__rate_interval])) * 100", + "format": "time_series", + "hide": false, + "intervalFactor": 1, + "legendFormat": "Busy System", + "refId": "A", + "step": 240 + }, + { + "expr": "sum by (instance)(rate(node_cpu_seconds_total{mode='user',instance=\"$node\",job=\"$job\"}[$__rate_interval])) * 100", + "format": "time_series", + "hide": false, + "intervalFactor": 1, + "legendFormat": "Busy User", + "refId": "B", + "step": 240 + }, + { + "expr": "sum by (instance)(rate(node_cpu_seconds_total{mode='iowait',instance=\"$node\",job=\"$job\"}[$__rate_interval])) * 100", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "Busy Iowait", + "refId": "C", + "step": 240 + }, + { + "expr": "sum by (instance)(rate(node_cpu_seconds_total{mode=~\".*irq\",instance=\"$node\",job=\"$job\"}[$__rate_interval])) * 100", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "Busy IRQs", + "refId": "D", + "step": 240 + }, + { + "expr": "sum (rate(node_cpu_seconds_total{mode!='idle',mode!='user',mode!='system',mode!='iowait',mode!='irq',mode!='softirq',instance=\"$node\",job=\"$job\"}[$__rate_interval])) * 100", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "Busy Other", + "refId": "E", + "step": 240 + }, + { + "expr": "sum by (mode)(rate(node_cpu_seconds_total{mode='idle',instance=\"$node\",job=\"$job\"}[$__rate_interval])) * 100", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "Idle", + "refId": "F", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "CPU Basic", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:123", + "format": "short", + "label": "", + "logBase": 1, + "max": "100", + "min": "0", + "show": true + }, + { + "$$hashKey": "object:124", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": { + "Apps": "#629E51", + "Buffers": "#614D93", + "Cache": "#6D1F62", + "Cached": "#511749", + "Committed": "#508642", + "Free": "#0A437C", + "Hardware Corrupted - Amount of RAM that the kernel identified as corrupted / not working": "#CFFAFF", + "Inactive": "#584477", + "PageTables": "#0A50A1", + "Page_Tables": "#0A50A1", + "RAM_Free": "#E0F9D7", + "SWAP Used": "#BF1B00", + "Slab": "#806EB7", + "Slab_Cache": "#E0752D", + "Swap": "#BF1B00", + "Swap Used": "#BF1B00", + "Swap_Cache": "#C15C17", + "Swap_Free": "#2F575E", + "Unused": "#EAB839" + }, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "decimals": 2, + "description": "Basic memory usage", + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 4, + "fillGradient": 0, + "gridPos": { + "h": 7, + "w": 12, + "x": 12, + "y": 6 + }, + "hiddenSeries": false, + "id": 78, + "legend": { + "alignAsTable": false, + "avg": false, + "current": false, + "max": false, + "min": false, + "rightSide": false, + "show": true, + "sideWidth": 350, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "alias": "RAM Total", + "color": "#E0F9D7", + "fill": 0, + "stack": false + }, + { + "alias": "RAM Cache + Buffer", + "color": "#052B51" + }, + { + "alias": "RAM Free", + "color": "#7EB26D" + }, + { + "alias": "Avaliable", + "color": "#DEDAF7", + "fill": 0, + "stack": false + } + ], + "spaceLength": 10, + "stack": true, + "steppedLine": false, + "targets": [ + { + "expr": "node_memory_MemTotal_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "hide": false, + "intervalFactor": 1, + "legendFormat": "RAM Total", + "refId": "A", + "step": 240 + }, + { + "expr": "node_memory_MemTotal_bytes{instance=\"$node\",job=\"$job\"} - node_memory_MemFree_bytes{instance=\"$node\",job=\"$job\"} - (node_memory_Cached_bytes{instance=\"$node\",job=\"$job\"} + node_memory_Buffers_bytes{instance=\"$node\",job=\"$job\"})", + "format": "time_series", + "hide": false, + "intervalFactor": 1, + "legendFormat": "RAM Used", + "refId": "B", + "step": 240 + }, + { + "expr": "node_memory_Cached_bytes{instance=\"$node\",job=\"$job\"} + node_memory_Buffers_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "RAM Cache + Buffer", + "refId": "C", + "step": 240 + }, + { + "expr": "node_memory_MemFree_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "RAM Free", + "refId": "D", + "step": 240 + }, + { + "expr": "(node_memory_SwapTotal_bytes{instance=\"$node\",job=\"$job\"} - node_memory_SwapFree_bytes{instance=\"$node\",job=\"$job\"})", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "SWAP Used", + "refId": "E", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Memory Basic", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "bytes", + "label": "", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": { + "Recv_bytes_eth2": "#7EB26D", + "Recv_bytes_lo": "#0A50A1", + "Recv_drop_eth2": "#6ED0E0", + "Recv_drop_lo": "#E0F9D7", + "Recv_errs_eth2": "#BF1B00", + "Recv_errs_lo": "#CCA300", + "Trans_bytes_eth2": "#7EB26D", + "Trans_bytes_lo": "#0A50A1", + "Trans_drop_eth2": "#6ED0E0", + "Trans_drop_lo": "#E0F9D7", + "Trans_errs_eth2": "#BF1B00", + "Trans_errs_lo": "#CCA300", + "recv_bytes_lo": "#0A50A1", + "recv_drop_eth0": "#99440A", + "recv_drop_lo": "#967302", + "recv_errs_eth0": "#BF1B00", + "recv_errs_lo": "#890F02", + "trans_bytes_eth0": "#7EB26D", + "trans_bytes_lo": "#0A50A1", + "trans_drop_eth0": "#99440A", + "trans_drop_lo": "#967302", + "trans_errs_eth0": "#BF1B00", + "trans_errs_lo": "#890F02" + }, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "description": "Basic network info per interface", + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 4, + "fillGradient": 0, + "gridPos": { + "h": 7, + "w": 12, + "x": 0, + "y": 13 + }, + "hiddenSeries": false, + "id": 74, + "legend": { + "alignAsTable": false, + "avg": false, + "current": false, + "hideEmpty": false, + "hideZero": false, + "max": false, + "min": false, + "rightSide": false, + "show": true, + "sort": "current", + "sortDesc": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "alias": "/.*trans.*/", + "transform": "negative-Y" + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_network_receive_bytes_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])*8", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "recv {{device}}", + "refId": "A", + "step": 240 + }, + { + "expr": "rate(node_network_transmit_bytes_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])*8", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "trans {{device}} ", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Network Traffic Basic", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "bps", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "format": "pps", + "label": "", + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "decimals": 3, + "description": "Disk space used of all filesystems mounted", + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 4, + "fillGradient": 0, + "gridPos": { + "h": 7, + "w": 12, + "x": 12, + "y": 13 + }, + "height": "", + "hiddenSeries": false, + "id": 152, + "legend": { + "alignAsTable": false, + "avg": false, + "current": false, + "max": false, + "min": false, + "rightSide": false, + "show": true, + "sort": "current", + "sortDesc": false, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "100 - ((node_filesystem_avail_bytes{instance=\"$node\",job=\"$job\",device!~'rootfs'} * 100) / node_filesystem_size_bytes{instance=\"$node\",job=\"$job\",device!~'rootfs'})", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "{{mountpoint}}", + "refId": "A", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Disk Space Used Basic", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "percent", + "label": null, + "logBase": 1, + "max": "100", + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": true + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "collapsed": true, + "datasource": "${DS_PROMETHEUS}", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 20 + }, + "id": 265, + "panels": [ + { + "aliasColors": { + "Idle - Waiting for something to happen": "#052B51", + "guest": "#9AC48A", + "idle": "#052B51", + "iowait": "#EAB839", + "irq": "#BF1B00", + "nice": "#C15C17", + "softirq": "#E24D42", + "steal": "#FCE2DE", + "system": "#508642", + "user": "#5195CE" + }, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "decimals": 2, + "description": "", + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 4, + "fillGradient": 0, + "gridPos": { + "h": 12, + "w": 12, + "x": 0, + "y": 3 + }, + "hiddenSeries": false, + "id": 3, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sideWidth": 250, + "sort": null, + "sortDesc": null, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": true, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "repeat": null, + "seriesOverrides": [], + "spaceLength": 10, + "stack": true, + "steppedLine": false, + "targets": [ + { + "expr": "sum by (mode)(rate(node_cpu_seconds_total{mode=\"system\",instance=\"$node\",job=\"$job\"}[$__rate_interval])) * 100", + "format": "time_series", + "interval": "10s", + "intervalFactor": 1, + "legendFormat": "System - Processes executing in kernel mode", + "refId": "A", + "step": 240 + }, + { + "expr": "sum by (mode)(rate(node_cpu_seconds_total{mode='user',instance=\"$node\",job=\"$job\"}[$__rate_interval])) * 100", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "User - Normal processes executing in user mode", + "refId": "B", + "step": 240 + }, + { + "expr": "sum by (mode)(rate(node_cpu_seconds_total{mode='nice',instance=\"$node\",job=\"$job\"}[$__rate_interval])) * 100", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "Nice - Niced processes executing in user mode", + "refId": "C", + "step": 240 + }, + { + "expr": "sum by (mode)(rate(node_cpu_seconds_total{mode='idle',instance=\"$node\",job=\"$job\"}[$__rate_interval])) * 100", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "Idle - Waiting for something to happen", + "refId": "D", + "step": 240 + }, + { + "expr": "sum by (mode)(rate(node_cpu_seconds_total{mode='iowait',instance=\"$node\",job=\"$job\"}[$__rate_interval])) * 100", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "Iowait - Waiting for I/O to complete", + "refId": "E", + "step": 240 + }, + { + "expr": "sum by (mode)(rate(node_cpu_seconds_total{mode='irq',instance=\"$node\",job=\"$job\"}[$__rate_interval])) * 100", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "Irq - Servicing interrupts", + "refId": "F", + "step": 240 + }, + { + "expr": "sum by (mode)(rate(node_cpu_seconds_total{mode='softirq',instance=\"$node\",job=\"$job\"}[$__rate_interval])) * 100", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "Softirq - Servicing softirqs", + "refId": "G", + "step": 240 + }, + { + "expr": "sum by (mode)(rate(node_cpu_seconds_total{mode='steal',instance=\"$node\",job=\"$job\"}[$__rate_interval])) * 100", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "Steal - Time spent in other operating systems when running in a virtualized environment", + "refId": "H", + "step": 240 + }, + { + "expr": "sum by (mode)(rate(node_cpu_seconds_total{mode='guest',instance=\"$node\",job=\"$job\"}[$__rate_interval])) * 100", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "Guest - Time spent running a virtual CPU for a guest operating system", + "refId": "I", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "CPU", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "label": "percentage", + "logBase": 1, + "max": "100", + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": { + "Apps": "#629E51", + "Buffers": "#614D93", + "Cache": "#6D1F62", + "Cached": "#511749", + "Committed": "#508642", + "Free": "#0A437C", + "Hardware Corrupted - Amount of RAM that the kernel identified as corrupted / not working": "#CFFAFF", + "Inactive": "#584477", + "PageTables": "#0A50A1", + "Page_Tables": "#0A50A1", + "RAM_Free": "#E0F9D7", + "Slab": "#806EB7", + "Slab_Cache": "#E0752D", + "Swap": "#BF1B00", + "Swap - Swap memory usage": "#BF1B00", + "Swap_Cache": "#C15C17", + "Swap_Free": "#2F575E", + "Unused": "#EAB839", + "Unused - Free memory unassigned": "#052B51" + }, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "decimals": 2, + "description": "", + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 4, + "fillGradient": 0, + "gridPos": { + "h": 12, + "w": 12, + "x": 12, + "y": 3 + }, + "hiddenSeries": false, + "id": 24, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sideWidth": 350, + "sort": null, + "sortDesc": null, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "alias": "/.*Hardware Corrupted - *./", + "stack": false + } + ], + "spaceLength": 10, + "stack": true, + "steppedLine": false, + "targets": [ + { + "expr": "node_memory_MemTotal_bytes{instance=\"$node\",job=\"$job\"} - node_memory_MemFree_bytes{instance=\"$node\",job=\"$job\"} - node_memory_Buffers_bytes{instance=\"$node\",job=\"$job\"} - node_memory_Cached_bytes{instance=\"$node\",job=\"$job\"} - node_memory_Slab_bytes{instance=\"$node\",job=\"$job\"} - node_memory_PageTables_bytes{instance=\"$node\",job=\"$job\"} - node_memory_SwapCached_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "hide": false, + "intervalFactor": 1, + "legendFormat": "Apps - Memory used by user-space applications", + "refId": "A", + "step": 240 + }, + { + "expr": "node_memory_PageTables_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "hide": false, + "intervalFactor": 1, + "legendFormat": "PageTables - Memory used to map between virtual and physical memory addresses", + "refId": "B", + "step": 240 + }, + { + "expr": "node_memory_SwapCached_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "SwapCache - Memory that keeps track of pages that have been fetched from swap but not yet been modified", + "refId": "C", + "step": 240 + }, + { + "expr": "node_memory_Slab_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "hide": false, + "intervalFactor": 1, + "legendFormat": "Slab - Memory used by the kernel to cache data structures for its own use (caches like inode, dentry, etc)", + "refId": "D", + "step": 240 + }, + { + "expr": "node_memory_Cached_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "hide": false, + "intervalFactor": 1, + "legendFormat": "Cache - Parked file data (file content) cache", + "refId": "E", + "step": 240 + }, + { + "expr": "node_memory_Buffers_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "hide": false, + "intervalFactor": 1, + "legendFormat": "Buffers - Block device (e.g. harddisk) cache", + "refId": "F", + "step": 240 + }, + { + "expr": "node_memory_MemFree_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "hide": false, + "intervalFactor": 1, + "legendFormat": "Unused - Free memory unassigned", + "refId": "G", + "step": 240 + }, + { + "expr": "(node_memory_SwapTotal_bytes{instance=\"$node\",job=\"$job\"} - node_memory_SwapFree_bytes{instance=\"$node\",job=\"$job\"})", + "format": "time_series", + "hide": false, + "intervalFactor": 1, + "legendFormat": "Swap - Swap space used", + "refId": "H", + "step": 240 + }, + { + "expr": "node_memory_HardwareCorrupted_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "hide": false, + "intervalFactor": 1, + "legendFormat": "Hardware Corrupted - Amount of RAM that the kernel identified as corrupted / not working", + "refId": "I", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Memory Stack", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "bytes", + "label": "bytes", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": { + "receive_packets_eth0": "#7EB26D", + "receive_packets_lo": "#E24D42", + "transmit_packets_eth0": "#7EB26D", + "transmit_packets_lo": "#E24D42" + }, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 4, + "fillGradient": 0, + "gridPos": { + "h": 12, + "w": 12, + "x": 0, + "y": 15 + }, + "hiddenSeries": false, + "id": 84, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "$$hashKey": "object:5871", + "alias": "/.*Trans.*/", + "transform": "negative-Y" + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_network_receive_bytes_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])*8", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "{{device}} - Receive", + "refId": "A", + "step": 240 + }, + { + "expr": "rate(node_network_transmit_bytes_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])*8", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "{{device}} - Transmit", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Network Traffic", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:5884", + "format": "bps", + "label": "bits out (-) / in (+)", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "$$hashKey": "object:5885", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "decimals": 3, + "description": "", + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 4, + "fillGradient": 0, + "gridPos": { + "h": 12, + "w": 12, + "x": 12, + "y": 15 + }, + "height": "", + "hiddenSeries": false, + "id": 156, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sort": "current", + "sortDesc": false, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_filesystem_size_bytes{instance=\"$node\",job=\"$job\",device!~'rootfs'} - node_filesystem_avail_bytes{instance=\"$node\",job=\"$job\",device!~'rootfs'}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "{{mountpoint}}", + "refId": "A", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Disk Space Used", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "bytes", + "label": "bytes", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "description": "", + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 12, + "w": 12, + "x": 0, + "y": 27 + }, + "hiddenSeries": false, + "id": 229, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideZero": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "alias": "/.*Read.*/", + "transform": "negative-Y" + }, + { + "alias": "/.*sda_.*/", + "color": "#7EB26D" + }, + { + "alias": "/.*sdb_.*/", + "color": "#EAB839" + }, + { + "alias": "/.*sdc_.*/", + "color": "#6ED0E0" + }, + { + "alias": "/.*sdd_.*/", + "color": "#EF843C" + }, + { + "alias": "/.*sde_.*/", + "color": "#E24D42" + }, + { + "alias": "/.*sda1.*/", + "color": "#584477" + }, + { + "alias": "/.*sda2_.*/", + "color": "#BA43A9" + }, + { + "alias": "/.*sda3_.*/", + "color": "#F4D598" + }, + { + "alias": "/.*sdb1.*/", + "color": "#0A50A1" + }, + { + "alias": "/.*sdb2.*/", + "color": "#BF1B00" + }, + { + "alias": "/.*sdb2.*/", + "color": "#BF1B00" + }, + { + "alias": "/.*sdb3.*/", + "color": "#E0752D" + }, + { + "alias": "/.*sdc1.*/", + "color": "#962D82" + }, + { + "alias": "/.*sdc2.*/", + "color": "#614D93" + }, + { + "alias": "/.*sdc3.*/", + "color": "#9AC48A" + }, + { + "alias": "/.*sdd1.*/", + "color": "#65C5DB" + }, + { + "alias": "/.*sdd2.*/", + "color": "#F9934E" + }, + { + "alias": "/.*sdd3.*/", + "color": "#EA6460" + }, + { + "alias": "/.*sde1.*/", + "color": "#E0F9D7" + }, + { + "alias": "/.*sdd2.*/", + "color": "#FCEACA" + }, + { + "alias": "/.*sde3.*/", + "color": "#F9E2D2" + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_disk_reads_completed_total{instance=\"$node\",job=\"$job\",device=~\"$diskdevices\"}[$__rate_interval])", + "intervalFactor": 4, + "legendFormat": "{{device}} - Reads completed", + "refId": "A", + "step": 240 + }, + { + "expr": "rate(node_disk_writes_completed_total{instance=\"$node\",job=\"$job\",device=~\"$diskdevices\"}[$__rate_interval])", + "intervalFactor": 1, + "legendFormat": "{{device}} - Writes completed", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Disk IOps", + "tooltip": { + "shared": false, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "iops", + "label": "IO read (-) / write (+)", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": { + "io time": "#890F02" + }, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "decimals": 3, + "description": "", + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 4, + "fillGradient": 0, + "gridPos": { + "h": 12, + "w": 12, + "x": 12, + "y": 27 + }, + "hiddenSeries": false, + "id": 42, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sort": null, + "sortDesc": null, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "alias": "/.*read*./", + "transform": "negative-Y" + }, + { + "alias": "/.*sda.*/", + "color": "#7EB26D" + }, + { + "alias": "/.*sdb.*/", + "color": "#EAB839" + }, + { + "alias": "/.*sdc.*/", + "color": "#6ED0E0" + }, + { + "alias": "/.*sdd.*/", + "color": "#EF843C" + }, + { + "alias": "/.*sde.*/", + "color": "#E24D42" + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_disk_read_bytes_total{instance=\"$node\",job=\"$job\",device=~\"$diskdevices\"}[$__rate_interval])", + "format": "time_series", + "hide": false, + "intervalFactor": 1, + "legendFormat": "{{device}} - Successfully read bytes", + "refId": "A", + "step": 240 + }, + { + "expr": "rate(node_disk_written_bytes_total{instance=\"$node\",job=\"$job\",device=~\"$diskdevices\"}[$__rate_interval])", + "format": "time_series", + "hide": false, + "intervalFactor": 1, + "legendFormat": "{{device}} - Successfully written bytes", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "I/O Usage Read / Write", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": false, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:965", + "format": "Bps", + "label": "bytes read (-) / write (+)", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "$$hashKey": "object:966", + "format": "ms", + "label": "", + "logBase": 1, + "max": null, + "min": null, + "show": true + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": { + "io time": "#890F02" + }, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "decimals": 3, + "description": "", + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 4, + "fillGradient": 0, + "gridPos": { + "h": 12, + "w": 12, + "x": 0, + "y": 39 + }, + "hiddenSeries": false, + "id": 127, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sort": null, + "sortDesc": null, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_disk_io_time_seconds_total{instance=\"$node\",job=\"$job\",device=~\"$diskdevices\"} [$__rate_interval])", + "format": "time_series", + "hide": false, + "interval": "", + "intervalFactor": 1, + "legendFormat": "{{device}}", + "refId": "A", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "I/O Utilization", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": false, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:1041", + "format": "percentunit", + "label": "%util", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "$$hashKey": "object:1042", + "format": "s", + "label": "", + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + } + ], + "repeat": null, + "title": "CPU / Memory / Net / Disk", + "type": "row" + }, + { + "collapsed": true, + "datasource": "${DS_PROMETHEUS}", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 21 + }, + "id": 266, + "panels": [ + { + "aliasColors": { + "Apps": "#629E51", + "Buffers": "#614D93", + "Cache": "#6D1F62", + "Cached": "#511749", + "Committed": "#508642", + "Free": "#0A437C", + "Hardware Corrupted - Amount of RAM that the kernel identified as corrupted / not working": "#CFFAFF", + "Inactive": "#584477", + "PageTables": "#0A50A1", + "Page_Tables": "#0A50A1", + "RAM_Free": "#E0F9D7", + "Slab": "#806EB7", + "Slab_Cache": "#E0752D", + "Swap": "#BF1B00", + "Swap_Cache": "#C15C17", + "Swap_Free": "#2F575E", + "Unused": "#EAB839" + }, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "decimals": 2, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 70 + }, + "hiddenSeries": false, + "id": 136, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sideWidth": 350, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 2, + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": true, + "steppedLine": false, + "targets": [ + { + "expr": "node_memory_Inactive_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "Inactive - Memory which has been less recently used. It is more eligible to be reclaimed for other purposes", + "refId": "A", + "step": 240 + }, + { + "expr": "node_memory_Active_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "Active - Memory that has been used more recently and usually not reclaimed unless absolutely necessary", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Memory Active / Inactive", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "cumulative" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "bytes", + "label": "bytes", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": { + "Apps": "#629E51", + "Buffers": "#614D93", + "Cache": "#6D1F62", + "Cached": "#511749", + "Committed": "#508642", + "Free": "#0A437C", + "Hardware Corrupted - Amount of RAM that the kernel identified as corrupted / not working": "#CFFAFF", + "Inactive": "#584477", + "PageTables": "#0A50A1", + "Page_Tables": "#0A50A1", + "RAM_Free": "#E0F9D7", + "Slab": "#806EB7", + "Slab_Cache": "#E0752D", + "Swap": "#BF1B00", + "Swap_Cache": "#C15C17", + "Swap_Free": "#2F575E", + "Unused": "#EAB839" + }, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "decimals": 2, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 70 + }, + "hiddenSeries": false, + "id": 135, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sideWidth": 350, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "alias": "/.*Committed_AS - *./" + }, + { + "alias": "/.