EDGEPOD-157 Upgrade Aether dashboards for better support of multi-sites
Change-Id: I525b8a75739a0e8271dd0a532ba52faf2627534c
diff --git a/nem-monitoring/grafana-dashboards/multi-site-k8s.json b/nem-monitoring/grafana-dashboards/multi-site-k8s.json
new file mode 100644
index 0000000..acc4fe6
--- /dev/null
+++ b/nem-monitoring/grafana-dashboards/multi-site-k8s.json
@@ -0,0 +1,2329 @@
+{
+ "annotations": {
+ "list": [
+ {
+ "builtIn": 1,
+ "datasource": "-- Grafana --",
+ "enable": true,
+ "hide": true,
+ "iconColor": "rgba(0, 211, 255, 1)",
+ "name": "Annotations & Alerts",
+ "type": "dashboard"
+ }
+ ]
+ },
+ "description": "Monitors K8S cluster using Prometheus.",
+ "editable": true,
+ "gnetId": 8721,
+ "graphTooltip": 1,
+ "id": 5,
+ "iteration": 1580676696605,
+ "links": [],
+ "panels": [
+ {
+ "collapsed": false,
+ "datasource": null,
+ "gridPos": {
+ "h": 1,
+ "w": 24,
+ "x": 0,
+ "y": 0
+ },
+ "id": 34,
+ "panels": [],
+ "title": "Total usage",
+ "type": "row"
+ },
+ {
+ "cacheTimeout": null,
+ "colorBackground": false,
+ "colorValue": true,
+ "colors": [
+ "rgba(50, 172, 45, 0.97)",
+ "rgba(237, 129, 40, 0.89)",
+ "rgba(245, 54, 54, 0.9)"
+ ],
+ "datasource": "Prometheus",
+ "decimals": 2,
+ "editable": true,
+ "error": false,
+ "format": "percent",
+ "gauge": {
+ "maxValue": 100,
+ "minValue": 0,
+ "show": true,
+ "thresholdLabels": false,
+ "thresholdMarkers": true
+ },
+ "gridPos": {
+ "h": 5,
+ "w": 8,
+ "x": 0,
+ "y": 1
+ },
+ "height": "180px",
+ "id": 6,
+ "interval": null,
+ "isNew": true,
+ "links": [],
+ "mappingType": 1,
+ "mappingTypes": [
+ {
+ "name": "value to text",
+ "value": 1
+ },
+ {
+ "name": "range to text",
+ "value": 2
+ }
+ ],
+ "maxDataPoints": 100,
+ "nullPointMode": "connected",
+ "nullText": null,
+ "options": {},
+ "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": "sum (rate (container_cpu_usage_seconds_total{id=\"/\",datacenter=\"$Site\",kubernetes_io_hostname=~\"^$Node$\"}[5m])) / sum (machine_cpu_cores{datacenter=\"$Site\",kubernetes_io_hostname=~\"^$Node$\"}) * 100",
+ "format": "time_series",
+ "interval": "10s",
+ "intervalFactor": 1,
+ "refId": "A",
+ "step": 10
+ }
+ ],
+ "thresholds": "65, 90",
+ "title": "Cluster CPU usage (5m avg)",
+ "type": "singlestat",
+ "valueFontSize": "50%",
+ "valueMaps": [
+ {
+ "op": "=",
+ "text": "N/A",
+ "value": "null"
+ }
+ ],
+ "valueName": "current"
+ },
+ {
+ "cacheTimeout": null,
+ "colorBackground": false,
+ "colorValue": true,
+ "colors": [
+ "rgba(50, 172, 45, 0.97)",
+ "rgba(237, 129, 40, 0.89)",
+ "rgba(245, 54, 54, 0.9)"
+ ],
+ "datasource": "Prometheus",
+ "editable": true,
+ "error": false,
+ "format": "percent",
+ "gauge": {
+ "maxValue": 100,
+ "minValue": 0,
+ "show": true,
+ "thresholdLabels": false,
+ "thresholdMarkers": true
+ },
+ "gridPos": {
+ "h": 5,
+ "w": 8,
+ "x": 8,
+ "y": 1
+ },
+ "height": "180px",
+ "id": 4,
+ "interval": null,
+ "isNew": true,
+ "links": [],
+ "mappingType": 1,
+ "mappingTypes": [
+ {
+ "name": "value to text",
+ "value": 1
+ },
+ {
+ "name": "range to text",
+ "value": 2
+ }
+ ],
+ "maxDataPoints": 100,
+ "nullPointMode": "connected",
+ "nullText": null,
+ "options": {},
+ "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": "sum (container_memory_working_set_bytes{id=\"/\",datacenter=\"$Site\",kubernetes_io_hostname=~\"^$Node$\"}) / sum (machine_memory_bytes{datacenter=\"$Site\",kubernetes_io_hostname=~\"^$Node$\"}) * 100",
+ "interval": "10s",
+ "intervalFactor": 1,
+ "refId": "A",
+ "step": 10
+ }
+ ],
+ "thresholds": "65, 90",
+ "title": "Cluster memory usage",
+ "type": "singlestat",
+ "valueFontSize": "50%",
+ "valueMaps": [
+ {
+ "op": "=",
+ "text": "N/A",
+ "value": "null"
+ }
+ ],
+ "valueName": "current"
+ },
+ {
+ "cacheTimeout": null,
+ "colorBackground": false,
+ "colorValue": true,
+ "colors": [
+ "rgba(50, 172, 45, 0.97)",
+ "rgba(237, 129, 40, 0.89)",
+ "rgba(245, 54, 54, 0.9)"
+ ],
+ "datasource": "Prometheus",
+ "decimals": 2,
+ "editable": true,
+ "error": false,
+ "format": "percent",
+ "gauge": {
+ "maxValue": 100,
+ "minValue": 0,
+ "show": true,
+ "thresholdLabels": false,
+ "thresholdMarkers": true
+ },
+ "gridPos": {
+ "h": 5,
+ "w": 8,
+ "x": 16,
+ "y": 1
+ },
+ "height": "180px",
+ "id": 7,
+ "interval": null,
+ "isNew": true,
+ "links": [],
+ "mappingType": 1,
+ "mappingTypes": [
+ {
+ "name": "value to text",
+ "value": 1
+ },
+ {
+ "name": "range to text",
+ "value": 2
+ }
+ ],
+ "maxDataPoints": 100,
+ "nullPointMode": "connected",
+ "nullText": null,
+ "options": {},
+ "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": "sum (container_fs_usage_bytes{device=~\"^/dev/[sv]d[a-z][1-9]$\",id=\"/\",datacenter=\"$Site\",kubernetes_io_hostname=~\"^$Node$\"}) / sum (container_fs_limit_bytes{device=~\"^/dev/[sv]d[a-z][1-9]$\",id=\"/\",datacenter=\"$Site\",kubernetes_io_hostname=~\"^$Node$\"}) * 100",
+ "format": "time_series",
+ "interval": "10s",
+ "intervalFactor": 1,
+ "legendFormat": "",
+ "metric": "",
+ "refId": "A",
+ "step": 10
+ }
+ ],
+ "thresholds": "65, 90",
+ "title": "