*CommitLimit - *./", + "color": "#BF1B00", + "fill": 0 + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_memory_Committed_AS_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "Committed_AS - Amount of memory presently allocated on the system", + "refId": "A", + "step": 240 + }, + { + "expr": "node_memory_CommitLimit_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "CommitLimit - Amount of memory currently available to be allocated on the system", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Memory Commited", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "cumulative" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "bytes", + "label": "bytes", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": { + "Apps": "#629E51", + "Buffers": "#614D93", + "Cache": "#6D1F62", + "Cached": "#511749", + "Committed": "#508642", + "Free": "#0A437C", + "Hardware Corrupted - Amount of RAM that the kernel identified as corrupted / not working": "#CFFAFF", + "Inactive": "#584477", + "PageTables": "#0A50A1", + "Page_Tables": "#0A50A1", + "RAM_Free": "#E0F9D7", + "Slab": "#806EB7", + "Slab_Cache": "#E0752D", + "Swap": "#BF1B00", + "Swap_Cache": "#C15C17", + "Swap_Free": "#2F575E", + "Unused": "#EAB839" + }, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "decimals": 2, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 80 + }, + "hiddenSeries": false, + "id": 191, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sideWidth": 350, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": true, + "steppedLine": false, + "targets": [ + { + "expr": "node_memory_Inactive_file_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "hide": false, + "intervalFactor": 1, + "legendFormat": "Inactive_file - File-backed memory on inactive LRU list", + "refId": "A", + "step": 240 + }, + { + "expr": "node_memory_Inactive_anon_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "hide": false, + "intervalFactor": 1, + "legendFormat": "Inactive_anon - Anonymous and swap cache on inactive LRU list, including tmpfs (shmem)", + "refId": "B", + "step": 240 + }, + { + "expr": "node_memory_Active_file_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "hide": false, + "intervalFactor": 1, + "legendFormat": "Active_file - File-backed memory on active LRU list", + "refId": "C", + "step": 240 + }, + { + "expr": "node_memory_Active_anon_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "hide": false, + "intervalFactor": 1, + "legendFormat": "Active_anon - Anonymous and swap cache on active least-recently-used (LRU) list, including tmpfs", + "refId": "D", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Memory Active / Inactive Detail", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "cumulative" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "bytes", + "label": "bytes", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "bytes", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": true + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": { + "Active": "#99440A", + "Buffers": "#58140C", + "Cache": "#6D1F62", + "Cached": "#511749", + "Committed": "#508642", + "Dirty": "#6ED0E0", + "Free": "#B7DBAB", + "Inactive": "#EA6460", + "Mapped": "#052B51", + "PageTables": "#0A50A1", + "Page_Tables": "#0A50A1", + "Slab_Cache": "#EAB839", + "Swap": "#BF1B00", + "Swap_Cache": "#C15C17", + "Total": "#511749", + "Total RAM": "#052B51", + "Total RAM + Swap": "#052B51", + "Total Swap": "#614D93", + "VmallocUsed": "#EA6460" + }, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "decimals": 2, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 80 + }, + "hiddenSeries": false, + "id": 130, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sideWidth": null, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 2, + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_memory_Writeback_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "Writeback - Memory which is actively being written back to disk", + "refId": "A", + "step": 240 + }, + { + "expr": "node_memory_WritebackTmp_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "WritebackTmp - Memory used by FUSE for temporary writeback buffers", + "refId": "B", + "step": 240 + }, + { + "expr": "node_memory_Dirty_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "Dirty - Memory which is waiting to get written back to the disk", + "refId": "C", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Memory Writeback and Dirty", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "bytes", + "label": "bytes", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": { + "Apps": "#629E51", + "Buffers": "#614D93", + "Cache": "#6D1F62", + "Cached": "#511749", + "Committed": "#508642", + "Free": "#0A437C", + "Hardware Corrupted - Amount of RAM that the kernel identified as corrupted / not working": "#CFFAFF", + "Inactive": "#584477", + "PageTables": "#0A50A1", + "Page_Tables": "#0A50A1", + "RAM_Free": "#E0F9D7", + "Slab": "#806EB7", + "Slab_Cache": "#E0752D", + "Swap": "#BF1B00", + "Swap_Cache": "#C15C17", + "Swap_Free": "#2F575E", + "Unused": "#EAB839" + }, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "decimals": 2, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 90 + }, + "hiddenSeries": false, + "id": 138, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sideWidth": 350, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "$$hashKey": "object:4131", + "alias": "ShmemHugePages - Memory used by shared memory (shmem) and tmpfs allocated with huge pages", + "fill": 0 + }, + { + "$$hashKey": "object:4138", + "alias": "ShmemHugePages - Memory used by shared memory (shmem) and tmpfs allocated with huge pages", + "fill": 0 + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_memory_Mapped_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "Mapped - Used memory in mapped pages files which have been mmaped, such as libraries", + "refId": "A", + "step": 240 + }, + { + "expr": "node_memory_Shmem_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "Shmem - Used shared memory (shared between several processes, thus including RAM disks)", + "refId": "B", + "step": 240 + }, + { + "expr": "node_memory_ShmemHugePages_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "ShmemHugePages - Memory used by shared memory (shmem) and tmpfs allocated with huge pages", + "refId": "C", + "step": 240 + }, + { + "expr": "node_memory_ShmemPmdMapped_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "ShmemPmdMapped - Ammount of shared (shmem/tmpfs) memory backed by huge pages", + "refId": "D", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Memory Shared and Mapped", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "cumulative" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:4106", + "format": "bytes", + "label": "bytes", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "$$hashKey": "object:4107", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": { + "Active": "#99440A", + "Buffers": "#58140C", + "Cache": "#6D1F62", + "Cached": "#511749", + "Committed": "#508642", + "Dirty": "#6ED0E0", + "Free": "#B7DBAB", + "Inactive": "#EA6460", + "Mapped": "#052B51", + "PageTables": "#0A50A1", + "Page_Tables": "#0A50A1", + "Slab_Cache": "#EAB839", + "Swap": "#BF1B00", + "Swap_Cache": "#C15C17", + "Total": "#511749", + "Total RAM": "#052B51", + "Total RAM + Swap": "#052B51", + "Total Swap": "#614D93", + "VmallocUsed": "#EA6460" + }, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "decimals": 2, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 90 + }, + "hiddenSeries": false, + "id": 131, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sideWidth": null, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 2, + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": true, + "steppedLine": false, + "targets": [ + { + "expr": "node_memory_SUnreclaim_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "SUnreclaim - Part of Slab, that cannot be reclaimed on memory pressure", + "refId": "A", + "step": 240 + }, + { + "expr": "node_memory_SReclaimable_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "SReclaimable - Part of Slab, that might be reclaimed, such as caches", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Memory Slab", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "bytes", + "label": "bytes", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": { + "Active": "#99440A", + "Buffers": "#58140C", + "Cache": "#6D1F62", + "Cached": "#511749", + "Committed": "#508642", + "Dirty": "#6ED0E0", + "Free": "#B7DBAB", + "Inactive": "#EA6460", + "Mapped": "#052B51", + "PageTables": "#0A50A1", + "Page_Tables": "#0A50A1", + "Slab_Cache": "#EAB839", + "Swap": "#BF1B00", + "Swap_Cache": "#C15C17", + "Total": "#511749", + "Total RAM": "#052B51", + "Total RAM + Swap": "#052B51", + "VmallocUsed": "#EA6460" + }, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "decimals": 2, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 100 + }, + "hiddenSeries": false, + "id": 70, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sideWidth": null, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_memory_VmallocChunk_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "hide": false, + "intervalFactor": 1, + "legendFormat": "VmallocChunk - Largest contigious block of vmalloc area which is free", + "refId": "A", + "step": 240 + }, + { + "expr": "node_memory_VmallocTotal_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "hide": false, + "intervalFactor": 1, + "legendFormat": "VmallocTotal - Total size of vmalloc memory area", + "refId": "B", + "step": 240 + }, + { + "expr": "node_memory_VmallocUsed_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "hide": false, + "intervalFactor": 1, + "legendFormat": "VmallocUsed - Amount of vmalloc area which is used", + "refId": "C", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Memory Vmalloc", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "bytes", + "label": "bytes", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": { + "Apps": "#629E51", + "Buffers": "#614D93", + "Cache": "#6D1F62", + "Cached": "#511749", + "Committed": "#508642", + "Free": "#0A437C", + "Hardware Corrupted - Amount of RAM that the kernel identified as corrupted / not working": "#CFFAFF", + "Inactive": "#584477", + "PageTables": "#0A50A1", + "Page_Tables": "#0A50A1", + "RAM_Free": "#E0F9D7", + "Slab": "#806EB7", + "Slab_Cache": "#E0752D", + "Swap": "#BF1B00", + "Swap_Cache": "#C15C17", + "Swap_Free": "#2F575E", + "Unused": "#EAB839" + }, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "decimals": 2, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 100 + }, + "hiddenSeries": false, + "id": 159, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sideWidth": 350, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_memory_Bounce_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "Bounce - Memory used for block device bounce buffers", + "refId": "A", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Memory Bounce", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "cumulative" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "bytes", + "label": "bytes", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": { + "Active": "#99440A", + "Buffers": "#58140C", + "Cache": "#6D1F62", + "Cached": "#511749", + "Committed": "#508642", + "Dirty": "#6ED0E0", + "Free": "#B7DBAB", + "Inactive": "#EA6460", + "Mapped": "#052B51", + "PageTables": "#0A50A1", + "Page_Tables": "#0A50A1", + "Slab_Cache": "#EAB839", + "Swap": "#BF1B00", + "Swap_Cache": "#C15C17", + "Total": "#511749", + "Total RAM": "#052B51", + "Total RAM + Swap": "#052B51", + "VmallocUsed": "#EA6460" + }, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "decimals": 2, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 110 + }, + "hiddenSeries": false, + "id": 129, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sideWidth": null, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "alias": "/.*Inactive *./", + "transform": "negative-Y" + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_memory_AnonHugePages_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "AnonHugePages - Memory in anonymous huge pages", + "refId": "A", + "step": 240 + }, + { + "expr": "node_memory_AnonPages_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "AnonPages - Memory in user pages not backed by files", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Memory Anonymous", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "bytes", + "label": "bytes", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": { + "Apps": "#629E51", + "Buffers": "#614D93", + "Cache": "#6D1F62", + "Cached": "#511749", + "Committed": "#508642", + "Free": "#0A437C", + "Hardware Corrupted - Amount of RAM that the kernel identified as corrupted / not working": "#CFFAFF", + "Inactive": "#584477", + "PageTables": "#0A50A1", + "Page_Tables": "#0A50A1", + "RAM_Free": "#E0F9D7", + "Slab": "#806EB7", + "Slab_Cache": "#E0752D", + "Swap": "#BF1B00", + "Swap_Cache": "#C15C17", + "Swap_Free": "#2F575E", + "Unused": "#EAB839" + }, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "decimals": 2, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 110 + }, + "hiddenSeries": false, + "id": 160, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sideWidth": 350, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 2, + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_memory_KernelStack_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "KernelStack - Kernel memory stack. This is not reclaimable", + "refId": "A", + "step": 240 + }, + { + "expr": "node_memory_Percpu_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "PerCPU - Per CPU memory allocated dynamically by loadable modules", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Memory Kernel / CPU", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "cumulative" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "bytes", + "label": "bytes", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": { + "Active": "#99440A", + "Buffers": "#58140C", + "Cache": "#6D1F62", + "Cached": "#511749", + "Committed": "#508642", + "Dirty": "#6ED0E0", + "Free": "#B7DBAB", + "Inactive": "#EA6460", + "Mapped": "#052B51", + "PageTables": "#0A50A1", + "Page_Tables": "#0A50A1", + "Slab_Cache": "#EAB839", + "Swap": "#BF1B00", + "Swap_Cache": "#C15C17", + "Total": "#511749", + "Total RAM": "#806EB7", + "Total RAM + Swap": "#806EB7", + "VmallocUsed": "#EA6460" + }, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "decimals": 2, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 120 + }, + "hiddenSeries": false, + "id": 140, + "legend": { + "alignAsTable": true, + "avg": false, + "current": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sideWidth": null, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_memory_HugePages_Free{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "HugePages_Free - Huge pages in the pool that are not yet allocated", + "refId": "A", + "step": 240 + }, + { + "expr": "node_memory_HugePages_Rsvd{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "HugePages_Rsvd - Huge pages for which a commitment to allocate from the pool has been made, but no allocation has yet been made", + "refId": "B", + "step": 240 + }, + { + "expr": "node_memory_HugePages_Surp{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "HugePages_Surp - Huge pages in the pool above the value in /proc/sys/vm/nr_hugepages", + "refId": "C", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Memory HugePages Counter", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "label": "pages", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": "", + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": { + "Active": "#99440A", + "Buffers": "#58140C", + "Cache": "#6D1F62", + "Cached": "#511749", + "Committed": "#508642", + "Dirty": "#6ED0E0", + "Free": "#B7DBAB", + "Inactive": "#EA6460", + "Mapped": "#052B51", + "PageTables": "#0A50A1", + "Page_Tables": "#0A50A1", + "Slab_Cache": "#EAB839", + "Swap": "#BF1B00", + "Swap_Cache": "#C15C17", + "Total": "#511749", + "Total RAM": "#806EB7", + "Total RAM + Swap": "#806EB7", + "VmallocUsed": "#EA6460" + }, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "decimals": 2, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 120 + }, + "hiddenSeries": false, + "id": 71, + "legend": { + "alignAsTable": true, + "avg": false, + "current": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sideWidth": null, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 2, + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_memory_HugePages_Total{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "HugePages - Total size of the pool of huge pages", + "refId": "A", + "step": 240 + }, + { + "expr": "node_memory_Hugepagesize_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "Hugepagesize - Huge Page size", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Memory HugePages Size", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "bytes", + "label": "bytes", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": "", + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": { + "Active": "#99440A", + "Buffers": "#58140C", + "Cache": "#6D1F62", + "Cached": "#511749", + "Committed": "#508642", + "Dirty": "#6ED0E0", + "Free": "#B7DBAB", + "Inactive": "#EA6460", + "Mapped": "#052B51", + "PageTables": "#0A50A1", + "Page_Tables": "#0A50A1", + "Slab_Cache": "#EAB839", + "Swap": "#BF1B00", + "Swap_Cache": "#C15C17", + "Total": "#511749", + "Total RAM": "#052B51", + "Total RAM + Swap": "#052B51", + "VmallocUsed": "#EA6460" + }, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "decimals": 2, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 130 + }, + "hiddenSeries": false, + "id": 128, + "legend": { + "alignAsTable": true, + "avg": true, + "current": false, + "hideEmpty": false, + "hideZero": false, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sideWidth": null, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_memory_DirectMap1G_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "DirectMap1G - Amount of pages mapped as this size", + "refId": "A", + "step": 240 + }, + { + "expr": "node_memory_DirectMap2M_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "DirectMap2M - Amount of pages mapped as this size", + "refId": "B", + "step": 240 + }, + { + "expr": "node_memory_DirectMap4k_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "DirectMap4K - Amount of pages mapped as this size", + "refId": "C", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Memory DirectMap", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "bytes", + "label": "bytes", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": { + "Apps": "#629E51", + "Buffers": "#614D93", + "Cache": "#6D1F62", + "Cached": "#511749", + "Committed": "#508642", + "Free": "#0A437C", + "Hardware Corrupted - Amount of RAM that the kernel identified as corrupted / not working": "#CFFAFF", + "Inactive": "#584477", + "PageTables": "#0A50A1", + "Page_Tables": "#0A50A1", + "RAM_Free": "#E0F9D7", + "Slab": "#806EB7", + "Slab_Cache": "#E0752D", + "Swap": "#BF1B00", + "Swap_Cache": "#C15C17", + "Swap_Free": "#2F575E", + "Unused": "#EAB839" + }, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "decimals": 2, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 130 + }, + "hiddenSeries": false, + "id": 137, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sideWidth": 350, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_memory_Unevictable_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "Unevictable - Amount of unevictable memory that can't be swapped out for a variety of reasons", + "refId": "A", + "step": 240 + }, + { + "expr": "node_memory_Mlocked_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "MLocked - Size of pages locked to memory using the mlock() system call", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Memory Unevictable and MLocked", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "cumulative" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "bytes", + "label": "bytes", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": { + "Active": "#99440A", + "Buffers": "#58140C", + "Cache": "#6D1F62", + "Cached": "#511749", + "Committed": "#508642", + "Dirty": "#6ED0E0", + "Free": "#B7DBAB", + "Inactive": "#EA6460", + "Mapped": "#052B51", + "PageTables": "#0A50A1", + "Page_Tables": "#0A50A1", + "Slab_Cache": "#EAB839", + "Swap": "#BF1B00", + "Swap_Cache": "#C15C17", + "Total": "#511749", + "Total RAM": "#052B51", + "Total RAM + Swap": "#052B51", + "Total Swap": "#614D93", + "VmallocUsed": "#EA6460" + }, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "decimals": 2, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 140 + }, + "hiddenSeries": false, + "id": 132, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sideWidth": null, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_memory_NFS_Unstable_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "NFS Unstable - Memory in NFS pages sent to the server, but not yet commited to the storage", + "refId": "A", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Memory NFS", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "bytes", + "label": "bytes", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + } + ], + "repeat": null, + "title": "Memory Meminfo", + "type": "row" + }, + { + "collapsed": true, + "datasource": "${DS_PROMETHEUS}", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 22 + }, + "id": 267, + "panels": [ + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 23 + }, + "hiddenSeries": false, + "id": 176, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "alias": "/.*out/", + "transform": "negative-Y" + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_vmstat_pgpgin{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "Pagesin - Page in operations", + "refId": "A", + "step": 240 + }, + { + "expr": "rate(node_vmstat_pgpgout{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "Pagesout - Page out operations", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Memory Pages In / Out", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "label": "pages out (-) / in (+)", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 23 + }, + "hiddenSeries": false, + "id": 22, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "alias": "/.*out/", + "transform": "negative-Y" + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_vmstat_pswpin{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "Pswpin - Pages swapped in", + "refId": "A", + "step": 240 + }, + { + "expr": "rate(node_vmstat_pswpout{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "Pswpout - Pages swapped out", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Memory Pages Swap In / Out", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "label": "pages out (-) / in (+)", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": { + "Apps": "#629E51", + "Buffers": "#614D93", + "Cache": "#6D1F62", + "Cached": "#511749", + "Committed": "#508642", + "Free": "#0A437C", + "Hardware Corrupted - Amount of RAM that the kernel identified as corrupted / not working": "#CFFAFF", + "Inactive": "#584477", + "PageTables": "#0A50A1", + "Page_Tables": "#0A50A1", + "RAM_Free": "#E0F9D7", + "Slab": "#806EB7", + "Slab_Cache": "#E0752D", + "Swap": "#BF1B00", + "Swap_Cache": "#C15C17", + "Swap_Free": "#2F575E", + "Unused": "#EAB839" + }, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "decimals": 2, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 33 + }, + "hiddenSeries": false, + "id": 175, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sideWidth": 350, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "$$hashKey": "object:6118", + "alias": "Pgfault - Page major and minor fault operations", + "fill": 0, + "stack": false + } + ], + "spaceLength": 10, + "stack": true, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_vmstat_pgfault{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "Pgfault - Page major and minor fault operations", + "refId": "A", + "step": 240 + }, + { + "expr": "rate(node_vmstat_pgmajfault{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "Pgmajfault - Major page fault operations", + "refId": "B", + "step": 240 + }, + { + "expr": "rate(node_vmstat_pgfault{instance=\"$node\",job=\"$job\"}[$__rate_interval]) - rate(node_vmstat_pgmajfault{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "Pgminfault - Minor page fault operations", + "refId": "C", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Memory Page Faults", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "cumulative" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:6133", + "format": "short", + "label": "faults", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "$$hashKey": "object:6134", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": { + "Active": "#99440A", + "Buffers": "#58140C", + "Cache": "#6D1F62", + "Cached": "#511749", + "Committed": "#508642", + "Dirty": "#6ED0E0", + "Free": "#B7DBAB", + "Inactive": "#EA6460", + "Mapped": "#052B51", + "PageTables": "#0A50A1", + "Page_Tables": "#0A50A1", + "Slab_Cache": "#EAB839", + "Swap": "#BF1B00", + "Swap_Cache": "#C15C17", + "Total": "#511749", + "Total RAM": "#052B51", + "Total RAM + Swap": "#052B51", + "Total Swap": "#614D93", + "VmallocUsed": "#EA6460" + }, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "decimals": 2, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 33 + }, + "hiddenSeries": false, + "id": 307, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sideWidth": null, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_vmstat_oom_kill{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "oom killer invocations ", + "refId": "A", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "OOM Killer", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:5373", + "format": "short", + "label": "counter", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "$$hashKey": "object:5374", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + } + ], + "repeat": null, + "title": "Memory Vmstat", + "type": "row" + }, + { + "collapsed": true, + "datasource": "${DS_PROMETHEUS}", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 23 + }, + "id": 293, + "panels": [ + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "description": "", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 24 + }, + "hiddenSeries": false, + "id": 260, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "alias": "/.*Variation*./", + "color": "#890F02" + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_timex_estimated_error_seconds{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "hide": false, + "interval": "", + "intervalFactor": 1, + "legendFormat": "Estimated error in seconds", + "refId": "A", + "step": 240 + }, + { + "expr": "node_timex_offset_seconds{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "hide": false, + "interval": "", + "intervalFactor": 1, + "legendFormat": "Time offset in between local system and reference clock", + "refId": "B", + "step": 240 + }, + { + "expr": "node_timex_maxerror_seconds{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "hide": false, + "interval": "", + "intervalFactor": 1, + "legendFormat": "Maximum error in seconds", + "refId": "C", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Time Syncronized Drift", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "s", + "label": "seconds", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "format": "short", + "label": "counter", + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "description": "", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 24 + }, + "hiddenSeries": false, + "id": 291, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_timex_loop_time_constant{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "Phase-locked loop time adjust", + "refId": "A", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Time PLL Adjust", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "label": "counter", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "description": "", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 34 + }, + "hiddenSeries": false, + "id": 168, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "alias": "/.*Variation*./", + "color": "#890F02" + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_timex_sync_status{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "Is clock synchronized to a reliable server (1 = yes, 0 = no)", + "refId": "A", + "step": 240 + }, + { + "expr": "node_timex_frequency_adjustment_ratio{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "Local clock frequency adjustment", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Time Syncronized Status", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "label": "counter", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "description": "", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 34 + }, + "hiddenSeries": false, + "id": 294, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_timex_tick_seconds{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "Seconds between clock ticks", + "refId": "A", + "step": 240 + }, + { + "expr": "node_timex_tai_offset_seconds{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "International Atomic Time (TAI) offset", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Time Misc", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "s", + "label": "seconds", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + } + ], + "title": "System Timesync", + "type": "row" + }, + { + "collapsed": true, + "datasource": "${DS_PROMETHEUS}", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 24 + }, + "id": 312, + "panels": [ + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 7 + }, + "hiddenSeries": false, + "id": 62, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_procs_blocked{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "Processes blocked waiting for I/O to complete", + "refId": "A", + "step": 240 + }, + { + "expr": "node_procs_running{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "Processes in runnable state", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Processes Status", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:6500", + "format": "short", + "label": "counter", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "$$hashKey": "object:6501", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 7 + }, + "hiddenSeries": false, + "id": 315, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": true, + "steppedLine": false, + "targets": [ + { + "expr": "node_processes_state{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "{{ state }}", + "refId": "A", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Processes State", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:6500", + "format": "short", + "label": "counter", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "$$hashKey": "object:6501", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 17 + }, + "hiddenSeries": false, + "id": 148, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_forks_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "hide": false, + "intervalFactor": 1, + "legendFormat": "Processes forks second", + "refId": "A", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Processes Forks", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:6640", + "format": "short", + "label": "forks / sec", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "$$hashKey": "object:6641", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 17 + }, + "hiddenSeries": false, + "id": 149, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "alias": "/.*Max.*/", + "fill": 0 + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(process_virtual_memory_bytes{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "hide": false, + "interval": "", + "intervalFactor": 1, + "legendFormat": "Processes virtual memory size in bytes", + "refId": "A", + "step": 240 + }, + { + "expr": "process_resident_memory_max_bytes{instance=\"$node\",job=\"$job\"}", + "hide": false, + "interval": "", + "intervalFactor": 1, + "legendFormat": "Maximum amount of virtual memory available in bytes", + "refId": "B", + "step": 240 + }, + { + "expr": "rate(process_virtual_memory_bytes{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "hide": false, + "interval": "", + "intervalFactor": 1, + "legendFormat": "Processes virtual memory size in bytes", + "refId": "C", + "step": 240 + }, + { + "expr": "rate(process_virtual_memory_max_bytes{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "hide": false, + "interval": "", + "intervalFactor": 1, + "legendFormat": "Maximum amount of virtual memory available in bytes", + "refId": "D", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Processes Memory", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "decbytes", + "label": "bytes", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 27 + }, + "hiddenSeries": false, + "id": 313, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "$$hashKey": "object:709", + "alias": "PIDs limit", + "color": "#F2495C", + "fill": 0 + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_processes_pids{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "Number of PIDs", + "refId": "A", + "step": 240 + }, + { + "expr": "node_processes_max_processes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "PIDs limit", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "PIDs Number and Limit", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:6500", + "format": "short", + "label": "counter", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "$$hashKey": "object:6501", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 27 + }, + "hiddenSeries": false, + "id": 305, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "$$hashKey": "object:4963", + "alias": "/.*waiting.*/", + "transform": "negative-Y" + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_schedstat_running_seconds_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "CPU {{ cpu }} - seconds spent running a process", + "refId": "A", + "step": 240 + }, + { + "expr": "rate(node_schedstat_waiting_seconds_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "CPU {{ cpu }} - seconds spent by processing waiting for this CPU", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Process schedule stats Running / Waiting", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:4860", + "format": "s", + "label": "seconds", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "$$hashKey": "object:4861", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 37 + }, + "hiddenSeries": false, + "id": 314, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "$$hashKey": "object:709", + "alias": "Threads limit", + "color": "#F2495C", + "fill": 0 + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_processes_threads{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "Allocated threads", + "refId": "A", + "step": 240 + }, + { + "expr": "node_processes_max_threads{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "Threads limit", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Threads Number and Limit", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:6500", + "format": "short", + "label": "counter", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "$$hashKey": "object:6501", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + } + ], + "title": "System Processes", + "type": "row" + }, + { + "collapsed": true, + "datasource": "${DS_PROMETHEUS}", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 25 + }, + "id": 269, + "panels": [ + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 8 + }, + "hiddenSeries": false, + "id": 8, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "repeat": null, + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_context_switches_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "Context switches", + "refId": "A", + "step": 240 + }, + { + "expr": "rate(node_intr_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "hide": false, + "intervalFactor": 1, + "legendFormat": "Interrupts", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Context Switches / Interrupts", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "label": "counter", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 8 + }, + "hiddenSeries": false, + "id": 7, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "repeat": null, + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_load1{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 4, + "legendFormat": "Load 1m", + "refId": "A", + "step": 240 + }, + { + "expr": "node_load5{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 4, + "legendFormat": "Load 5m", + "refId": "B", + "step": 240 + }, + { + "expr": "node_load15{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 4, + "legendFormat": "Load 15m", + "refId": "C", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "System Load", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:6261", + "format": "short", + "label": "counter", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "$$hashKey": "object:6262", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 18 + }, + "hiddenSeries": false, + "id": 259, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "alias": "/.*Critical*./", + "color": "#E24D42", + "fill": 0 + }, + { + "alias": "/.*Max*./", + "color": "#EF843C", + "fill": 0 + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_interrupts_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "{{ type }} - {{ info }}", + "refId": "A", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Interrupts Detail", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "label": "counter", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 18 + }, + "hiddenSeries": false, + "id": 306, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_schedstat_timeslices_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "CPU {{ cpu }}", + "refId": "A", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Schedule timeslices executed by each cpu", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:4860", + "format": "short", + "label": "counter", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "$$hashKey": "object:4861", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 28 + }, + "hiddenSeries": false, + "id": 151, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_entropy_available_bits{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "Entropy available to random number generators", + "refId": "A", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Entropy", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:6568", + "format": "short", + "label": "counter", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "$$hashKey": "object:6569", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 28 + }, + "hiddenSeries": false, + "id": 308, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(process_cpu_seconds_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "Time spent", + "refId": "A", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "CPU time spent in user and system contexts", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:4860", + "format": "s", + "label": "seconds", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "$$hashKey": "object:4861", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 38 + }, + "hiddenSeries": false, + "id": 64, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "$$hashKey": "object:6323", + "alias": "/.*Max*./", + "color": "#890F02", + "fill": 0 + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "process_max_fds{instance=\"$node\",job=\"$job\"}", + "interval": "", + "intervalFactor": 1, + "legendFormat": "Maximum open file descriptors", + "refId": "A", + "step": 240 + }, + { + "expr": "process_open_fds{instance=\"$node\",job=\"$job\"}", + "interval": "", + "intervalFactor": 1, + "legendFormat": "Open file descriptors", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "File Descriptors", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:6338", + "format": "short", + "label": "counter", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "$$hashKey": "object:6339", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + } + ], + "repeat": null, + "title": "System Misc", + "type": "row" + }, + { + "collapsed": true, + "datasource": "${DS_PROMETHEUS}", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 26 + }, + "id": 304, + "panels": [ + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 26 + }, + "hiddenSeries": false, + "id": 158, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "$$hashKey": "object:6726", + "alias": "/.*Critical*./", + "color": "#E24D42", + "fill": 0 + }, + { + "$$hashKey": "object:6727", + "alias": "/.*Max*./", + "color": "#EF843C", + "fill": 0 + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_hwmon_temp_celsius{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "{{ chip }} {{ sensor }} temp", + "refId": "A", + "step": 240 + }, + { + "expr": "node_hwmon_temp_crit_alarm_celsius{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "hide": true, + "interval": "", + "intervalFactor": 1, + "legendFormat": "{{ chip }} {{ sensor }} Critical Alarm", + "refId": "B", + "step": 240 + }, + { + "expr": "node_hwmon_temp_crit_celsius{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "{{ chip }} {{ sensor }} Critical", + "refId": "C", + "step": 240 + }, + { + "expr": "node_hwmon_temp_crit_hyst_celsius{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "hide": true, + "interval": "", + "intervalFactor": 1, + "legendFormat": "{{ chip }} {{ sensor }} Critical Historical", + "refId": "D", + "step": 240 + }, + { + "expr": "node_hwmon_temp_max_celsius{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "hide": true, + "interval": "", + "intervalFactor": 1, + "legendFormat": "{{ chip }} {{ sensor }} Max", + "refId": "E", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Hardware temperature monitor", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:6750", + "format": "celsius", + "label": "temperature", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "$$hashKey": "object:6751", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 26 + }, + "hiddenSeries": false, + "id": 300, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "$$hashKey": "object:1655", + "alias": "/.*Max*./", + "color": "#EF843C", + "fill": 0 + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_cooling_device_cur_state{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "hide": false, + "interval": "", + "intervalFactor": 1, + "legendFormat": "Current {{ name }} in {{ type }}", + "refId": "A", + "step": 240 + }, + { + "expr": "node_cooling_device_max_state{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "Max {{ name }} in {{ type }}", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Throttle cooling device", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:1678", + "format": "short", + "label": "counter", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "$$hashKey": "object:1679", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 36 + }, + "hiddenSeries": false, + "id": 302, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_power_supply_online{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "hide": false, + "interval": "", + "intervalFactor": 1, + "legendFormat": "{{ power_supply }} online", + "refId": "A", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Power supply", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:1678", + "format": "short", + "label": "counter", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "$$hashKey": "object:1679", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + } + ], + "title": "Hardware Misc", + "type": "row" + }, + { + "collapsed": true, + "datasource": "${DS_PROMETHEUS}", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 27 + }, + "id": 296, + "panels": [ + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 10 + }, + "hiddenSeries": false, + "id": 297, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_systemd_socket_accepted_connections_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "{{ name }} Connections", + "refId": "A", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Systemd Sockets", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "label": "counter", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 10 + }, + "hiddenSeries": false, + "id": 298, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "alias": "Failed", + "color": "#F2495C" + }, + { + "alias": "Inactive", + "color": "#FF9830" + }, + { + "alias": "Active", + "color": "#73BF69" + }, + { + "alias": "Deactivating", + "color": "#FFCB7D" + }, + { + "alias": "Activating", + "color": "#C8F2C2" + } + ], + "spaceLength": 10, + "stack": true, + "steppedLine": false, + "targets": [ + { + "expr": "node_systemd_units{instance=\"$node\",job=\"$job\",state=\"activating\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "Activating", + "refId": "A", + "step": 240 + }, + { + "expr": "node_systemd_units{instance=\"$node\",job=\"$job\",state=\"active\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "Active", + "refId": "B", + "step": 240 + }, + { + "expr": "node_systemd_units{instance=\"$node\",job=\"$job\",state=\"deactivating\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "Deactivating", + "refId": "C", + "step": 240 + }, + { + "expr": "node_systemd_units{instance=\"$node\",job=\"$job\",state=\"failed\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "Failed", + "refId": "D", + "step": 240 + }, + { + "expr": "node_systemd_units{instance=\"$node\",job=\"$job\",state=\"inactive\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "Inactive", + "refId": "E", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Systemd Units State", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "label": "counter", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + } + ], + "title": "Systemd", + "type": "row" + }, + { + "collapsed": true, + "datasource": "${DS_PROMETHEUS}", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 28 + }, + "id": 270, + "panels": [ + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "description": "The number (after merges) of I/O requests completed per second for the device", + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 11 + }, + "hiddenSeries": false, + "id": 9, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideZero": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "repeat": null, + "seriesOverrides": [ + { + "$$hashKey": "object:2033", + "alias": "/.*Read.*/", + "transform": "negative-Y" + }, + { + "$$hashKey": "object:2034", + "alias": "/.*sda_.*/", + "color": "#7EB26D" + }, + { + "$$hashKey": "object:2035", + "alias": "/.*sdb_.*/", + "color": "#EAB839" + }, + { + "$$hashKey": "object:2036", + "alias": "/.*sdc_.*/", + "color": "#6ED0E0" + }, + { + "$$hashKey": "object:2037", + "alias": "/.*sdd_.*/", + "color": "#EF843C" + }, + { + "$$hashKey": "object:2038", + "alias": "/.*sde_.*/", + "color": "#E24D42" + }, + { + "$$hashKey": "object:2039", + "alias": "/.*sda1.*/", + "color": "#584477" + }, + { + "$$hashKey": "object:2040", + "alias": "/.*sda2_.*/", + "color": "#BA43A9" + }, + { + "$$hashKey": "object:2041", + "alias": "/.*sda3_.*/", + "color": "#F4D598" + }, + { + "$$hashKey": "object:2042", + "alias": "/.*sdb1.*/", + "color": "#0A50A1" + }, + { + "$$hashKey": "object:2043", + "alias": "/.*sdb2.*/", + "color": "#BF1B00" + }, + { + "$$hashKey": "object:2044", + "alias": "/.*sdb3.*/", + "color": "#E0752D" + }, + { + "$$hashKey": "object:2045", + "alias": "/.*sdc1.*/", + "color": "#962D82" + }, + { + "$$hashKey": "object:2046", + "alias": "/.*sdc2.*/", + "color": "#614D93" + }, + { + "$$hashKey": "object:2047", + "alias": "/.*sdc3.*/", + "color": "#9AC48A" + }, + { + "$$hashKey": "object:2048", + "alias": "/.*sdd1.*/", + "color": "#65C5DB" + }, + { + "$$hashKey": "object:2049", + "alias": "/.*sdd2.*/", + "color": "#F9934E" + }, + { + "$$hashKey": "object:2050", + "alias": "/.*sdd3.*/", + "color": "#EA6460" + }, + { + "$$hashKey": "object:2051", + "alias": "/.*sde1.*/", + "color": "#E0F9D7" + }, + { + "$$hashKey": "object:2052", + "alias": "/.*sdd2.*/", + "color": "#FCEACA" + }, + { + "$$hashKey": "object:2053", + "alias": "/.*sde3.*/", + "color": "#F9E2D2" + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_disk_reads_completed_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "intervalFactor": 4, + "legendFormat": "{{device}} - Reads completed", + "refId": "A", + "step": 240 + }, + { + "expr": "rate(node_disk_writes_completed_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "intervalFactor": 1, + "legendFormat": "{{device}} - Writes completed", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Disk IOps Completed", + "tooltip": { + "shared": false, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:2186", + "format": "iops", + "label": "IO read (-) / write (+)", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "$$hashKey": "object:2187", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "description": "The number of bytes read from or written to the device per second", + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 11 + }, + "hiddenSeries": false, + "id": 33, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideZero": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "alias": "/.*Read.*/", + "transform": "negative-Y" + }, + { + "alias": "/.*sda_.*/", + "color": "#7EB26D" + }, + { + "alias": "/.*sdb_.*/", + "color": "#EAB839" + }, + { + "alias": "/.*sdc_.*/", + "color": "#6ED0E0" + }, + { + "alias": "/.*sdd_.*/", + "color": "#EF843C" + }, + { + "alias": "/.*sde_.*/", + "color": "#E24D42" + }, + { + "alias": "/.*sda1.*/", + "color": "#584477" + }, + { + "alias": "/.*sda2_.*/", + "color": "#BA43A9" + }, + { + "alias": "/.*sda3_.*/", + "color": "#F4D598" + }, + { + "alias": "/.*sdb1.*/", + "color": "#0A50A1" + }, + { + "alias": "/.*sdb2.*/", + "color": "#BF1B00" + }, + { + "alias": "/.*sdb3.*/", + "color": "#E0752D" + }, + { + "alias": "/.*sdc1.*/", + "color": "#962D82" + }, + { + "alias": "/.*sdc2.*/", + "color": "#614D93" + }, + { + "alias": "/.*sdc3.*/", + "color": "#9AC48A" + }, + { + "alias": "/.*sdd1.*/", + "color": "#65C5DB" + }, + { + "alias": "/.*sdd2.*/", + "color": "#F9934E" + }, + { + "alias": "/.*sdd3.*/", + "color": "#EA6460" + }, + { + "alias": "/.*sde1.*/", + "color": "#E0F9D7" + }, + { + "alias": "/.*sdd2.*/", + "color": "#FCEACA" + }, + { + "alias": "/.*sde3.*/", + "color": "#F9E2D2" + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_disk_read_bytes_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "intervalFactor": 4, + "legendFormat": "{{device}} - Read bytes", + "refId": "A", + "step": 240 + }, + { + "expr": "rate(node_disk_written_bytes_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "{{device}} - Written bytes", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Disk R/W Data", + "tooltip": { + "shared": false, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:369", + "format": "Bps", + "label": "bytes read (-) / write (+)", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "$$hashKey": "object:370", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "description": "The average time for requests issued to the device to be served. This includes the time spent by the requests in queue and the time spent servicing them.", + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 3, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 21 + }, + "hiddenSeries": false, + "id": 37, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideZero": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sort": "current", + "sortDesc": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "alias": "/.*Read.*/", + "transform": "negative-Y" + }, + { + "alias": "/.*sda_.*/", + "color": "#7EB26D" + }, + { + "alias": "/.*sdb_.*/", + "color": "#EAB839" + }, + { + "alias": "/.*sdc_.*/", + "color": "#6ED0E0" + }, + { + "alias": "/.*sdd_.*/", + "color": "#EF843C" + }, + { + "alias": "/.*sde_.*/", + "color": "#E24D42" + }, + { + "alias": "/.*sda1.*/", + "color": "#584477" + }, + { + "alias": "/.*sda2_.*/", + "color": "#BA43A9" + }, + { + "alias": "/.*sda3_.*/", + "color": "#F4D598" + }, + { + "alias": "/.*sdb1.*/", + "color": "#0A50A1" + }, + { + "alias": "/.*sdb2.*/", + "color": "#BF1B00" + }, + { + "alias": "/.*sdb3.*/", + "color": "#E0752D" + }, + { + "alias": "/.*sdc1.*/", + "color": "#962D82" + }, + { + "alias": "/.*sdc2.*/", + "color": "#614D93" + }, + { + "alias": "/.*sdc3.*/", + "color": "#9AC48A" + }, + { + "alias": "/.*sdd1.*/", + "color": "#65C5DB" + }, + { + "alias": "/.*sdd2.*/", + "color": "#F9934E" + }, + { + "alias": "/.*sdd3.*/", + "color": "#EA6460" + }, + { + "alias": "/.*sde1.*/", + "color": "#E0F9D7" + }, + { + "alias": "/.*sdd2.*/", + "color": "#FCEACA" + }, + { + "alias": "/.*sde3.*/", + "color": "#F9E2D2" + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_disk_read_time_seconds_total{instance=\"$node\",job=\"$job\"}[$__rate_interval]) / rate(node_disk_reads_completed_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "hide": false, + "interval": "", + "intervalFactor": 4, + "legendFormat": "{{device}} - Read wait time avg", + "refId": "A", + "step": 240 + }, + { + "expr": "rate(node_disk_write_time_seconds_total{instance=\"$node\",job=\"$job\"}[$__rate_interval]) / rate(node_disk_writes_completed_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "hide": false, + "interval": "", + "intervalFactor": 1, + "legendFormat": "{{device}} - Write wait time avg", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Disk Average Wait Time", + "tooltip": { + "shared": false, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:441", + "format": "s", + "label": "time. read (-) / write (+)", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "$$hashKey": "object:442", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "description": "The average queue length of the requests that were issued to the device", + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 21 + }, + "hiddenSeries": false, + "id": 35, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideZero": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sort": "current", + "sortDesc": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "alias": "/.*sda_.*/", + "color": "#7EB26D" + }, + { + "alias": "/.*sdb_.*/", + "color": "#EAB839" + }, + { + "alias": "/.*sdc_.*/", + "color": "#6ED0E0" + }, + { + "alias": "/.*sdd_.*/", + "color": "#EF843C" + }, + { + "alias": "/.*sde_.*/", + "color": "#E24D42" + }, + { + "alias": "/.*sda1.*/", + "color": "#584477" + }, + { + "alias": "/.*sda2_.*/", + "color": "#BA43A9" + }, + { + "alias": "/.*sda3_.*/", + "color": "#F4D598" + }, + { + "alias": "/.*sdb1.*/", + "color": "#0A50A1" + }, + { + "alias": "/.*sdb2.*/", + "color": "#BF1B00" + }, + { + "alias": "/.*sdb3.*/", + "color": "#E0752D" + }, + { + "alias": "/.*sdc1.*/", + "color": "#962D82" + }, + { + "alias": "/.*sdc2.*/", + "color": "#614D93" + }, + { + "alias": "/.*sdc3.*/", + "color": "#9AC48A" + }, + { + "alias": "/.*sdd1.*/", + "color": "#65C5DB" + }, + { + "alias": "/.*sdd2.*/", + "color": "#F9934E" + }, + { + "alias": "/.*sdd3.*/", + "color": "#EA6460" + }, + { + "alias": "/.*sde1.*/", + "color": "#E0F9D7" + }, + { + "alias": "/.*sdd2.*/", + "color": "#FCEACA" + }, + { + "alias": "/.*sde3.*/", + "color": "#F9E2D2" + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_disk_io_time_weighted_seconds_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "interval": "", + "intervalFactor": 4, + "legendFormat": "{{device}}", + "refId": "A", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Average Queue Size", + "tooltip": { + "shared": false, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:513", + "format": "none", + "label": "aqu-sz", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "$$hashKey": "object:514", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "description": "The number of read and write requests merged per second that were queued to the device", + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 31 + }, + "hiddenSeries": false, + "id": 133, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideZero": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sort": "current", + "sortDesc": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "alias": "/.*Read.*/", + "transform": "negative-Y" + }, + { + "alias": "/.*sda_.*/", + "color": "#7EB26D" + }, + { + "alias": "/.*sdb_.*/", + "color": "#EAB839" + }, + { + "alias": "/.*sdc_.*/", + "color": "#6ED0E0" + }, + { + "alias": "/.*sdd_.*/", + "color": "#EF843C" + }, + { + "alias": "/.*sde_.*/", + "color": "#E24D42" + }, + { + "alias": "/.*sda1.*/", + "color": "#584477" + }, + { + "alias": "/.*sda2_.*/", + "color": "#BA43A9" + }, + { + "alias": "/.*sda3_.*/", + "color": "#F4D598" + }, + { + "alias": "/.*sdb1.*/", + "color": "#0A50A1" + }, + { + "alias": "/.*sdb2.*/", + "color": "#BF1B00" + }, + { + "alias": "/.*sdb3.*/", + "color": "#E0752D" + }, + { + "alias": "/.*sdc1.*/", + "color": "#962D82" + }, + { + "alias": "/.*sdc2.*/", + "color": "#614D93" + }, + { + "alias": "/.*sdc3.*/", + "color": "#9AC48A" + }, + { + "alias": "/.*sdd1.*/", + "color": "#65C5DB" + }, + { + "alias": "/.*sdd2.*/", + "color": "#F9934E" + }, + { + "alias": "/.*sdd3.*/", + "color": "#EA6460" + }, + { + "alias": "/.*sde1.*/", + "color": "#E0F9D7" + }, + { + "alias": "/.*sdd2.*/", + "color": "#FCEACA" + }, + { + "alias": "/.*sde3.*/", + "color": "#F9E2D2" + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_disk_reads_merged_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "intervalFactor": 1, + "legendFormat": "{{device}} - Read merged", + "refId": "A", + "step": 240 + }, + { + "expr": "rate(node_disk_writes_merged_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "intervalFactor": 1, + "legendFormat": "{{device}} - Write merged", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Disk R/W Merged", + "tooltip": { + "shared": false, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:585", + "format": "iops", + "label": "I/Os", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "$$hashKey": "object:586", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "description": "Percentage of elapsed time during which I/O requests were issued to the device (bandwidth utilization for the device). Device saturation occurs when this value is close to 100% for devices serving requests serially. But for devices serving requests in parallel, such as RAID arrays and modern SSDs, this number does not reflect their performance limits.", + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 3, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 31 + }, + "hiddenSeries": false, + "id": 36, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideZero": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sort": "current", + "sortDesc": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "alias": "/.*sda_.*/", + "color": "#7EB26D" + }, + { + "alias": "/.*sdb_.*/", + "color": "#EAB839" + }, + { + "alias": "/.*sdc_.*/", + "color": "#6ED0E0" + }, + { + "alias": "/.*sdd_.*/", + "color": "#EF843C" + }, + { + "alias": "/.*sde_.*/", + "color": "#E24D42" + }, + { + "alias": "/.*sda1.*/", + "color": "#584477" + }, + { + "alias": "/.*sda2_.*/", + "color": "#BA43A9" + }, + { + "alias": "/.*sda3_.*/", + "color": "#F4D598" + }, + { + "alias": "/.*sdb1.*/", + "color": "#0A50A1" + }, + { + "alias": "/.*sdb2.*/", + "color": "#BF1B00" + }, + { + "alias": "/.*sdb3.*/", + "color": "#E0752D" + }, + { + "alias": "/.*sdc1.*/", + "color": "#962D82" + }, + { + "alias": "/.*sdc2.*/", + "color": "#614D93" + }, + { + "alias": "/.*sdc3.*/", + "color": "#9AC48A" + }, + { + "alias": "/.*sdd1.*/", + "color": "#65C5DB" + }, + { + "alias": "/.*sdd2.*/", + "color": "#F9934E" + }, + { + "alias": "/.*sdd3.*/", + "color": "#EA6460" + }, + { + "alias": "/.*sde1.*/", + "color": "#E0F9D7" + }, + { + "alias": "/.*sdd2.*/", + "color": "#FCEACA" + }, + { + "alias": "/.*sde3.*/", + "color": "#F9E2D2" + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_disk_io_time_seconds_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "interval": "", + "intervalFactor": 4, + "legendFormat": "{{device}} - IO", + "refId": "A", + "step": 240 + }, + { + "expr": "rate(node_disk_discard_time_seconds_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "interval": "", + "intervalFactor": 4, + "legendFormat": "{{device}} - discard", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Time Spent Doing I/Os", + "tooltip": { + "shared": false, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:657", + "format": "percentunit", + "label": "%util", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "$$hashKey": "object:658", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "description": "The number of outstanding requests at the instant the sample was taken. Incremented as requests are given to appropriate struct request_queue and decremented as they finish.", + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 41 + }, + "hiddenSeries": false, + "id": 34, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideZero": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sort": "current", + "sortDesc": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "alias": "/.*sda_.*/", + "color": "#7EB26D" + }, + { + "alias": "/.*sdb_.*/", + "color": "#EAB839" + }, + { + "alias": "/.*sdc_.*/", + "color": "#6ED0E0" + }, + { + "alias": "/.*sdd_.*/", + "color": "#EF843C" + }, + { + "alias": "/.*sde_.*/", + "color": "#E24D42" + }, + { + "alias": "/.*sda1.*/", + "color": "#584477" + }, + { + "alias": "/.*sda2_.*/", + "color": "#BA43A9" + }, + { + "alias": "/.*sda3_.*/", + "color": "#F4D598" + }, + { + "alias": "/.*sdb1.*/", + "color": "#0A50A1" + }, + { + "alias": "/.*sdb2.*/", + "color": "#BF1B00" + }, + { + "alias": "/.*sdb3.*/", + "color": "#E0752D" + }, + { + "alias": "/.*sdc1.*/", + "color": "#962D82" + }, + { + "alias": "/.*sdc2.*/", + "color": "#614D93" + }, + { + "alias": "/.*sdc3.*/", + "color": "#9AC48A" + }, + { + "alias": "/.*sdd1.*/", + "color": "#65C5DB" + }, + { + "alias": "/.*sdd2.*/", + "color": "#F9934E" + }, + { + "alias": "/.*sdd3.*/", + "color": "#EA6460" + }, + { + "alias": "/.*sde1.*/", + "color": "#E0F9D7" + }, + { + "alias": "/.*sdd2.*/", + "color": "#FCEACA" + }, + { + "alias": "/.*sde3.*/", + "color": "#F9E2D2" + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_disk_io_now{instance=\"$node\",job=\"$job\"}", + "interval": "", + "intervalFactor": 4, + "legendFormat": "{{device}} - IO now", + "refId": "A", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Instantaneous Queue Size", + "tooltip": { + "shared": false, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:729", + "format": "none", + "label": "Outstanding req.", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "$$hashKey": "object:730", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "description": "", + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 41 + }, + "hiddenSeries": false, + "id": 301, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideZero": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null as zero", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "$$hashKey": "object:2034", + "alias": "/.*sda_.*/", + "color": "#7EB26D" + }, + { + "$$hashKey": "object:2035", + "alias": "/.*sdb_.*/", + "color": "#EAB839" + }, + { + "$$hashKey": "object:2036", + "alias": "/.*sdc_.*/", + "color": "#6ED0E0" + }, + { + "$$hashKey": "object:2037", + "alias": "/.*sdd_.*/", + "color": "#EF843C" + }, + { + "$$hashKey": "object:2038", + "alias": "/.*sde_.*/", + "color": "#E24D42" + }, + { + "$$hashKey": "object:2039", + "alias": "/.*sda1.*/", + "color": "#584477" + }, + { + "$$hashKey": "object:2040", + "alias": "/.*sda2_.*/", + "color": "#BA43A9" + }, + { + "$$hashKey": "object:2041", + "alias": "/.*sda3_.*/", + "color": "#F4D598" + }, + { + "$$hashKey": "object:2042", + "alias": "/.*sdb1.*/", + "color": "#0A50A1" + }, + { + "$$hashKey": "object:2043", + "alias": "/.*sdb2.*/", + "color": "#BF1B00" + }, + { + "$$hashKey": "object:2044", + "alias": "/.*sdb3.*/", + "color": "#E0752D" + }, + { + "$$hashKey": "object:2045", + "alias": "/.*sdc1.*/", + "color": "#962D82" + }, + { + "$$hashKey": "object:2046", + "alias": "/.*sdc2.*/", + "color": "#614D93" + }, + { + "$$hashKey": "object:2047", + "alias": "/.*sdc3.*/", + "color": "#9AC48A" + }, + { + "$$hashKey": "object:2048", + "alias": "/.*sdd1.*/", + "color": "#65C5DB" + }, + { + "$$hashKey": "object:2049", + "alias": "/.*sdd2.*/", + "color": "#F9934E" + }, + { + "$$hashKey": "object:2050", + "alias": "/.*sdd3.*/", + "color": "#EA6460" + }, + { + "$$hashKey": "object:2051", + "alias": "/.*sde1.*/", + "color": "#E0F9D7" + }, + { + "$$hashKey": "object:2052", + "alias": "/.*sdd2.*/", + "color": "#FCEACA" + }, + { + "$$hashKey": "object:2053", + "alias": "/.*sde3.*/", + "color": "#F9E2D2" + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_disk_discards_completed_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "interval": "", + "intervalFactor": 4, + "legendFormat": "{{device}} - Discards completed", + "refId": "A", + "step": 240 + }, + { + "expr": "rate(node_disk_discards_merged_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "interval": "", + "intervalFactor": 1, + "legendFormat": "{{device}} - Discards merged", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Disk IOps Discards completed / merged", + "tooltip": { + "shared": false, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:2186", + "format": "iops", + "label": "IOs", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "$$hashKey": "object:2187", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + } + ], + "repeat": null, + "title": "Storage Disk", + "type": "row" + }, + { + "collapsed": true, + "datasource": "${DS_PROMETHEUS}", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 29 + }, + "id": 271, + "panels": [ + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "decimals": 3, + "description": "", + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 12 + }, + "hiddenSeries": false, + "id": 43, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_filesystem_avail_bytes{instance=\"$node\",job=\"$job\",device!~'rootfs'}", + "format": "time_series", + "hide": false, + "intervalFactor": 1, + "legendFormat": "{{mountpoint}} - Available", + "metric": "", + "refId": "A", + "step": 240 + }, + { + "expr": "node_filesystem_free_bytes{instance=\"$node\",job=\"$job\",device!~'rootfs'}", + "format": "time_series", + "hide": true, + "intervalFactor": 1, + "legendFormat": "{{mountpoint}} - Free", + "refId": "B", + "step": 240 + }, + { + "expr": "node_filesystem_size_bytes{instance=\"$node\",job=\"$job\",device!~'rootfs'}", + "format": "time_series", + "hide": true, + "intervalFactor": 1, + "legendFormat": "{{mountpoint}} - Size", + "refId": "C", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Filesystem space available", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:3826", + "format": "bytes", + "label": "bytes", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "$$hashKey": "object:3827", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "description": "", + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 12 + }, + "hiddenSeries": false, + "id": 41, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideZero": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_filesystem_files_free{instance=\"$node\",job=\"$job\",device!~'rootfs'}", + "format": "time_series", + "hide": false, + "intervalFactor": 1, + "legendFormat": "{{mountpoint}} - Free file nodes", + "refId": "A", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "File Nodes Free", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:3894", + "format": "short", + "label": "file nodes", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "$$hashKey": "object:3895", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "description": "", + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 22 + }, + "hiddenSeries": false, + "id": 28, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_filefd_maximum{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 4, + "legendFormat": "Max open files", + "refId": "A", + "step": 240 + }, + { + "expr": "node_filefd_allocated{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "Open files", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "File Descriptor", + "tooltip": { + "shared": false, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "label": "files", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "description": "", + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 22 + }, + "hiddenSeries": false, + "id": 219, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideZero": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_filesystem_files{instance=\"$node\",job=\"$job\",device!~'rootfs'}", + "format": "time_series", + "hide": false, + "intervalFactor": 1, + "legendFormat": "{{mountpoint}} - File nodes total", + "refId": "A", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "File Nodes Size", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "label": "file Nodes", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": { + "/ ReadOnly": "#890F02" + }, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "decimals": null, + "description": "", + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 32 + }, + "hiddenSeries": false, + "id": 44, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideEmpty": true, + "hideZero": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 6, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": true, + "steppedLine": false, + "targets": [ + { + "expr": "node_filesystem_readonly{instance=\"$node\",job=\"$job\",device!~'rootfs'}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "{{mountpoint}} - ReadOnly", + "refId": "A", + "step": 240 + }, + { + "expr": "node_filesystem_device_error{instance=\"$node\",job=\"$job\",device!~'rootfs',fstype!~'tmpfs'}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "{{mountpoint}} - Device error", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Filesystem in ReadOnly / Error", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:3670", + "format": "short", + "label": "counter", + "logBase": 1, + "max": "1", + "min": "0", + "show": true + }, + { + "$$hashKey": "object:3671", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + } + ], + "repeat": null, + "title": "Storage Filesystem", + "type": "row" + }, + { + "collapsed": true, + "datasource": "${DS_PROMETHEUS}", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 30 + }, + "id": 272, + "panels": [ + { + "aliasColors": { + "receive_packets_eth0": "#7EB26D", + "receive_packets_lo": "#E24D42", + "transmit_packets_eth0": "#7EB26D", + "transmit_packets_lo": "#E24D42" + }, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 30 + }, + "hiddenSeries": false, + "id": 60, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sideWidth": 300, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "alias": "/.*Trans.*/", + "transform": "negative-Y" + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_network_receive_packets_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "{{device}} - Receive", + "refId": "A", + "step": 240 + }, + { + "expr": "rate(node_network_transmit_packets_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "{{device}} - Transmit", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Network Traffic by Packets", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "pps", + "label": "packets out (-) / in (+)", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 30 + }, + "hiddenSeries": false, + "id": 142, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideEmpty": false, + "hideZero": false, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sideWidth": 300, + "sort": "current", + "sortDesc": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "alias": "/.*Trans.*/", + "transform": "negative-Y" + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_network_receive_errs_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "{{device}} - Receive errors", + "refId": "A", + "step": 240 + }, + { + "expr": "rate(node_network_transmit_errs_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "{{device}} - Rransmit errors", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Network Traffic Errors", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "pps", + "label": "packets out (-) / in (+)", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 40 + }, + "hiddenSeries": false, + "id": 143, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideEmpty": false, + "hideZero": false, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sideWidth": 300, + "sort": "current", + "sortDesc": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "alias": "/.*Trans.*/", + "transform": "negative-Y" + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_network_receive_drop_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "{{device}} - Receive drop", + "refId": "A", + "step": 240 + }, + { + "expr": "rate(node_network_transmit_drop_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "{{device}} - Transmit drop", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Network Traffic Drop", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "pps", + "label": "packets out (-) / in (+)", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 40 + }, + "hiddenSeries": false, + "id": 141, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideEmpty": false, + "hideZero": false, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sideWidth": 300, + "sort": "current", + "sortDesc": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "alias": "/.*Trans.*/", + "transform": "negative-Y" + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_network_receive_compressed_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "{{device}} - Receive compressed", + "refId": "A", + "step": 240 + }, + { + "expr": "rate(node_network_transmit_compressed_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "{{device}} - Transmit compressed", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Network Traffic Compressed", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "pps", + "label": "packets out (-) / in (+)", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 50 + }, + "hiddenSeries": false, + "id": 146, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideEmpty": false, + "hideZero": false, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sideWidth": 300, + "sort": "current", + "sortDesc": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "alias": "/.*Trans.*/", + "transform": "negative-Y" + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_network_receive_multicast_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "{{device}} - Receive multicast", + "refId": "A", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Network Traffic Multicast", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "pps", + "label": "packets out (-) / in (+)", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 50 + }, + "hiddenSeries": false, + "id": 144, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideEmpty": false, + "hideZero": false, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sideWidth": 300, + "sort": "current", + "sortDesc": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "alias": "/.*Trans.*/", + "transform": "negative-Y" + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_network_receive_fifo_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "{{device}} - Receive fifo", + "refId": "A", + "step": 240 + }, + { + "expr": "rate(node_network_transmit_fifo_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "{{device}} - Transmit fifo", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Network Traffic Fifo", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "pps", + "label": "packets out (-) / in (+)", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 60 + }, + "hiddenSeries": false, + "id": 145, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideEmpty": false, + "hideZero": false, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sideWidth": 300, + "sort": "current", + "sortDesc": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "$$hashKey": "object:576", + "alias": "/.*Trans.*/", + "transform": "negative-Y" + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_network_receive_frame_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "hide": false, + "intervalFactor": 1, + "legendFormat": "{{device}} - Receive frame", + "refId": "A", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Network Traffic Frame", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:589", + "format": "pps", + "label": "packets out (-) / in (+)", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "$$hashKey": "object:590", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 60 + }, + "hiddenSeries": false, + "id": 231, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideEmpty": false, + "hideZero": false, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sideWidth": 300, + "sort": "current", + "sortDesc": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_network_transmit_carrier_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "{{device}} - Statistic transmit_carrier", + "refId": "A", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Network Traffic Carrier", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "label": "counter", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 70 + }, + "hiddenSeries": false, + "id": 232, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideEmpty": false, + "hideZero": false, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sideWidth": 300, + "sort": "current", + "sortDesc": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "alias": "/.*Trans.*/", + "transform": "negative-Y" + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_network_transmit_colls_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "{{device}} - Transmit colls", + "refId": "A", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Network Traffic Colls", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "label": "counter", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 70 + }, + "hiddenSeries": false, + "id": 61, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "$$hashKey": "object:663", + "alias": "NF conntrack limit", + "color": "#890F02", + "fill": 0 + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_nf_conntrack_entries{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "NF conntrack entries", + "refId": "A", + "step": 240 + }, + { + "expr": "node_nf_conntrack_entries_limit{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "NF conntrack limit", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "NF Contrack", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:678", + "format": "short", + "label": "entries", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "$$hashKey": "object:679", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 80 + }, + "hiddenSeries": false, + "id": 230, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_arp_entries{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "{{ device }} - ARP entries", + "refId": "A", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "ARP Entries", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "label": "Entries", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 80 + }, + "hiddenSeries": false, + "id": 288, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 1, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_network_mtu_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "{{ device }} - Bytes", + "refId": "A", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "MTU", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "decimals": 0, + "format": "bytes", + "label": "bytes", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 90 + }, + "hiddenSeries": false, + "id": 280, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 1, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_network_speed_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "{{ device }} - Speed", + "refId": "A", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Speed", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "decimals": 0, + "format": "bytes", + "label": "bytes", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 90 + }, + "hiddenSeries": false, + "id": 289, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 1, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_network_transmit_queue_length{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "{{ device }} - Interface transmit queue length", + "refId": "A", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Queue Length", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "decimals": 0, + "format": "none", + "label": "packets", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 100 + }, + "hiddenSeries": false, + "id": 290, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideEmpty": false, + "hideZero": false, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sideWidth": 300, + "sort": "current", + "sortDesc": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "$$hashKey": "object:232", + "alias": "/.*Dropped.*/", + "transform": "negative-Y" + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_softnet_processed_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "CPU {{cpu}} - Processed", + "refId": "A", + "step": 240 + }, + { + "expr": "rate(node_softnet_dropped_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "CPU {{cpu}} - Dropped", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Softnet Packets", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:207", + "format": "short", + "label": "packetes drop (-) / process (+)", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "$$hashKey": "object:208", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 100 + }, + "hiddenSeries": false, + "id": 310, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideEmpty": false, + "hideZero": false, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sideWidth": 300, + "sort": "current", + "sortDesc": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_softnet_times_squeezed_total{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "CPU {{cpu}} - Squeezed", + "refId": "A", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Softnet Out of Quota", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:207", + "format": "short", + "label": "counter", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "$$hashKey": "object:208", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 110 + }, + "hiddenSeries": false, + "id": 309, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideEmpty": false, + "hideZero": false, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sideWidth": 300, + "sort": "current", + "sortDesc": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_network_up{operstate=\"up\",instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "{{interface}} - Operational state UP", + "refId": "A", + "step": 240 + }, + { + "expr": "node_network_carrier{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "instant": false, + "legendFormat": "{{device}} - Physical link state", + "refId": "B" + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Network Operational Status", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "label": "counter", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + } + ], + "repeat": null, + "title": "Network Traffic", + "type": "row" + }, + { + "collapsed": true, + "datasource": "${DS_PROMETHEUS}", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 31 + }, + "id": 273, + "panels": [ + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 32 + }, + "hiddenSeries": false, + "id": 63, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideEmpty": false, + "hideZero": false, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sideWidth": 300, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_sockstat_TCP_alloc{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "TCP_alloc - Allocated sockets", + "refId": "A", + "step": 240 + }, + { + "expr": "node_sockstat_TCP_inuse{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "TCP_inuse - Tcp sockets currently in use", + "refId": "B", + "step": 240 + }, + { + "expr": "node_sockstat_TCP_mem{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "hide": true, + "interval": "", + "intervalFactor": 1, + "legendFormat": "TCP_mem - Used memory for tcp", + "refId": "C", + "step": 240 + }, + { + "expr": "node_sockstat_TCP_orphan{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "TCP_orphan - Orphan sockets", + "refId": "D", + "step": 240 + }, + { + "expr": "node_sockstat_TCP_tw{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "TCP_tw - Sockets wating close", + "refId": "E", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Sockstat TCP", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "label": "counter", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 32 + }, + "hiddenSeries": false, + "id": 124, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideEmpty": false, + "hideZero": false, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sideWidth": 300, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_sockstat_UDPLITE_inuse{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "UDPLITE_inuse - Udplite sockets currently in use", + "refId": "A", + "step": 240 + }, + { + "expr": "node_sockstat_UDP_inuse{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "UDP_inuse - Udp sockets currently in use", + "refId": "B", + "step": 240 + }, + { + "expr": "node_sockstat_UDP_mem{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "UDP_mem - Used memory for udp", + "refId": "C", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Sockstat UDP", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "label": "counter", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 42 + }, + "hiddenSeries": false, + "id": 125, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideEmpty": false, + "hideZero": false, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sideWidth": 300, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_sockstat_FRAG_inuse{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "FRAG_inuse - Frag sockets currently in use", + "refId": "A", + "step": 240 + }, + { + "expr": "node_sockstat_RAW_inuse{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "RAW_inuse - Raw sockets currently in use", + "refId": "C", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Sockstat FRAG / RAW", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:1572", + "format": "short", + "label": "counter", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "$$hashKey": "object:1573", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 42 + }, + "hiddenSeries": false, + "id": 220, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideEmpty": false, + "hideZero": false, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sideWidth": 300, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_sockstat_TCP_mem_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "mem_bytes - TCP sockets in that state", + "refId": "A", + "step": 240 + }, + { + "expr": "node_sockstat_UDP_mem_bytes{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "mem_bytes - UDP sockets in that state", + "refId": "B", + "step": 240 + }, + { + "expr": "node_sockstat_FRAG_memory{instance=\"$node\",job=\"$job\"}", + "interval": "", + "intervalFactor": 1, + "legendFormat": "FRAG_memory - Used memory for frag", + "refId": "C" + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Sockstat Memory Size", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "bytes", + "label": "bytes", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 52 + }, + "hiddenSeries": false, + "id": 126, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideEmpty": false, + "hideZero": false, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sideWidth": 300, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_sockstat_sockets_used{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "Sockets_used - Sockets currently in use", + "refId": "A", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Sockstat Used", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "label": "sockets", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + } + ], + "repeat": null, + "title": "Network Sockstat", + "type": "row" + }, + { + "collapsed": true, + "datasource": "${DS_PROMETHEUS}", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 32 + }, + "id": 274, + "panels": [ + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 33 + }, + "height": "", + "hiddenSeries": false, + "id": 221, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideEmpty": false, + "hideZero": false, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sideWidth": 300, + "sort": "current", + "sortDesc": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 12, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "$$hashKey": "object:1876", + "alias": "/.*Out.*/", + "transform": "negative-Y" + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_netstat_IpExt_InOctets{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "InOctets - Received octets", + "refId": "A", + "step": 240 + }, + { + "expr": "rate(node_netstat_IpExt_OutOctets{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "OutOctets - Sent octets", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Netstat IP In / Out Octets", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:1889", + "format": "short", + "label": "octects out (-) / in (+)", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "$$hashKey": "object:1890", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 33 + }, + "height": "", + "hiddenSeries": false, + "id": 81, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideEmpty": false, + "hideZero": false, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sideWidth": 300, + "sort": "current", + "sortDesc": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_netstat_Ip_Forwarding{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "Forwarding - IP forwarding", + "refId": "A", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Netstat IP Forwarding", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:1957", + "format": "short", + "label": "datagrams", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "$$hashKey": "object:1958", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "decimals": null, + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 43 + }, + "height": "", + "hiddenSeries": false, + "id": 115, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideZero": false, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sort": "current", + "sortDesc": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 12, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "alias": "/.*Out.*/", + "transform": "negative-Y" + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_netstat_Icmp_InMsgs{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "InMsgs - Messages which the entity received. Note that this counter includes all those counted by icmpInErrors", + "refId": "A", + "step": 240 + }, + { + "expr": "rate(node_netstat_Icmp_OutMsgs{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "OutMsgs - Messages which this entity attempted to send. Note that this counter includes all those counted by icmpOutErrors", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "ICMP In / Out", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "label": "messages out (-) / in (+)", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "decimals": null, + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 43 + }, + "height": "", + "hiddenSeries": false, + "id": 50, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideZero": false, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sort": "current", + "sortDesc": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 12, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "alias": "/.*Out.*/", + "transform": "negative-Y" + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_netstat_Icmp_InErrors{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "InErrors - Messages which the entity received but determined as having ICMP-specific errors (bad ICMP checksums, bad length, etc.)", + "refId": "A", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "ICMP Errors", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "label": "messages out (-) / in (+)", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "decimals": null, + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 53 + }, + "height": "", + "hiddenSeries": false, + "id": 55, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideZero": false, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sort": "current", + "sortDesc": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 12, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "alias": "/.*Out.*/", + "transform": "negative-Y" + }, + { + "alias": "/.*Snd.*/", + "transform": "negative-Y" + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_netstat_Udp_InDatagrams{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "InDatagrams - Datagrams received", + "refId": "A", + "step": 240 + }, + { + "expr": "rate(node_netstat_Udp_OutDatagrams{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "OutDatagrams - Datagrams sent", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "UDP In / Out", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "label": "datagrams out (-) / in (+)", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 53 + }, + "height": "", + "hiddenSeries": false, + "id": 109, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideZero": false, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sort": "current", + "sortDesc": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 12, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_netstat_Udp_InErrors{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "InErrors - UDP Datagrams that could not be delivered to an application", + "refId": "A", + "step": 240 + }, + { + "expr": "rate(node_netstat_Udp_NoPorts{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "NoPorts - UDP Datagrams received on a port with no listener", + "refId": "B", + "step": 240 + }, + { + "expr": "rate(node_netstat_UdpLite_InErrors{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "interval": "", + "legendFormat": "InErrors Lite - UDPLite Datagrams that could not be delivered to an application", + "refId": "C" + }, + { + "expr": "rate(node_netstat_Udp_RcvbufErrors{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "RcvbufErrors - UDP buffer errors received", + "refId": "D", + "step": 240 + }, + { + "expr": "rate(node_netstat_Udp_SndbufErrors{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "SndbufErrors - UDP buffer errors send", + "refId": "E", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "UDP Errors", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:4232", + "format": "short", + "label": "datagrams", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "$$hashKey": "object:4233", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "decimals": null, + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 63 + }, + "height": "", + "hiddenSeries": false, + "id": 299, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideZero": false, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sort": "current", + "sortDesc": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 12, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "alias": "/.*Out.*/", + "transform": "negative-Y" + }, + { + "alias": "/.*Snd.*/", + "transform": "negative-Y" + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_netstat_Tcp_InSegs{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "instant": false, + "interval": "", + "intervalFactor": 1, + "legendFormat": "InSegs - Segments received, including those received in error. This count includes segments received on currently established connections", + "refId": "A", + "step": 240 + }, + { + "expr": "rate(node_netstat_Tcp_OutSegs{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "OutSegs - Segments sent, including those on current connections but excluding those containing only retransmitted octets", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "TCP In / Out", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "label": "datagrams out (-) / in (+)", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "description": "", + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 63 + }, + "height": "", + "hiddenSeries": false, + "id": 104, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideEmpty": false, + "hideZero": false, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sort": "current", + "sortDesc": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 12, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_netstat_TcpExt_ListenOverflows{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "hide": false, + "interval": "", + "intervalFactor": 1, + "legendFormat": "ListenOverflows - Times the listen queue of a socket overflowed", + "refId": "A", + "step": 240 + }, + { + "expr": "rate(node_netstat_TcpExt_ListenDrops{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "hide": false, + "interval": "", + "intervalFactor": 1, + "legendFormat": "ListenDrops - SYNs to LISTEN sockets ignored", + "refId": "B", + "step": 240 + }, + { + "expr": "rate(node_netstat_TcpExt_TCPSynRetrans{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "TCPSynRetrans - SYN-SYN/ACK retransmits to break down retransmissions in SYN, fast/timeout retransmits", + "refId": "C", + "step": 240 + }, + { + "expr": "rate(node_netstat_Tcp_RetransSegs{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "interval": "", + "legendFormat": "RetransSegs - Segments retransmitted - that is, the number of TCP segments transmitted containing one or more previously transmitted octets", + "refId": "D" + }, + { + "expr": "rate(node_netstat_Tcp_InErrs{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "interval": "", + "legendFormat": "InErrs - Segments received in error (e.g., bad TCP checksums)", + "refId": "E" + }, + { + "expr": "rate(node_netstat_Tcp_OutRsts{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "interval": "", + "legendFormat": "OutRsts - Segments sent with RST flag", + "refId": "F" + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "TCP Errors", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "label": "counter", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 73 + }, + "height": "", + "hiddenSeries": false, + "id": 85, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideZero": false, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sortDesc": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 12, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "$$hashKey": "object:454", + "alias": "/.*MaxConn *./", + "color": "#890F02", + "fill": 0 + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "node_netstat_Tcp_CurrEstab{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "hide": false, + "interval": "", + "intervalFactor": 1, + "legendFormat": "CurrEstab - TCP connections for which the current state is either ESTABLISHED or CLOSE- WAIT", + "refId": "A", + "step": 240 + }, + { + "expr": "node_netstat_Tcp_MaxConn{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "hide": false, + "interval": "", + "intervalFactor": 1, + "legendFormat": "MaxConn - Limit on the total number of TCP connections the entity can support (Dinamic is \"-1\")", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "TCP Connections", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:469", + "format": "short", + "label": "connections", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "$$hashKey": "object:470", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "description": "", + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 73 + }, + "height": "", + "hiddenSeries": false, + "id": 91, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideEmpty": false, + "hideZero": false, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sort": "current", + "sortDesc": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 12, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "alias": "/.*Sent.*/", + "transform": "negative-Y" + } + ], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_netstat_TcpExt_SyncookiesFailed{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "hide": false, + "interval": "", + "intervalFactor": 1, + "legendFormat": "SyncookiesFailed - Invalid SYN cookies received", + "refId": "A", + "step": 240 + }, + { + "expr": "rate(node_netstat_TcpExt_SyncookiesRecv{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "hide": false, + "interval": "", + "intervalFactor": 1, + "legendFormat": "SyncookiesRecv - SYN cookies received", + "refId": "B", + "step": 240 + }, + { + "expr": "rate(node_netstat_TcpExt_SyncookiesSent{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "hide": false, + "interval": "", + "intervalFactor": 1, + "legendFormat": "SyncookiesSent - SYN cookies sent", + "refId": "C", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "TCP SynCookie", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "label": "counter out (-) / in (+)", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [] + }, + "overrides": [] + }, + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 83 + }, + "height": "", + "hiddenSeries": false, + "id": 82, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "hideZero": false, + "max": true, + "min": true, + "rightSide": false, + "show": true, + "sortDesc": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "maxPerRow": 12, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.7", + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(node_netstat_Tcp_ActiveOpens{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "ActiveOpens - TCP connections that have made a direct transition to the SYN-SENT state from the CLOSED state", + "refId": "A", + "step": 240 + }, + { + "expr": "rate(node_netstat_Tcp_PassiveOpens{instance=\"$node\",job=\"$job\"}[$__rate_interval])", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "PassiveOpens - TCP connections that have made a direct transition to the SYN-RCVD state from the LISTEN state", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "TCP Direct Transition", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "label": "connections", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + } + ], + "repeat": null, + "title": "Network Netstat", + "type": "row" + }, + { + "collapsed": true, + "datasource": "${DS_PROMETHEUS}", + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 33 + }, + "id": 279, + "panels": [ + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "description": "", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 0, + "y": 54 + }, + "hiddenSeries": false, + "id": 40, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "show": true, + "sort": "current", + "sortDesc": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": true, + "steppedLine": false, + "targets": [ + { + "expr": "node_scrape_collector_duration_seconds{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "hide": false, + "interval": "", + "intervalFactor": 1, + "legendFormat": "{{collector}} - Scrape duration", + "refId": "A", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Node Exporter Scrape Time", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "s", + "label": "seconds", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "description": "", + "fill": 2, + "fillGradient": 0, + "gridPos": { + "h": 10, + "w": 12, + "x": 12, + "y": 54 + }, + "hiddenSeries": false, + "id": 157, + "legend": { + "alignAsTable": true, + "avg": true, + "current": true, + "max": true, + "min": true, + "show": true, + "total": false, + "values": true + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null", + "options": { + "dataLinks": [] + }, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [ + { + "$$hashKey": "object:1969", + "alias": "/.*error.*/", + "color": "#F2495C", + "transform": "negative-Y" + } + ], + "spaceLength": 10, + "stack": true, + "steppedLine": false, + "targets": [ + { + "expr": "node_scrape_collector_success{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "hide": false, + "interval": "", + "intervalFactor": 1, + "legendFormat": "{{collector}} - Scrape success", + "refId": "A", + "step": 240 + }, + { + "expr": "node_textfile_scrape_error{instance=\"$node\",job=\"$job\"}", + "format": "time_series", + "hide": false, + "interval": "", + "intervalFactor": 1, + "legendFormat": "{{collector}} - Scrape textfile error (1 = true)", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Node Exporter Scrape", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:1484", + "format": "short", + "label": "counter", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "$$hashKey": "object:1485", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + } + ], + "repeat": null, + "title": "Node Exporter", + "type": "row" + } + ], + "refresh": "1m", + "schemaVersion": 26, + "style": "dark", + "tags": [ + "linux" + ], + "templating": { + "list": [ + { + "current": { + "selected": false, + "text": "default", + "value": "default" + }, + "error": null, + "hide": 0, + "includeAll": false, + "label": "datasource", + "multi": false, + "name": "DS_PROMETHEUS", + "options": [], + "query": "prometheus", + "refresh": 1, + "regex": "", + "skipUrlSync": false, + "type": "datasource" + }, + { + "allValue": null, + "current": {}, + "datasource": "${DS_PROMETHEUS}", + "definition": "", + "error": null, + "hide": 0, + "includeAll": false, + "label": "Job", + "multi": false, + "name": "job", + "options": [], + "query": "label_values(node_uname_info, job)", + "refresh": 1, + "regex": "", + "skipUrlSync": false, + "sort": 1, + "tagValuesQuery": "", + "tags": [], + "tagsQuery": "", + "type": "query", + "useTags": false + }, + { + "allValue": null, + "current": {}, + "datasource": "${DS_PROMETHEUS}", + "definition": "label_values(node_uname_info{job=\"$job\"}, instance)", + "error": null, + "hide": 0, + "includeAll": false, + "label": "Host:", + "multi": false, + "name": "node", + "options": [], + "query": "label_values(node_uname_info{job=\"$job\"}, instance)", + "refresh": 1, + "regex": "", + "skipUrlSync": false, + "sort": 1, + "tagValuesQuery": "", + "tags": [], + "tagsQuery": "", + "type": "query", + "useTags": false + }, + { + "allValue": null, + "current": { + "selected": false, + "text": "[a-z]+|nvme[0-9]+n[0-9]+|mmcblk[0-9]+", + "value": "[a-z]+|nvme[0-9]+n[0-9]+|mmcblk[0-9]+" + }, + "error": null, + "hide": 2, + "includeAll": false, + "label": null, + "multi": false, + "name": "diskdevices", + "options": [ + { + "selected": true, + "text": "[a-z]+|nvme[0-9]+n[0-9]+|mmcblk[0-9]+", + "value": "[a-z]+|nvme[0-9]+n[0-9]+|mmcblk[0-9]+" + } + ], + "query": "[a-z]+|nvme[0-9]+n[0-9]+|mmcblk[0-9]+", + "skipUrlSync": false, + "type": "custom" + } + ] + }, + "time": { + "from": "now-24h", + "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": "browser", + "title": "Node Exporter Full", + "uid": "rYdddlPWk", + "version": 67 +} \ No newline at end of file diff --git a/conf/grafana/provisioning/dashboards/blackbox.json b/conf/grafana/provisioning/dashboards/blackbox.json new file mode 100644 index 0000000..d6a2296 --- /dev/null +++ b/conf/grafana/provisioning/dashboards/blackbox.json @@ -0,0 +1,1383 @@ +{ + "__inputs": [ + { + "name": "DS_PROMETHEUS", + "label": "Prometheus", + "description": "", + "type": "datasource", + "pluginId": "prometheus", + "pluginName": "Prometheus" + } + ], + "__requires": [ + { + "type": "grafana", + "id": "grafana", + "name": "Grafana", + "version": "8.1.0" + }, + { + "type": "datasource", + "id": "prometheus", + "name": "Prometheus", + "version": "1.0.0" + }, + { + "type": "panel", + "id": "stat", + "name": "Stat", + "version": "" + }, + { + "type": "panel", + "id": "table", + "name": "Table", + "version": "" + }, + { + "type": "panel", + "id": "timeseries", + "name": "Time series", + "version": "" + } + ], + "annotations": { + "list": [ + { + "builtIn": 1, + "datasource": "-- Grafana --", + "enable": true, + "hide": true, + "iconColor": "rgba(0, 211, 255, 1)", + "name": "Annotations & Alerts", + "target": { + "limit": 100, + "matchAny": false, + "tags": [ + "blackbox", + "prometheus" + ], + "type": "dashboard" + }, + "type": "dashboard" + } + ] + }, + "description": "Prometheus Blackbox Exporter Overview", + "editable": true, + "graphTooltip": 0, + "id": null, + "links": [], + "panels": [ + { + "datasource": "${DS_PROMETHEUS}", + "fieldConfig": { + "defaults": { + "color": { + "mode": "thresholds" + }, + "custom": { + "axisLabel": "", + "axisPlacement": "auto", + "axisSoftMax": 599, + "barAlignment": 0, + "drawStyle": "line", + "fillOpacity": 10, + "gradientMode": "none", + "hideFrom": { + "legend": false, + "tooltip": false, + "viz": false + }, + "lineInterpolation": "smooth", + "lineStyle": { + "fill": "solid" + }, + "lineWidth": 1, + "pointSize": 5, + "scaleDistribution": { + "type": "linear" + }, + "showPoints": "always", + "spanNulls": true, + "stacking": { + "group": "A", + "mode": "none" + }, + "thresholdsStyle": { + "mode": "area" + } + }, + "links": [], + "mappings": [], + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green", + "value": null + }, + { + "color": "blue", + "value": 300 + }, + { + "color": "orange", + "value": 400 + }, + { + "color": "red", + "value": 500 + } + ] + }, + "unit": "none" + }, + "overrides": [] + }, + "gridPos": { + "h": 6, + "w": 12, + "x": 0, + "y": 0 + }, + "id": 30, + "links": [], + "options": { + "legend": { + "calcs": [ + "lastNotNull" + ], + "displayMode": "table", + "placement": "right" + }, + "tooltip": { + "mode": "multi" + } + }, + "pluginVersion": "8.1.0", + "targets": [ + { + "exemplar": true, + "expr": "probe_http_status_code{target=~\"$target\"}", + "format": "time_series", + "interval": "$interval", + "intervalFactor": 1, + "legendFormat": "{{target}}", + "refId": "B" + } + ], + "timeFrom": null, + "timeShift": null, + "title": "HTTP Status Code", + "type": "timeseries" + }, + { + "cacheTimeout": null, + "datasource": "${DS_PROMETHEUS}", + "fieldConfig": { + "defaults": { + "color": { + "mode": "thresholds" + }, + "mappings": [ + { + "options": { + "0": { + "text": "DOWN" + }, + "1": { + "text": "UP" + } + }, + "type": "value" + }, + { + "options": { + "match": "null", + "result": { + "text": "N/A" + } + }, + "type": "special" + } + ], + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "#d44a3a", + "value": null + }, + { + "color": "rgba(237, 129, 40, 0.89)", + "value": 1 + }, + { + "color": "#299c46", + "value": 1 + } + ] + }, + "unit": "none" + }, + "overrides": [] + }, + "gridPos": { + "h": 6, + "w": 4, + "x": 12, + "y": 0 + }, + "id": 2, + "interval": null, + "links": [], + "maxDataPoints": 100, + "options": { + "colorMode": "background", + "graphMode": "none", + "justifyMode": "auto", + "orientation": "horizontal", + "reduceOptions": { + "calcs": [ + "lastNotNull" + ], + "fields": "", + "values": false + }, + "text": { + "titleSize": 12, + "valueSize": 12 + }, + "textMode": "auto" + }, + "pluginVersion": "8.1.0", + "repeat": null, + "repeatDirection": "v", + "targets": [ + { + "exemplar": true, + "expr": "probe_success{target=~\"$target\"}", + "format": "time_series", + "interval": "$interval", + "intervalFactor": 1, + "legendFormat": "{{target}}", + "refId": "A" + } + ], + "title": "Status", + "type": "stat" + }, + { + "cacheTimeout": null, + "datasource": "${DS_PROMETHEUS}", + "fieldConfig": { + "defaults": { + "color": { + "mode": "thresholds" + }, + "decimals": 0, + "mappings": [ + { + "options": { + "0": { + "text": "N/A" + }, + "1": { + "text": "YES" + } + }, + "type": "value" + }, + { + "options": { + "match": "null", + "result": { + "text": "N/A" + } + }, + "type": "special" + } + ], + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "#299c46", + "value": null + }, + { + "color": "rgba(237, 129, 40, 0.89)", + "value": 201 + }, + { + "color": "#d44a3a", + "value": 399 + } + ] + }, + "unit": "none" + }, + "overrides": [] + }, + "gridPos": { + "h": 6, + "w": 4, + "x": 16, + "y": 0 + }, + "id": 20, + "interval": null, + "links": [], + "maxDataPoints": 100, + "options": { + "colorMode": "value", + "graphMode": "none", + "justifyMode": "auto", + "orientation": "horizontal", + "reduceOptions": { + "calcs": [ + "lastNotNull" + ], + "fields": "", + "values": false + }, + "text": { + "titleSize": 12, + "valueSize": 12 + }, + "textMode": "auto" + }, + "pluginVersion": "8.1.0", + "repeat": null, + "repeatDirection": "h", + "targets": [ + { + "exemplar": true, + "expr": "probe_http_status_code{target=~\"$target\"}", + "format": "time_series", + "interval": "$interval", + "intervalFactor": 1, + "legendFormat": "{{target}}", + "refId": "A" + } + ], + "title": "HTTP Status Code", + "type": "stat" + }, + { + "cacheTimeout": null, + "datasource": "${DS_PROMETHEUS}", + "fieldConfig": { + "defaults": { + "color": { + "mode": "thresholds" + }, + "mappings": [ + { + "options": { + "match": "null", + "result": { + "text": "N/A" + } + }, + "type": "special" + } + ], + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green", + "value": null + }, + { + "color": "red", + "value": 80 + } + ] + }, + "unit": "none" + }, + "overrides": [] + }, + "gridPos": { + "h": 6, + "w": 4, + "x": 20, + "y": 0 + }, + "id": 27, + "interval": null, + "links": [], + "maxDataPoints": 100, + "options": { + "colorMode": "none", + "graphMode": "none", + "justifyMode": "auto", + "orientation": "horizontal", + "reduceOptions": { + "calcs": [ + "lastNotNull" + ], + "fields": "", + "values": false + }, + "text": { + "titleSize": 12, + "valueSize": 12 + }, + "textMode": "auto" + }, + "pluginVersion": "8.1.0", + "targets": [ + { + "exemplar": true, + "expr": "probe_http_version{target=~\"$target\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "{{target}}", + "refId": "A" + } + ], + "title": "HTTP Version", + "transformations": [], + "type": "stat" + }, + { + "datasource": "${DS_PROMETHEUS}", + "fieldConfig": { + "defaults": { + "color": { + "mode": "palette-classic" + }, + "custom": { + "axisLabel": "", + "axisPlacement": "auto", + "barAlignment": 0, + "drawStyle": "line", + "fillOpacity": 10, + "gradientMode": "none", + "hideFrom": { + "legend": false, + "tooltip": false, + "viz": false + }, + "lineInterpolation": "linear", + "lineWidth": 1, + "pointSize": 5, + "scaleDistribution": { + "type": "linear" + }, + "showPoints": "never", + "spanNulls": true, + "stacking": { + "group": "A", + "mode": "none" + }, + "thresholdsStyle": { + "mode": "off" + } + }, + "mappings": [], + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "dark-green", + "value": null + }, + { + "color": "orange", + "value": 0.2 + }, + { + "color": "red", + "value": 0.5 + } + ] + }, + "unit": "s" + }, + "overrides": [] + }, + "gridPos": { + "h": 6, + "w": 12, + "x": 0, + "y": 6 + }, + "id": 17, + "links": [], + "options": { + "legend": { + "calcs": [ + "mean" + ], + "displayMode": "table", + "placement": "right" + }, + "tooltip": { + "mode": "multi" + } + }, + "pluginVersion": "8.1.0", + "repeat": null, + "targets": [ + { + "exemplar": true, + "expr": "probe_duration_seconds{target=~\"$target\"}", + "format": "time_series", + "interval": "$interval", + "intervalFactor": 1, + "legendFormat": "{{target}}", + "refId": "A" + } + ], + "timeFrom": null, + "timeShift": null, + "title": "Probe Duration", + "type": "timeseries" + }, + { + "cacheTimeout": null, + "datasource": "${DS_PROMETHEUS}", + "fieldConfig": { + "defaults": { + "color": { + "mode": "thresholds" + }, + "mappings": [ + { + "options": { + "match": "null", + "result": { + "text": "N/A" + } + }, + "type": "special" + } + ], + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "dark-green", + "value": null + }, + { + "color": "orange", + "value": 0.2 + }, + { + "color": "red", + "value": 0.5 + } + ] + }, + "unit": "s" + }, + "overrides": [] + }, + "gridPos": { + "h": 6, + "w": 4, + "x": 12, + "y": 6 + }, + "id": 23, + "interval": null, + "links": [], + "maxDataPoints": 100, + "options": { + "colorMode": "value", + "graphMode": "none", + "justifyMode": "auto", + "orientation": "horizontal", + "reduceOptions": { + "calcs": [ + "lastNotNull" + ], + "fields": "", + "values": false + }, + "text": { + "titleSize": 12, + "valueSize": 12 + }, + "textMode": "auto" + }, + "pluginVersion": "8.1.0", + "repeat": null, + "targets": [ + { + "exemplar": true, + "expr": "avg(probe_duration_seconds{target=~\"$target\"}) by (target)", + "format": "time_series", + "interval": "$interval", + "intervalFactor": 1, + "legendFormat": "{{target}}", + "refId": "A" + } + ], + "title": "Average Probe Duration", + "type": "stat" + }, + { + "cacheTimeout": null, + "datasource": "${DS_PROMETHEUS}", + "fieldConfig": { + "defaults": { + "color": { + "mode": "thresholds" + }, + "decimals": 2, + "mappings": [ + { + "options": { + "match": "null", + "result": { + "text": "N/A" + } + }, + "type": "special" + } + ], + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "dark-green", + "value": null + }, + { + "color": "orange", + "value": 0.025 + }, + { + "color": "red", + "value": 0.05 + } + ] + }, + "unit": "s" + }, + "overrides": [] + }, + "gridPos": { + "h": 6, + "w": 4, + "x": 16, + "y": 6 + }, + "id": 24, + "interval": null, + "links": [], + "maxDataPoints": 100, + "options": { + "colorMode": "value", + "graphMode": "none", + "justifyMode": "auto", + "orientation": "horizontal", + "reduceOptions": { + "calcs": [ + "lastNotNull" + ], + "fields": "", + "values": false + }, + "text": { + "titleSize": 12, + "valueSize": 12 + }, + "textMode": "auto" + }, + "pluginVersion": "8.1.0", + "repeat": null, + "repeatDirection": "h", + "targets": [ + { + "exemplar": true, + "expr": "avg(probe_dns_lookup_time_seconds{target=~\"$target\"}) by (target)", + "format": "time_series", + "interval": "$interval", + "intervalFactor": 1, + "legendFormat": "{{target}}", + "refId": "A" + } + ], + "title": "Average DNS Lookup", + "type": "stat" + }, + { + "cacheTimeout": null, + "datasource": "${DS_PROMETHEUS}", + "fieldConfig": { + "defaults": { + "color": { + "mode": "thresholds" + }, + "mappings": [ + { + "options": { + "match": "null", + "result": { + "text": "N/A" + } + }, + "type": "special" + } + ], + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green", + "value": null + }, + { + "color": "red", + "value": 80 + } + ] + }, + "unit": "none" + }, + "overrides": [] + }, + "gridPos": { + "h": 6, + "w": 4, + "x": 20, + "y": 6 + }, + "id": 28, + "interval": null, + "links": [], + "maxDataPoints": 100, + "options": { + "colorMode": "none", + "graphMode": "none", + "justifyMode": "auto", + "orientation": "horizontal", + "reduceOptions": { + "calcs": [ + "lastNotNull" + ], + "fields": "", + "values": false + }, + "text": { + "titleSize": 12, + "valueSize": 12 + }, + "textMode": "auto" + }, + "pluginVersion": "8.1.0", + "targets": [ + { + "exemplar": true, + "expr": "probe_ip_protocol{target=~\"$target\"}", + "format": "time_series", + "interval": "", + "intervalFactor": 1, + "legendFormat": "{{target}}", + "refId": "A" + } + ], + "title": "IP protocol", + "type": "stat" + }, + { + "datasource": "${DS_PROMETHEUS}", + "fieldConfig": { + "defaults": { + "color": { + "mode": "palette-classic" + }, + "custom": { + "axisLabel": "", + "axisPlacement": "auto", + "barAlignment": 0, + "drawStyle": "line", + "fillOpacity": 10, + "gradientMode": "none", + "hideFrom": { + "legend": false, + "tooltip": false, + "viz": false + }, + "lineInterpolation": "smooth", + "lineWidth": 1, + "pointSize": 5, + "scaleDistribution": { + "type": "linear" + }, + "showPoints": "never", + "spanNulls": true, + "stacking": { + "group": "A", + "mode": "none" + }, + "thresholdsStyle": { + "mode": "off" + } + }, + "mappings": [], + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green", + "value": null + }, + { + "color": "red", + "value": 80 + } + ] + }, + "unit": "s" + }, + "overrides": [] + }, + "gridPos": { + "h": 6, + "w": 12, + "x": 0, + "y": 12 + }, + "id": 25, + "links": [], + "options": { + "legend": { + "calcs": [ + "mean" + ], + "displayMode": "table", + "placement": "right" + }, + "tooltip": { + "mode": "single" + } + }, + "pluginVersion": "8.1.0", + "targets": [ + { + "exemplar": true, + "expr": "probe_http_duration_seconds{target=~\"$target\", phase=~\"$phase\"}", + "format": "time_series", + "interval": "$interval", + "intervalFactor": 1, + "legendFormat": "{{target}} - {{ phase }}", + "refId": "B" + } + ], + "timeFrom": null, + "timeShift": null, + "title": "HTTP Duration (phase = $phase)", + "type": "timeseries" + }, + { + "cacheTimeout": null, + "datasource": "${DS_PROMETHEUS}", + "fieldConfig": { + "defaults": { + "color": { + "mode": "thresholds" + }, + "decimals": 2, + "mappings": [ + { + "options": { + "0": { + "text": "NO" + }, + "1": { + "text": "YES" + } + }, + "type": "value" + }, + { + "options": { + "match": "null", + "result": { + "text": "N/A" + } + }, + "type": "special" + } + ], + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "#d44a3a", + "value": null + }, + { + "color": "rgba(237, 129, 40, 0.89)", + "value": 0 + }, + { + "color": "#299c46", + "value": 1209600 + } + ] + }, + "unit": "dtdhms" + }, + "overrides": [] + }, + "gridPos": { + "h": 6, + "w": 4, + "x": 12, + "y": 12 + }, + "id": 19, + "interval": null, + "links": [], + "maxDataPoints": 100, + "options": { + "colorMode": "value", + "graphMode": "none", + "justifyMode": "auto", + "orientation": "horizontal", + "reduceOptions": { + "calcs": [ + "lastNotNull" + ], + "fields": "", + "values": false + }, + "text": { + "titleSize": 12, + "valueSize": 12 + }, + "textMode": "auto" + }, + "pluginVersion": "8.1.0", + "repeat": null, + "repeatDirection": "h", + "targets": [ + { + "exemplar": true, + "expr": "probe_ssl_earliest_cert_expiry{target=~\"$target\"} - time()", + "format": "time_series", + "interval": "$interval", + "intervalFactor": 1, + "legendFormat": "{{target}}", + "refId": "A" + } + ], + "timeFrom": null, + "title": "SSL Expiry", + "type": "stat" + }, + { + "cacheTimeout": null, + "datasource": "${DS_PROMETHEUS}", + "fieldConfig": { + "defaults": { + "color": { + "mode": "thresholds" + }, + "mappings": [ + { + "options": { + "0": { + "text": "NO" + }, + "1": { + "text": "YES" + } + }, + "type": "value" + }, + { + "options": { + "match": "null", + "result": { + "text": "N/A" + } + }, + "type": "special" + } + ], + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "#d44a3a", + "value": null + }, + { + "color": "rgba(237, 129, 40, 0.89)", + "value": 0 + }, + { + "color": "#299c46", + "value": 1 + } + ] + }, + "unit": "none" + }, + "overrides": [] + }, + "gridPos": { + "h": 6, + "w": 4, + "x": 16, + "y": 12 + }, + "id": 18, + "interval": null, + "links": [], + "maxDataPoints": 100, + "options": { + "colorMode": "value", + "graphMode": "none", + "justifyMode": "auto", + "orientation": "horizontal", + "reduceOptions": { + "calcs": [ + "lastNotNull" + ], + "fields": "", + "values": false + }, + "text": { + "titleSize": 12, + "valueSize": 12 + }, + "textMode": "auto" + }, + "pluginVersion": "8.1.0", + "repeat": null, + "repeatDirection": "v", + "targets": [ + { + "exemplar": true, + "expr": "probe_http_ssl{target=~\"$target\"}", + "format": "time_series", + "interval": "$interval", + "intervalFactor": 1, + "legendFormat": "{{target}}", + "refId": "A" + } + ], + "title": "SSL", + "type": "stat" + }, + { + "cacheTimeout": null, + "datasource": "${DS_PROMETHEUS}", + "fieldConfig": { + "defaults": { + "color": { + "mode": "thresholds" + }, + "custom": { + "align": "auto", + "displayMode": "auto", + "filterable": false + }, + "mappings": [ + { + "options": { + "match": "null", + "result": { + "text": "N/A" + } + }, + "type": "special" + } + ], + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green", + "value": null + }, + { + "color": "red", + "value": 80 + } + ] + }, + "unit": "none" + }, + "overrides": [] + }, + "gridPos": { + "h": 6, + "w": 4, + "x": 20, + "y": 12 + }, + "id": 29, + "interval": null, + "links": [], + "maxDataPoints": 100, + "options": { + "showHeader": false, + "sortBy": [ + { + "desc": false, + "displayName": "job" + } + ] + }, + "pluginVersion": "8.1.0", + "targets": [ + { + "exemplar": true, + "expr": "probe_tls_version_info{target=~\"$target\"}", + "format": "table", + "instant": true, + "interval": "", + "intervalFactor": 1, + "legendFormat": "{{version}}", + "refId": "A" + } + ], + "title": "TLS version", + "transformations": [ + { + "id": "organize", + "options": { + "excludeByName": { + "Time": true, + "Value": true, + "Value #A": true, + "__name__": true, + "instance": true, + "job": true, + "version": false + }, + "indexByName": {}, + "renameByName": { + "__name__": "" + } + } + } + ], + "type": "table" + } + ], + "refresh": false, + "schemaVersion": 30, + "style": "dark", + "tags": [ + "blackbox", + "prometheus" + ], + "templating": { + "list": [ + { + "auto": true, + "auto_count": 10, + "auto_min": "10s", + "current": { + "selected": false, + "text": "30s", + "value": "30s" + }, + "description": null, + "error": null, + "hide": 0, + "label": "Interval", + "name": "interval", + "options": [ + { + "selected": false, + "text": "auto", + "value": "$__auto_interval_interval" + }, + { + "selected": false, + "text": "5s", + "value": "5s" + }, + { + "selected": false, + "text": "10s", + "value": "10s" + }, + { + "selected": true, + "text": "30s", + "value": "30s" + }, + { + "selected": false, + "text": "1m", + "value": "1m" + }, + { + "selected": false, + "text": "10m", + "value": "10m" + }, + { + "selected": false, + "text": "30m", + "value": "30m" + }, + { + "selected": false, + "text": "1h", + "value": "1h" + }, + { + "selected": false, + "text": "6h", + "value": "6h" + }, + { + "selected": false, + "text": "12h", + "value": "12h" + }, + { + "selected": false, + "text": "1d", + "value": "1d" + }, + { + "selected": false, + "text": "7d", + "value": "7d" + }, + { + "selected": false, + "text": "14d", + "value": "14d" + }, + { + "selected": false, + "text": "30d", + "value": "30d" + } + ], + "query": "5s,10s,30s,1m,10m,30m,1h,6h,12h,1d,7d,14d,30d", + "queryValue": "", + "refresh": 2, + "skipUrlSync": false, + "type": "interval" + }, + { + "allValue": null, + "current": {}, + "datasource": "${DS_PROMETHEUS}", + "definition": "label_values(probe_success, target)", + "description": null, + "error": null, + "hide": 0, + "includeAll": true, + "label": null, + "multi": true, + "name": "target", + "options": [], + "query": { + "query": "label_values(probe_success, target)", + "refId": "StandardVariableQuery" + }, + "refresh": 1, + "regex": "", + "skipUrlSync": false, + "sort": 0, + "tagValuesQuery": "", + "tagsQuery": "", + "type": "query", + "useTags": false + }, + { + "allValue": null, + "current": {}, + "datasource": "${DS_PROMETHEUS}", + "definition": "label_values(probe_http_duration_seconds, phase)", + "description": null, + "error": null, + "hide": 0, + "includeAll": true, + "label": "HTTP phase", + "multi": true, + "name": "phase", + "options": [], + "query": { + "query": "label_values(probe_http_duration_seconds, phase)", + "refId": "StandardVariableQuery" + }, + "refresh": 1, + "regex": "", + "skipUrlSync": false, + "sort": 1, + "tagValuesQuery": "", + "tagsQuery": "", + "type": "query", + "useTags": false + }, + { + "current": { + "selected": false, + "text": "Prometheus", + "value": "Prometheus" + }, + "description": null, + "error": null, + "hide": 2, + "includeAll": false, + "label": null, + "multi": false, + "name": "DS_PROMETHEUS", + "options": [], + "query": "prometheus", + "refresh": 1, + "regex": "", + "skipUrlSync": false, + "type": "datasource" + } + ] + }, + "time": { + "from": "now-30m", + "to": "now" + }, + "timepicker": { + "refresh_intervals": [ + "5s", + "10s", + "30s", + "1m", + "5m", + "15m", + "30m", + "1h", + "2h", + "1d" + ], + "time_options": [ + "5m", + "15m", + "30m", + "1h", + "6h", + "12h", + "24h", + "2d", + "7d", + "30d" + ] + }, + "timezone": "", + "title": "Prometheus Blackbox Exporter", + "uid": "pS6ZuGV7z", + "version": 1, + "gnetId": 14928 +} \ No newline at end of file diff --git a/conf/grafana/provisioning/dashboards/dashboard.yml b/conf/grafana/provisioning/dashboards/dashboard.yml new file mode 100644 index 0000000..f0fb649 --- /dev/null +++ b/conf/grafana/provisioning/dashboards/dashboard.yml @@ -0,0 +1,11 @@ +apiVersion: 1 + +providers: +- name: 'General' + orgId: 1 + folder: '' + type: file + disableDeletion: false + editable: true + options: + path: /etc/grafana/provisioning/dashboards \ No newline at end of file diff --git a/conf/grafana/provisioning/dashboards/docker.json b/conf/grafana/provisioning/dashboards/docker.json new file mode 100644 index 0000000..86528f7 --- /dev/null +++ b/conf/grafana/provisioning/dashboards/docker.json @@ -0,0 +1,1103 @@ +{ + "__inputs": [ + { + "name": "DS_PROMETHEUS", + "label": "Prometheus", + "description": "", + "type": "datasource", + "pluginId": "prometheus", + "pluginName": "Prometheus" + } + ], + "__requires": [ + { + "type": "grafana", + "id": "grafana", + "name": "Grafana", + "version": "5.4.3" + }, + { + "type": "panel", + "id": "graph", + "name": "Graph", + "version": "5.0.0" + }, + { + "type": "datasource", + "id": "prometheus", + "name": "Prometheus", + "version": "5.0.0" + }, + { + "type": "panel", + "id": "singlestat", + "name": "Singlestat", + "version": "5.0.0" + }, + { + "type": "panel", + "id": "table", + "name": "Table", + "version": "5.0.0" + } + ], + "annotations": { + "list": [ + { + "builtIn": 1, + "datasource": "-- Grafana --", + "enable": true, + "hide": true, + "iconColor": "rgba(0, 211, 255, 1)", + "name": "Annotations & Alerts", + "type": "dashboard" + } + ] + }, + "description": "A simple overview of the most important Docker host and container metrics. (cAdvisor/Prometheus)", + "editable": true, + "gnetId": 10619, + "graphTooltip": 1, + "id": null, + "iteration": 1564137194972, + "links": [], + "panels": [ + { + "cacheTimeout": null, + "colorBackground": false, + "colorValue": false, + "colors": [ + "#299c46", + "rgba(237, 129, 40, 0.89)", + "#d44a3a" + ], + "datasource": "${DS_PROMETHEUS}", + "format": "none", + "gauge": { + "maxValue": 100, + "minValue": 0, + "show": false, + "thresholdLabels": false, + "thresholdMarkers": true + }, + "gridPos": { + "h": 6, + "w": 4, + "x": 0, + "y": 0 + }, + "id": 22, + "interval": null, + "links": [], + "mappingType": 1, + "mappingTypes": [ + { + "name": "value to text", + "value": 1 + }, + { + "name": "range to text", + "value": 2 + } + ], + "maxDataPoints": 100, + "nullPointMode": "connected", + "nullText": null, + "postfix": "", + "postfixFontSize": "50%", + "prefix": "", + "prefixFontSize": "50%", + "rangeMaps": [ + { + "from": "null", + "text": "N/A", + "to": "null" + } + ], + "sparkline": { + "fillColor": "rgba(31, 118, 189, 0.18)", + "full": false, + "lineColor": "rgb(31, 120, 193)", + "show": false + }, + "tableColumn": "", + "targets": [ + { + "bucketAggs": [ + { + "field": "@timestamp", + "id": "2", + "settings": { + "interval": "auto", + "min_doc_count": 0, + "trimEdges": 0 + }, + "type": "date_histogram" + } + ], + "expr": "count(container_last_seen{instance=~\"$node\",job=~\"$job\",image!=\"\"})", + "format": "time_series", + "intervalFactor": 1, + "metrics": [ + { + "field": "select field", + "id": "1", + "type": "count" + } + ], + "query": "count(container_last_seen{instance=~\"$node\",job=~\"$job\",image!=\"\"})", + "refId": "A", + "timeField": "@timestamp" + } + ], + "thresholds": "", + "title": "Running containers", + "transparent": true, + "type": "singlestat", + "valueFontSize": "80%", + "valueMaps": [ + { + "op": "=", + "text": "N/A", + "value": "null" + } + ], + "valueName": "avg" + }, + { + "aliasColors": { + "{id=\"/\",instance=\"cadvisor:8080\",job=\"prometheus\"}": "#BA43A9" + }, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "editable": true, + "error": false, + "fill": 3, + "grid": {}, + "gridPos": { + "h": 6, + "w": 8, + "x": 4, + "y": 0 + }, + "id": 5, + "isNew": true, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": false, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 2, + "links": [], + "nullPointMode": "null as zero", + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": true, + "steppedLine": false, + "targets": [ + { + "expr": "sum(rate(container_cpu_system_seconds_total[1m]))", + "hide": true, + "intervalFactor": 2, + "legendFormat": "a", + "refId": "B", + "step": 120 + }, + { + "expr": "sum(rate(container_cpu_system_seconds_total{name=~\".+\"}[1m]))", + "hide": true, + "interval": "", + "intervalFactor": 2, + "legendFormat": "nur container", + "refId": "F", + "step": 10 + }, + { + "expr": "sum(rate(container_cpu_system_seconds_total{id=\"/\"}[1m]))", + "hide": true, + "interval": "", + "intervalFactor": 2, + "legendFormat": "nur docker host", + "metric": "", + "refId": "A", + "step": 20 + }, + { + "expr": "sum(rate(process_cpu_seconds_total[$interval])) * 100", + "hide": false, + "interval": "", + "intervalFactor": 2, + "legendFormat": "host", + "metric": "", + "refId": "C", + "step": 4 + }, + { + "expr": "sum(rate(container_cpu_system_seconds_total{name=~\".+\"}[1m])) + sum(rate(container_cpu_system_seconds_total{id=\"/\"}[1m])) + sum(rate(process_cpu_seconds_total[1m]))", + "hide": true, + "intervalFactor": 2, + "legendFormat": "", + "refId": "D", + "step": 120 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "CPU Usage on Node", + "tooltip": { + "msResolution": true, + "shared": true, + "sort": 0, + "value_type": "cumulative" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": false, + "values": [] + }, + "yaxes": [ + { + "format": "percent", + "label": "", + "logBase": 1, + "max": 120, + "min": null, + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": { + "SENT": "#BF1B00" + }, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "editable": true, + "error": false, + "fill": 5, + "grid": {}, + "gridPos": { + "h": 4, + "w": 12, + "x": 0, + "y": 6 + }, + "id": 19, + "isNew": true, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": false, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null as zero", + "percentage": false, + "pointradius": 1, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "sum(rate(container_network_receive_bytes_total{id=\"/\"}[$interval])) by (id)", + "intervalFactor": 2, + "legendFormat": "RECEIVED", + "refId": "A", + "step": 4 + }, + { + "expr": "- sum(rate(container_network_transmit_bytes_total{id=\"/\"}[$interval])) by (id)", + "hide": false, + "intervalFactor": 2, + "legendFormat": "SENT", + "refId": "B", + "step": 4 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Network Traffic on Node", + "tooltip": { + "msResolution": true, + "shared": true, + "sort": 0, + "value_type": "cumulative" + }, + "transparent": false, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": false, + "values": [] + }, + "yaxes": [ + { + "format": "bytes", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "editable": true, + "error": false, + "fill": 5, + "grid": {}, + "gridPos": { + "h": 8, + "w": 12, + "x": 0, + "y": 10 + }, + "id": 1, + "isNew": true, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "links": [], + "nullPointMode": "null as zero", + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": true, + "steppedLine": false, + "targets": [ + { + "expr": "sum(rate(container_cpu_usage_seconds_total{name=~\".+\"}[$interval])) by (name) * 100", + "format": "time_series", + "hide": false, + "interval": "", + "intervalFactor": 2, + "legendFormat": "{{name}}", + "metric": "container_cp", + "refId": "F", + "step": 2 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "CPU Usage per Container (Stacked)", + "tooltip": { + "msResolution": true, + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "percent", + "label": "", + "logBase": 1, + "max": null, + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "editable": true, + "error": false, + "fill": 3, + "grid": {}, + "gridPos": { + "h": 6, + "w": 12, + "x": 12, + "y": 12 + }, + "id": 10, + "isNew": true, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 2, + "links": [], + "nullPointMode": "null as zero", + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": true, + "steppedLine": false, + "targets": [ + { + "expr": "sum(container_memory_rss{name=~\".+\"}) by (name)", + "hide": false, + "intervalFactor": 2, + "legendFormat": "{{name}}", + "refId": "A", + "step": 2 + }, + { + "expr": "container_memory_usage_bytes{name=~\".+\"}", + "hide": true, + "intervalFactor": 2, + "legendFormat": "{{name}}", + "refId": "B", + "step": 240 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Memory Usage per Container (Stacked)", + "tooltip": { + "msResolution": true, + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "bytes", + "label": "", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": true + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "editable": true, + "error": false, + "fill": 1, + "grid": {}, + "gridPos": { + "h": 6, + "w": 12, + "x": 12, + "y": 18 + }, + "id": 9, + "isNew": true, + "legend": { + "avg": false, + "current": false, + "hideEmpty": false, + "hideZero": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 2, + "links": [], + "nullPointMode": "null as zero", + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "sum(rate(container_network_transmit_bytes_total{name=~\".+\"}[$interval])) by (name)", + "intervalFactor": 2, + "legendFormat": "{{name}}", + "refId": "A", + "step": 2 + }, + { + "expr": "rate(container_network_transmit_bytes_total{id=\"/\"}[$interval])", + "hide": true, + "intervalFactor": 2, + "legendFormat": "", + "refId": "B", + "step": 10 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Sent Network Traffic per Container", + "tooltip": { + "msResolution": true, + "shared": true, + "sort": 0, + "value_type": "cumulative" + }, + "transparent": false, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "Bps", + "label": "", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "format": "short", + "label": "", + "logBase": 10, + "max": 8, + "min": 0, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "editable": true, + "error": false, + "fill": 3, + "grid": {}, + "gridPos": { + "h": 6, + "w": 12, + "x": 12, + "y": 18 + }, + "id": 11, + "isNew": true, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 2, + "links": [], + "nullPointMode": "null as zero", + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": true, + "steppedLine": false, + "targets": [ + { + "expr": "container_memory_rss{name=~\".+\"}", + "hide": true, + "intervalFactor": 2, + "legendFormat": "{{name}}", + "refId": "A", + "step": 20 + }, + { + "expr": "container_memory_usage_bytes{name=~\".+\"}", + "hide": true, + "intervalFactor": 2, + "legendFormat": "{{name}}", + "refId": "B", + "step": 20 + }, + { + "expr": "sum(container_memory_cache{name=~\".+\"}) by (name)", + "hide": false, + "intervalFactor": 2, + "legendFormat": "{{name}}", + "refId": "C", + "step": 2 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Cached Memory per Container (Stacked)", + "tooltip": { + "msResolution": true, + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "bytes", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "editable": true, + "error": false, + "fill": 1, + "grid": {}, + "gridPos": { + "h": 6, + "w": 12, + "x": 12, + "y": 24 + }, + "id": 8, + "isNew": true, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 2, + "links": [], + "nullPointMode": "null as zero", + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "sum(rate(container_network_receive_bytes_total{name=~\".+\"}[$interval])) by (name)", + "intervalFactor": 2, + "legendFormat": "{{name}}", + "refId": "A", + "step": 2 + }, + { + "expr": "- rate(container_network_transmit_bytes_total{name=~\".+\"}[$interval])", + "hide": true, + "intervalFactor": 2, + "legendFormat": "{{name}}", + "refId": "B", + "step": 10 + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Received Network Traffic per Container", + "tooltip": { + "msResolution": true, + "shared": true, + "sort": 0, + "value_type": "cumulative" + }, + "transparent": false, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "Bps", + "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": "10s", + "schemaVersion": 16, + "style": "dark", + "tags": [], + "templating": { + "list": [ + { + "auto": true, + "auto_count": 50, + "auto_min": "50s", + "current": { + "text": "auto", + "value": "$__auto_interval_interval" + }, + "datasource": null, + "hide": 0, + "includeAll": false, + "label": "Interval", + "multi": false, + "name": "interval", + "options": [ + { + "selected": true, + "text": "auto", + "value": "$__auto_interval_interval" + }, + { + "selected": false, + "text": "30s", + "value": "30s" + }, + { + "selected": false, + "text": "1m", + "value": "1m" + }, + { + "selected": false, + "text": "2m", + "value": "2m" + }, + { + "selected": false, + "text": "3m", + "value": "3m" + }, + { + "selected": false, + "text": "5m", + "value": "5m" + }, + { + "selected": false, + "text": "7m", + "value": "7m" + }, + { + "selected": false, + "text": "10m", + "value": "10m" + }, + { + "selected": false, + "text": "30m", + "value": "30m" + }, + { + "selected": false, + "text": "1h", + "value": "1h" + }, + { + "selected": false, + "text": "6h", + "value": "6h" + }, + { + "selected": false, + "text": "12h", + "value": "12h" + }, + { + "selected": false, + "text": "1d", + "value": "1d" + }, + { + "selected": false, + "text": "7d", + "value": "7d" + }, + { + "selected": false, + "text": "14d", + "value": "14d" + }, + { + "selected": false, + "text": "30d", + "value": "30d" + } + ], + "query": "30s,1m,2m,3m,5m,7m,10m,30m,1h,6h,12h,1d,7d,14d,30d", + "refresh": 2, + "skipUrlSync": false, + "type": "interval" + }, + { + "allValue": null, + "current": {}, + "datasource": "${DS_PROMETHEUS}", + "definition": "label_values(container_cpu_user_seconds_total, job)", + "hide": 0, + "includeAll": false, + "label": "Job", + "multi": false, + "name": "job", + "options": [], + "query": "label_values(container_cpu_user_seconds_total, job)", + "refresh": 1, + "regex": "", + "skipUrlSync": false, + "sort": 1, + "tagValuesQuery": "", + "tags": [], + "tagsQuery": "", + "type": "query", + "useTags": false + }, + { + "allValue": null, + "current": {}, + "datasource": "${DS_PROMETHEUS}", + "definition": "label_values(container_cpu_user_seconds_total{job=~\"$job\"}, instance)", + "hide": 0, + "includeAll": false, + "label": "Host:", + "multi": true, + "name": "node", + "options": [], + "query": "label_values(container_cpu_user_seconds_total{job=~\"$job\"}, instance)", + "refresh": 1, + "skipUrlSync": false, + "sort": 1, + "tagValuesQuery": "", + "tags": [], + "tagsQuery": "", + "type": "query", + "useTags": false + }, + { + "current": { + "selected": false, + "text": "Prometheus", + "value": "Prometheus" + }, + "description": null, + "error": null, + "hide": 2, + "includeAll": false, + "label": null, + "multi": false, + "name": "DS_PROMETHEUS", + "options": [], + "query": "prometheus", + "refresh": 1, + "regex": "", + "skipUrlSync": false, + "type": "datasource" + } + ] + }, + "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": "browser", + "title": "Docker Container", + "uid": "4dMaCsRZz", + "version": 27 +} \ No newline at end of file diff --git a/conf/grafana/provisioning/dashboards/smokeping.json.bak b/conf/grafana/provisioning/dashboards/smokeping.json.bak new file mode 100644 index 0000000..3245328 --- /dev/null +++ b/conf/grafana/provisioning/dashboards/smokeping.json.bak @@ -0,0 +1,230 @@ +{ + "__inputs": [ + { + "name": "DS_PROMETHEUS", + "label": "Prometheus", + "description": "", + "type": "datasource", + "pluginId": "prometheus", + "pluginName": "Prometheus" + } + ], + "__requires": [ + { + "type": "grafana", + "id": "grafana", + "name": "Grafana", + "version": "7.3.7" + }, + { + "type": "panel", + "id": "heatmap", + "name": "Heatmap", + "version": "" + }, + { + "type": "datasource", + "id": "prometheus", + "name": "Prometheus", + "version": "1.0.0" + } + ], + "annotations": { + "list": [ + { + "builtIn": 1, + "datasource": "-- Grafana --", + "enable": true, + "hide": true, + "iconColor": "rgba(0, 211, 255, 1)", + "name": "Annotations & Alerts", + "type": "dashboard" + } + ] + }, + "editable": true, + "fiscalYearStartMonth": 0, + "gnetId": null, + "graphTooltip": 0, + "id": 2, + "iteration": 1650541706979, + "links": [], + "liveNow": false, + "panels": [ + { + "cards": { + "cardPadding": null, + "cardRound": null + }, + "color": { + "cardColor": "#b4ff00", + "colorScale": "sqrt", + "colorScheme": "interpolateOranges", + "exponent": 0.5, + "mode": "opacity" + }, + "dataFormat": "tsbuckets", + "datasource": "${DS_PROMETHEUS}", + "gridPos": { + "h": 17, + "w": 24, + "x": 0, + "y": 0 + }, + "heatmap": {}, + "hideZeroBuckets": false, + "highlightCards": true, + "id": 2, + "legend": { + "show": false + }, + "links": [], + "reverseYBuckets": false, + "targets": [ + { + "expr": "sum(rate(smokeping_response_duration_seconds_bucket{instance=~\"$prober\",host=\"$target\"}[5m])) by (le)", + "format": "heatmap", + "intervalFactor": 1, + "legendFormat": "{{le}}", + "refId": "A" + } + ], + "title": "Smokeping", + "tooltip": { + "show": true, + "showHistogram": false + }, + "type": "heatmap", + "xAxis": { + "show": true + }, + "xBucketNumber": null, + "xBucketSize": null, + "yAxis": { + "decimals": 0, + "format": "s", + "logBase": 1, + "max": null, + "min": "0", + "show": true, + "splitFactor": null + }, + "yBucketBound": "auto", + "yBucketNumber": null, + "yBucketSize": null + } + ], + "refresh": "30s", + "schemaVersion": 32, + "style": "dark", + "tags": [], + "templating": { + "list": [ + { + "allValue": null, + "current": {}, + "datasource": "${DS_PROMETHEUS}", + "definition": "", + "description": null, + "error": null, + "hide": 0, + "includeAll": false, + "label": null, + "multi": true, + "name": "prober", + "options": [], + "query": { + "query": "label_values(smokeping_prober_build_info, instance)", + "refId": "Prometheus-prober-Variable-Query" + }, + "refresh": 1, + "regex": "", + "skipUrlSync": false, + "sort": 1, + "tagValuesQuery": "", + "tagsQuery": "", + "type": "query", + "useTags": false + }, + { + "allValue": null, + "current": {}, + "datasource": "${DS_PROMETHEUS}", + "definition": "label_values(smokeping_response_duration_seconds_bucket, host)", + "description": null, + "error": null, + "hide": 0, + "includeAll": false, + "label": null, + "multi": false, + "name": "target", + "options": [], + "query": { + "query": "label_values(smokeping_response_duration_seconds_bucket, host)", + "refId": "Prometheus-target-Variable-Query" + }, + "refresh": 1, + "regex": "", + "skipUrlSync": false, + "sort": 1, + "tagValuesQuery": "", + "tagsQuery": "", + "type": "query", + "useTags": false + }, + { + "current": { + "selected": false, + "text": "Prometheus", + "value": "Prometheus" + }, + "description": null, + "error": null, + "hide": 2, + "includeAll": false, + "label": null, + "multi": false, + "name": "DS_PROMETHEUS", + "options": [], + "query": "prometheus", + "refresh": 1, + "regex": "", + "skipUrlSync": false, + "type": "datasource" + } + ] + }, + "time": { + "from": "now-1h", + "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": "browser", + "title": "Smokeping", + "uid": "i5aRaLaik", + "version": 1 +} diff --git a/conf/grafana/provisioning/datasources/datasource.yml b/conf/grafana/provisioning/datasources/datasource.yml new file mode 100644 index 0000000..c02bb38 --- /dev/null +++ b/conf/grafana/provisioning/datasources/datasource.yml @@ -0,0 +1,50 @@ +# config file version +apiVersion: 1 + +# list of datasources that should be deleted from the database +deleteDatasources: + - name: Prometheus + orgId: 1 + +# list of datasources to insert/update depending +# whats available in the database +datasources: + # name of the datasource. Required +- name: Prometheus + # datasource type. Required + type: prometheus + # access mode. direct or proxy. Required + access: proxy + # org id. will default to orgId 1 if not specified + orgId: 1 + # url + url: http://prometheus:9090 + # database password, if used + password: + # database user, if used + user: + # database name, if used + database: + # enable/disable basic auth + basicAuth: false + # basic auth username, if used + basicAuthUser: + # basic auth password, if used + basicAuthPassword: + # enable/disable with credentials headers + withCredentials: + # mark as default datasource. Max one per org + isDefault: true + # fields that will be converted to json and stored in json_data + jsonData: + graphiteVersion: "1.1" + tlsAuth: false + tlsAuthWithCACert: false + # json object of data that will be encrypted. + secureJsonData: + tlsCACert: "..." + tlsClientCert: "..." + tlsClientKey: "..." + version: 1 + # allow users to edit datasources from the UI. + editable: true diff --git a/conf/prometheus/alerts/blackbox.yml b/conf/prometheus/alerts/blackbox.yml new file mode 100644 index 0000000..28aa8ea --- /dev/null +++ b/conf/prometheus/alerts/blackbox.yml @@ -0,0 +1,74 @@ +groups: + - name: blackbox + rules: + - alert: BlackboxProbeFailed + expr: probe_success == 0 + for: 0m + labels: + severity: critical + annotations: + summary: Blackbox probe failed (instance {{ $labels.instance }}) + description: "Probe failed\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: BlackboxSlowProbe + expr: avg_over_time(probe_duration_seconds[1m]) > 5 + for: 1m + labels: + severity: warning + annotations: + summary: Blackbox slow probe (instance {{ $labels.instance }}) + description: "Blackbox probe took more than 5s to complete\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: BlackboxProbeHttpFailure + expr: probe_http_status_code <= 199 OR probe_http_status_code >= 400 + for: 0m + labels: + severity: critical + annotations: + summary: Blackbox probe HTTP failure (instance {{ $labels.instance }}) + description: "HTTP status code is not 200-399\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: BlackboxSslCertificateWillExpireSoon + expr: probe_ssl_earliest_cert_expiry - time() < 86400 * 30 + for: 0m + labels: + severity: warning + annotations: + summary: Blackbox SSL certificate will expire soon (instance {{ $labels.instance }}) + description: "SSL certificate expires in 30 days\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: BlackboxSslCertificateWillExpireSoon + expr: probe_ssl_earliest_cert_expiry - time() < 86400 * 3 + for: 0m + labels: + severity: critical + annotations: + summary: Blackbox SSL certificate will expire soon (instance {{ $labels.instance }}) + description: "SSL certificate expires in 3 days\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: BlackboxSslCertificateExpired + expr: probe_ssl_earliest_cert_expiry - time() <= 0 + for: 0m + labels: + severity: critical + annotations: + summary: Blackbox SSL certificate expired (instance {{ $labels.instance }}) + description: "SSL certificate has expired already\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: BlackboxProbeSlowHttp + expr: avg_over_time(probe_http_duration_seconds[1m]) > 1 + for: 1m + labels: + severity: warning + annotations: + summary: Blackbox probe slow HTTP (instance {{ $labels.instance }}) + description: "HTTP request took more than 1s\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: BlackboxProbeSlowPing + expr: avg_over_time(probe_icmp_duration_seconds[1m]) > 1 + for: 1m + labels: + severity: warning + annotations: + summary: Blackbox probe slow ping (instance {{ $labels.instance }}) + description: "Blackbox ping took more than 1s\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" diff --git a/conf/prometheus/alerts/cadvisor.yml b/conf/prometheus/alerts/cadvisor.yml new file mode 100644 index 0000000..b3bc5c4 --- /dev/null +++ b/conf/prometheus/alerts/cadvisor.yml @@ -0,0 +1,68 @@ +groups: + - name: cadvisor + rules: + # This rule can be very noisy in dynamic infra with legitimate container start/stop/deployment. + - alert: ContainerKilled + expr: time() - container_last_seen > 60 + for: 0m + labels: + severity: warning + annotations: + summary: Container killed (instance {{ $labels.instance }}) + description: "A container has disappeared\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + # This rule can be very noisy in dynamic infra with legitimate container start/stop/deployment. + - alert: ContainerAbsent + expr: absent(container_last_seen) + for: 5m + labels: + severity: warning + annotations: + summary: Container absent (instance {{ $labels.instance }}) + description: "A container is absent for 5 min\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: ContainerCpuUsage + expr: (sum(rate(container_cpu_usage_seconds_total{name!=""}[3m])) BY (instance, name) * 100) > 80 + for: 2m + labels: + severity: warning + annotations: + summary: Container CPU usage (instance {{ $labels.instance }}) + description: "Container CPU usage is above 80%\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + # See https://medium.com/faun/how-much-is-too-much-the-linux-oomkiller-and-used-memory-d32186f29c9d + - alert: ContainerMemoryUsage + expr: (sum(container_memory_working_set_bytes{name!=""}) BY (instance, name) / sum(container_spec_memory_limit_bytes > 0) BY (instance, name) * 100) > 80 + for: 2m + labels: + severity: warning + annotations: + summary: Container Memory usage (instance {{ $labels.instance }}) + description: "Container Memory usage is above 80%\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: ContainerVolumeUsage + expr: (1 - (sum(container_fs_inodes_free{name!=""}) BY (instance) / sum(container_fs_inodes_total) BY (instance))) * 100 > 80 + for: 2m + labels: + severity: warning + annotations: + summary: Container Volume usage (instance {{ $labels.instance }}) + description: "Container Volume usage is above 80%\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: ContainerVolumeIoUsage + expr: (sum(container_fs_io_current{name!=""}) BY (instance, name) * 100) > 80 + for: 2m + labels: + severity: warning + annotations: + summary: Container Volume IO usage (instance {{ $labels.instance }}) + description: "Container Volume IO usage is above 80%\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: ContainerHighThrottleRate + expr: rate(container_cpu_cfs_throttled_seconds_total[3m]) > 1 + for: 2m + labels: + severity: warning + annotations: + summary: Container high throttle rate (instance {{ $labels.instance }}) + description: "Container is being throttled\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" diff --git a/conf/prometheus/alerts/node.yml b/conf/prometheus/alerts/node.yml new file mode 100644 index 0000000..1fe550a --- /dev/null +++ b/conf/prometheus/alerts/node.yml @@ -0,0 +1,308 @@ +groups: + - name: Node + rules: + - alert: HostOutOfMemory + expr: node_memory_MemAvailable_bytes / node_memory_MemTotal_bytes * 100 < 10 + for: 2m + labels: + severity: warning + annotations: + summary: Host out of memory (instance {{ $labels.instance }}) + description: "Node memory is filling up (< 10% left)\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: HostMemoryUnderMemoryPressure + expr: rate(node_vmstat_pgmajfault[1m]) > 1000 + for: 2m + labels: + severity: warning + annotations: + summary: Host memory under memory pressure (instance {{ $labels.instance }}) + description: "The node is under heavy memory pressure. High rate of major page faults\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: HostUnusualNetworkThroughputIn + expr: sum by (instance) (rate(node_network_receive_bytes_total[2m])) / 1024 / 1024 > 100 + for: 5m + labels: + severity: warning + annotations: + summary: Host unusual network throughput in (instance {{ $labels.instance }}) + description: "Host network interfaces are probably receiving too much data (> 100 MB/s)\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: HostUnusualNetworkThroughputOut + expr: sum by (instance) (rate(node_network_transmit_bytes_total[2m])) / 1024 / 1024 > 100 + for: 5m + labels: + severity: warning + annotations: + summary: Host unusual network throughput out (instance {{ $labels.instance }}) + description: "Host network interfaces are probably sending too much data (> 100 MB/s)\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: HostUnusualDiskReadRate + expr: sum by (instance) (rate(node_disk_read_bytes_total[2m])) / 1024 / 1024 > 50 + for: 5m + labels: + severity: warning + annotations: + summary: Host unusual disk read rate (instance {{ $labels.instance }}) + description: "Disk is probably reading too much data (> 50 MB/s)\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: HostUnusualDiskWriteRate + expr: sum by (instance) (rate(node_disk_written_bytes_total[2m])) / 1024 / 1024 > 50 + for: 2m + labels: + severity: warning + annotations: + summary: Host unusual disk write rate (instance {{ $labels.instance }}) + description: "Disk is probably writing too much data (> 50 MB/s)\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + # Please add ignored mountpoints in node_exporter parameters like + # "--collector.filesystem.ignored-mount-points=^/(sys|proc|dev|run)($|/)". + # Same rule using "node_filesystem_free_bytes" will fire when disk fills for non-root users. + - alert: HostOutOfDiskSpace + expr: (node_filesystem_avail_bytes * 100) / node_filesystem_size_bytes < 10 and ON (instance, device, mountpoint) node_filesystem_readonly == 0 + for: 2m + labels: + severity: warning + annotations: + summary: Host out of disk space (instance {{ $labels.instance }}) + description: "Disk is almost full (< 10% left)\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + # Please add ignored mountpoints in node_exporter parameters like + # "--collector.filesystem.ignored-mount-points=^/(sys|proc|dev|run)($|/)". + # Same rule using "node_filesystem_free_bytes" will fire when disk fills for non-root users. + - alert: HostDiskWillFillIn24Hours + expr: (node_filesystem_avail_bytes * 100) / node_filesystem_size_bytes < 10 and ON (instance, device, mountpoint) predict_linear(node_filesystem_avail_bytes{fstype!~"tmpfs"}[1h], 24 * 3600) < 0 and ON (instance, device, mountpoint) node_filesystem_readonly == 0 + for: 2m + labels: + severity: warning + annotations: + summary: Host disk will fill in 24 hours (instance {{ $labels.instance }}) + description: "Filesystem is predicted to run out of space within the next 24 hours at current write rate\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: HostOutOfInodes + expr: node_filesystem_files_free{mountpoint ="/rootfs"} / node_filesystem_files{mountpoint="/rootfs"} * 100 < 10 and ON (instance, device, mountpoint) node_filesystem_readonly{mountpoint="/rootfs"} == 0 + for: 2m + labels: + severity: warning + annotations: + summary: Host out of inodes (instance {{ $labels.instance }}) + description: "Disk is almost running out of available inodes (< 10% left)\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: HostInodesWillFillIn24Hours + expr: node_filesystem_files_free{mountpoint ="/rootfs"} / node_filesystem_files{mountpoint="/rootfs"} * 100 < 10 and predict_linear(node_filesystem_files_free{mountpoint="/rootfs"}[1h], 24 * 3600) < 0 and ON (instance, device, mountpoint) node_filesystem_readonly{mountpoint="/rootfs"} == 0 + for: 2m + labels: + severity: warning + annotations: + summary: Host inodes will fill in 24 hours (instance {{ $labels.instance }}) + description: "Filesystem is predicted to run out of inodes within the next 24 hours at current write rate\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: HostUnusualDiskReadLatency + expr: rate(node_disk_read_time_seconds_total[1m]) / rate(node_disk_reads_completed_total[1m]) > 0.1 and rate(node_disk_reads_completed_total[1m]) > 0 + for: 2m + labels: + severity: warning + annotations: + summary: Host unusual disk read latency (instance {{ $labels.instance }}) + description: "Disk latency is growing (read operations > 100ms)\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: HostUnusualDiskWriteLatency + expr: rate(node_disk_write_time_seconds_total[1m]) / rate(node_disk_writes_completed_total[1m]) > 0.1 and rate(node_disk_writes_completed_total[1m]) > 0 + for: 2m + labels: + severity: warning + annotations: + summary: Host unusual disk write latency (instance {{ $labels.instance }}) + description: "Disk latency is growing (write operations > 100ms)\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: HostHighCpuLoad + expr: 100 - (avg by(instance) (rate(node_cpu_seconds_total{mode="idle"}[2m])) * 100) > 80 + for: 0m + labels: + severity: warning + annotations: + summary: Host high CPU load (instance {{ $labels.instance }}) + description: "CPU load is > 80%\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: HostCpuStealNoisyNeighbor + expr: avg by(instance) (rate(node_cpu_seconds_total{mode="steal"}[5m])) * 100 > 10 + for: 0m + labels: + severity: warning + annotations: + summary: Host CPU steal noisy neighbor (instance {{ $labels.instance }}) + description: "CPU steal is > 10%. A noisy neighbor is killing VM performances or a spot instance may be out of credit.\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + # 1000 context switches is an arbitrary number. + # Alert threshold depends on nature of application. + # Please read: https://github.com/samber/awesome-prometheus-alerts/issues/58 + - alert: HostContextSwitching + expr: (rate(node_context_switches_total[5m])) / (count without(cpu, mode) (node_cpu_seconds_total{mode="idle"})) > 2000 + for: 0m + labels: + severity: warning + annotations: + summary: Host context switching (instance {{ $labels.instance }}) + description: "Context switching is growing on node (> 2000 / s)\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: HostSwapIsFillingUp + expr: (1 - (node_memory_SwapFree_bytes / node_memory_SwapTotal_bytes)) * 100 > 80 + for: 2m + labels: + severity: warning + annotations: + summary: Host swap is filling up (instance {{ $labels.instance }}) + description: "Swap is filling up (>80%)\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: HostSystemdServiceCrashed + expr: node_systemd_unit_state{state="failed"} == 1 + for: 0m + labels: + severity: warning + annotations: + summary: Host systemd service crashed (instance {{ $labels.instance }}) + description: "systemd service crashed\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: HostPhysicalComponentTooHot + expr: node_hwmon_temp_celsius > 75 + for: 5m + labels: + severity: warning + annotations: + summary: Host physical component too hot (instance {{ $labels.instance }}) + description: "Physical hardware component too hot\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: HostNodeOvertemperatureAlarm + expr: node_hwmon_temp_crit_alarm_celsius == 1 + for: 0m + labels: + severity: critical + annotations: + summary: Host node overtemperature alarm (instance {{ $labels.instance }}) + description: "Physical node temperature alarm triggered\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: HostRaidArrayGotInactive + expr: node_md_state{state="inactive"} > 0 + for: 0m + labels: + severity: critical + annotations: + summary: Host RAID array got inactive (instance {{ $labels.instance }}) + description: "RAID array {{ $labels.device }} is in degraded state due to one or more disks failures. Number of spare drives is insufficient to fix issue automatically.\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: HostRaidDiskFailure + expr: node_md_disks{state="failed"} > 0 + for: 2m + labels: + severity: warning + annotations: + summary: Host RAID disk failure (instance {{ $labels.instance }}) + description: "At least one device in RAID array on {{ $labels.instance }} failed. Array {{ $labels.md_device }} needs attention and possibly a disk swap\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + #- alert: HostKernelVersionDeviations + # expr: count(sum(label_replace(node_uname_info, "kernel", "$1", "release", "([0-9]+.[0-9]+.[0-9]+).*")) by (kernel)) > 1 + # for: 6h + # labels: + # severity: warning + # annotations: + # summary: Host kernel version deviations (instance {{ $labels.instance }}) + # description: "Different kernel versions are running\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: HostOomKillDetected + expr: increase(node_vmstat_oom_kill[1m]) > 0 + for: 0m + labels: + severity: warning + annotations: + summary: Host OOM kill detected (instance {{ $labels.instance }}) + description: "OOM kill detected\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: HostEdacCorrectableErrorsDetected + expr: increase(node_edac_correctable_errors_total[1m]) > 0 + for: 0m + labels: + severity: info + annotations: + summary: Host EDAC Correctable Errors detected (instance {{ $labels.instance }}) + description: "Host {{ $labels.instance }} has had {{ printf \"%.0f\" $value }} correctable memory errors reported by EDAC in the last 5 minutes.\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: HostEdacUncorrectableErrorsDetected + expr: node_edac_uncorrectable_errors_total > 0 + for: 0m + labels: + severity: warning + annotations: + summary: Host EDAC Uncorrectable Errors detected (instance {{ $labels.instance }}) + description: "Host {{ $labels.instance }} has had {{ printf \"%.0f\" $value }} uncorrectable memory errors reported by EDAC in the last 5 minutes.\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: HostNetworkReceiveErrors + expr: rate(node_network_receive_errs_total[2m]) / rate(node_network_receive_packets_total[2m]) > 0.01 + for: 2m + labels: + severity: warning + annotations: + summary: Host Network Receive Errors (instance {{ $labels.instance }}) + description: "Host {{ $labels.instance }} interface {{ $labels.device }} has encountered {{ printf \"%.0f\" $value }} receive errors in the last two minutes.\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: HostNetworkTransmitErrors + expr: rate(node_network_transmit_errs_total[2m]) / rate(node_network_transmit_packets_total[2m]) > 0.01 + for: 2m + labels: + severity: warning + annotations: + summary: Host Network Transmit Errors (instance {{ $labels.instance }}) + description: "Host {{ $labels.instance }} interface {{ $labels.device }} has encountered {{ printf \"%.0f\" $value }} transmit errors in the last two minutes.\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: HostNetworkInterfaceSaturated + expr: (rate(node_network_receive_bytes_total{device!~"^tap.*"}[1m]) + rate(node_network_transmit_bytes_total{device!~"^tap.*"}[1m])) / node_network_speed_bytes{device!~"^tap.*"} > 0.8 < 10000 + for: 1m + labels: + severity: warning + annotations: + summary: Host Network Interface Saturated (instance {{ $labels.instance }}) + description: "The network interface \"{{ $labels.device }}\" on \"{{ $labels.instance }}\" is getting overloaded.\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: HostNetworkBondDegraded + expr: (node_bonding_active - node_bonding_slaves) != 0 + for: 2m + labels: + severity: warning + annotations: + summary: Host Network Bond Degraded (instance {{ $labels.instance }}) + description: "Bond \"{{ $labels.device }}\" degraded on \"{{ $labels.instance }}\".\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: HostConntrackLimit + expr: node_nf_conntrack_entries / node_nf_conntrack_entries_limit > 0.8 + for: 5m + labels: + severity: warning + annotations: + summary: Host conntrack limit (instance {{ $labels.instance }}) + description: "The number of conntrack is approaching limit\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: HostClockSkew + expr: (node_timex_offset_seconds > 0.05 and deriv(node_timex_offset_seconds[5m]) >= 0) or (node_timex_offset_seconds < -0.05 and deriv(node_timex_offset_seconds[5m]) <= 0) + for: 2m + labels: + severity: warning + annotations: + summary: Host clock skew (instance {{ $labels.instance }}) + description: "Clock skew detected. Clock is out of sync. Ensure NTP is configured correctly on this host.\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: HostClockNotSynchronising + expr: min_over_time(node_timex_sync_status[1m]) == 0 and node_timex_maxerror_seconds >= 16 + for: 2m + labels: + severity: warning + annotations: + summary: Host clock not synchronising (instance {{ $labels.instance }}) + description: "Clock not synchronising. Ensure NTP is configured on this host.\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: HostRequiresReboot + expr: node_reboot_required > 0 + for: 4h + labels: + severity: info + annotations: + summary: Host requires reboot (instance {{ $labels.instance }}) + description: "{{ $labels.instance }} requires a reboot.\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" diff --git a/conf/prometheus/alerts/prometheus.yml b/conf/prometheus/alerts/prometheus.yml new file mode 100644 index 0000000..f55aabb --- /dev/null +++ b/conf/prometheus/alerts/prometheus.yml @@ -0,0 +1,236 @@ +groups: + - name: Prometheus + rules: + - alert: PrometheusJobMissing + expr: absent(up{job="prometheus"}) + for: 0m + labels: + severity: warning + annotations: + summary: Prometheus job missing (instance {{ $labels.instance }}) + description: "A Prometheus job has disappeared\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: PrometheusTargetMissing + expr: up == 0 + for: 0m + labels: + severity: critical + annotations: + summary: Prometheus target missing (instance {{ $labels.instance }}) + description: "A Prometheus target has disappeared. An exporter might be crashed.\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: PrometheusAllTargetsMissing + expr: count by (job) (up) == 0 + for: 0m + labels: + severity: critical + annotations: + summary: Prometheus all targets missing (instance {{ $labels.instance }}) + description: "A Prometheus job does not have living target anymore.\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: PrometheusConfigurationReloadFailure + expr: prometheus_config_last_reload_successful != 1 + for: 0m + labels: + severity: warning + annotations: + summary: Prometheus configuration reload failure (instance {{ $labels.instance }}) + description: "Prometheus configuration reload error\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: PrometheusTooManyRestarts + expr: changes(process_start_time_seconds{job=~"prometheus|pushgateway|alertmanager"}[15m]) > 2 + for: 0m + labels: + severity: warning + annotations: + summary: Prometheus too many restarts (instance {{ $labels.instance }}) + description: "Prometheus has restarted more than twice in the last 15 minutes. It might be crashlooping.\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: PrometheusAlertmanagerJobMissing + expr: absent(up{job="alertmanager"}) + for: 0m + labels: + severity: warning + annotations: + summary: Prometheus AlertManager job missing (instance {{ $labels.instance }}) + description: "A Prometheus AlertManager job has disappeared\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: PrometheusAlertmanagerConfigurationReloadFailure + expr: alertmanager_config_last_reload_successful != 1 + for: 0m + labels: + severity: warning + annotations: + summary: Prometheus AlertManager configuration reload failure (instance {{ $labels.instance }}) + description: "AlertManager configuration reload error\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: PrometheusAlertmanagerConfigNotSynced + expr: count(count_values("config_hash", alertmanager_config_hash)) > 1 + for: 0m + labels: + severity: warning + annotations: + summary: Prometheus AlertManager config not synced (instance {{ $labels.instance }}) + description: "Configurations of AlertManager cluster instances are out of sync\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: PrometheusAlertmanagerE2eDeadManSwitch + expr: vector(1) + for: 0m + labels: + severity: critical + annotations: + summary: Prometheus AlertManager E2E dead man switch (instance {{ $labels.instance }}) + description: "Prometheus DeadManSwitch is an always-firing alert. It's used as an end-to-end test of Prometheus through the Alertmanager.\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: PrometheusNotConnectedToAlertmanager + expr: prometheus_notifications_alertmanagers_discovered < 1 + for: 0m + labels: + severity: critical + annotations: + summary: Prometheus not connected to alertmanager (instance {{ $labels.instance }}) + description: "Prometheus cannot connect the alertmanager\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: PrometheusRuleEvaluationFailures + expr: increase(prometheus_rule_evaluation_failures_total[3m]) > 0 + for: 0m + labels: + severity: critical + annotations: + summary: Prometheus rule evaluation failures (instance {{ $labels.instance }}) + description: "Prometheus encountered {{ $value }} rule evaluation failures, leading to potentially ignored alerts.\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: PrometheusTemplateTextExpansionFailures + expr: increase(prometheus_template_text_expansion_failures_total[3m]) > 0 + for: 0m + labels: + severity: critical + annotations: + summary: Prometheus template text expansion failures (instance {{ $labels.instance }}) + description: "Prometheus encountered {{ $value }} template text expansion failures\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: PrometheusRuleEvaluationSlow + expr: prometheus_rule_group_last_duration_seconds > prometheus_rule_group_interval_seconds + for: 5m + labels: + severity: warning + annotations: + summary: Prometheus rule evaluation slow (instance {{ $labels.instance }}) + description: "Prometheus rule evaluation took more time than the scheduled interval. It indicates a slower storage backend access or too complex query.\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: PrometheusNotificationsBacklog + expr: min_over_time(prometheus_notifications_queue_length[10m]) > 0 + for: 0m + labels: + severity: warning + annotations: + summary: Prometheus notifications backlog (instance {{ $labels.instance }}) + description: "The Prometheus notification queue has not been empty for 10 minutes\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: PrometheusAlertmanagerNotificationFailing + expr: rate(alertmanager_notifications_failed_total[1m]) > 0 + for: 0m + labels: + severity: critical + annotations: + summary: Prometheus AlertManager notification failing (instance {{ $labels.instance }}) + description: "Alertmanager is failing sending notifications\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: PrometheusTargetEmpty + expr: prometheus_sd_discovered_targets == 0 + for: 0m + labels: + severity: critical + annotations: + summary: Prometheus target empty (instance {{ $labels.instance }}) + description: "Prometheus has no target in service discovery\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: PrometheusTargetScrapingSlow + expr: prometheus_target_interval_length_seconds{quantile="0.9"} > 60 + for: 5m + labels: + severity: warning + annotations: + summary: Prometheus target scraping slow (instance {{ $labels.instance }}) + description: "Prometheus is scraping exporters slowly\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: PrometheusLargeScrape + expr: increase(prometheus_target_scrapes_exceeded_sample_limit_total[10m]) > 10 + for: 5m + labels: + severity: warning + annotations: + summary: Prometheus large scrape (instance {{ $labels.instance }}) + description: "Prometheus has many scrapes that exceed the sample limit\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: PrometheusTargetScrapeDuplicate + expr: increase(prometheus_target_scrapes_sample_duplicate_timestamp_total[5m]) > 0 + for: 0m + labels: + severity: warning + annotations: + summary: Prometheus target scrape duplicate (instance {{ $labels.instance }}) + description: "Prometheus has many samples rejected due to duplicate timestamps but different values\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: PrometheusTsdbCheckpointCreationFailures + expr: increase(prometheus_tsdb_checkpoint_creations_failed_total[1m]) > 0 + for: 0m + labels: + severity: critical + annotations: + summary: Prometheus TSDB checkpoint creation failures (instance {{ $labels.instance }}) + description: "Prometheus encountered {{ $value }} checkpoint creation failures\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: PrometheusTsdbCheckpointDeletionFailures + expr: increase(prometheus_tsdb_checkpoint_deletions_failed_total[1m]) > 0 + for: 0m + labels: + severity: critical + annotations: + summary: Prometheus TSDB checkpoint deletion failures (instance {{ $labels.instance }}) + description: "Prometheus encountered {{ $value }} checkpoint deletion failures\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: PrometheusTsdbCompactionsFailed + expr: increase(prometheus_tsdb_compactions_failed_total[1m]) > 0 + for: 0m + labels: + severity: critical + annotations: + summary: Prometheus TSDB compactions failed (instance {{ $labels.instance }}) + description: "Prometheus encountered {{ $value }} TSDB compactions failures\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: PrometheusTsdbHeadTruncationsFailed + expr: increase(prometheus_tsdb_head_truncations_failed_total[1m]) > 0 + for: 0m + labels: + severity: critical + annotations: + summary: Prometheus TSDB head truncations failed (instance {{ $labels.instance }}) + description: "Prometheus encountered {{ $value }} TSDB head truncation failures\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: PrometheusTsdbReloadFailures + expr: increase(prometheus_tsdb_reloads_failures_total[1m]) > 0 + for: 0m + labels: + severity: critical + annotations: + summary: Prometheus TSDB reload failures (instance {{ $labels.instance }}) + description: "Prometheus encountered {{ $value }} TSDB reload failures\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: PrometheusTsdbWalCorruptions + expr: increase(prometheus_tsdb_wal_corruptions_total[1m]) > 0 + for: 0m + labels: + severity: critical + annotations: + summary: Prometheus TSDB WAL corruptions (instance {{ $labels.instance }}) + description: "Prometheus encountered {{ $value }} TSDB WAL corruptions\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" + + - alert: PrometheusTsdbWalTruncationsFailed + expr: increase(prometheus_tsdb_wal_truncations_failed_total[1m]) > 0 + for: 0m + labels: + severity: critical + annotations: + summary: Prometheus TSDB WAL truncations failed (instance {{ $labels.instance }}) + description: "Prometheus encountered {{ $value }} TSDB WAL truncation failures\n VALUE = {{ $value }}\n LABELS = {{ $labels }}" diff --git a/conf/prometheus/prometheus.yml b/conf/prometheus/prometheus.yml new file mode 100644 index 0000000..c083c47 --- /dev/null +++ b/conf/prometheus/prometheus.yml @@ -0,0 +1,78 @@ +global: + scrape_interval: 20s + +rule_files: + - alerts/*.yml + +alerting: + alertmanagers: + - static_configs: + - targets: + - alertmanager:9093 + +scrape_configs: + - job_name: "prometheus" + scrape_interval: 5s + static_configs: + - targets: ["localhost:9090"] + + - job_name: "alertmanager" + scrape_interval: 5s + static_configs: + - targets: ["alertmanager:9093"] + + - job_name: "node" + static_configs: + - targets: ["node-exporter:9100"] + + - job_name: "cadvisor" + static_configs: + - targets: ["cadvisor:8080"] + + - job_name: "node_nyyu" + scheme: https + metrics_path: /node/metrics + static_configs: + - targets: ["nyyu.dev:443"] + + # - job_name: "smokeping" + # static_configs: + # - targets: ["smokeping:9374"] + + - job_name: "blackbox" + metrics_path: /probe + params: + module: [http_2xx] + dns_sd_configs: + - names: + - nyyu.dev + type: A + port: 443 + relabel_configs: + - source_labels: [__address__] + target_label: __param_target + replacement: https://$1/ + - source_labels: [__param_target] + target_label: target + - target_label: __address__ + replacement: blackbox:9115 + - source_labels: [__meta_dns_name] + target_label: __param_hostname + - source_labels: [__meta_dns_name] + target_label: vhost + + - job_name: "blackbox_icmp" + metrics_path: /probe + params: + module: [icmp] + static_configs: + - targets: + - 1.1.1.1 + - nyyu.dev + relabel_configs: + - source_labels: [__address__] + target_label: __param_target + - source_labels: [__param_target] + target_label: target + - target_label: __address__ + replacement: blackbox:9115 \ No newline at end of file diff --git a/docker-compose.yml b/docker-compose.yml new file mode 100644 index 0000000..7919e80 --- /dev/null +++ b/docker-compose.yml @@ -0,0 +1,120 @@ +version: '3.8' + +networks: + monitoring: + driver: bridge + +services: + grafana: + # ARM bug in rate interval : https://github.com/grafana/grafana/issues/43002 + image: grafana/grafana:8.2.7 + container_name: grafana + restart: unless-stopped + volumes: + - ./data/grafana:/var/lib/grafana + - ./conf/grafana/provisioning:/etc/grafana/provisioning + ports: + - 3000:3000 + environment: + - GF_SECURITY_ADMIN_USER=admin + - GF_SECURITY_ADMIN_PASSWORD=${GRAFANA_PASSWORD} + - GF_USERS_ALLOW_SIGN_UP=false + networks: + - monitoring + prometheus: + image: prom/prometheus + container_name: prometheus + restart: unless-stopped + volumes: + - ./conf/prometheus:/etc/prometheus + - ./data/prometheus:/prometheus + ports: + - 9090:9090 + command: + - '--config.file=/etc/prometheus/prometheus.yml' + - '--storage.tsdb.path=/prometheus' + - '--web.console.libraries=/etc/prometheus/console_libraries' + - '--web.console.templates=/etc/prometheus/consoles' + - '--web.enable-lifecycle' + - "--web.external-url=http://${EXTERNAL_IP}:9090" + expose: + - 9090 + networks: + - monitoring + alertmanager: + image: prom/alertmanager + container_name: alertmanager + restart: unless-stopped + volumes: + - ./conf/alertmanager:/etc/alertmanager + - ./data/alertmanager:/alertmanager + ports: + - 9093:9093 + command: + - '--config.file=/etc/alertmanager/alertmanager.yml' + - '--storage.path=/alertmanager' + - "--web.external-url=http://${EXTERNAL_IP}:9093" + expose: + - 9093 + networks: + - monitoring + node-exporter: + image: prom/node-exporter + container_name: node-exporter + restart: unless-stopped + volumes: + - /proc:/host/proc:ro + - /sys:/host/sys:ro + - /:/rootfs:ro + command: + - '--path.procfs=/host/proc' + - '--path.rootfs=/rootfs' + - '--path.sysfs=/host/sys' + - '--collector.filesystem.mount-points-exclude=^/(sys|proc|dev|host|etc)($$|/)' + expose: + - 9100 + networks: + - monitoring + blackbox_exporter: + image: prom/blackbox-exporter + container_name: blackbox + restart: unless-stopped + expose: + - 9115 + networks: + - monitoring + # smokeping: + # image: quay.io/superq/smokeping-prober + # container_name: smokeping + # restart: unless-stopped + # command: nyyu.dev + # privileged: true + # expose: + # - 9374 + # networks: + # - monitoring + cadvisor: + # ARM image + image: justrobin/cadvisor:v0.44.0 + container_name: cadvisor + restart: unless-stopped + command: + - '--housekeeping_interval=10s' + - '--raw_cgroup_prefix_whitelist=/docker/' + - '--disable_metrics=cpu_topology,hugetlb' + privileged: true + pid: 'host' + ports: + - '8040:8080' + volumes: + - '/:/rootfs:ro' + - '/var/run:/var/run:ro' + - '/sys:/sys:ro' + - '/var/lib/docker/:/var/lib/docker:ro' + - '/dev/disk/:/dev/disk:ro' + devices: + - '/dev/kmsg:/dev/kmsg' + expose: + - 8080 + networks: + - monitoring \ No newline at end of file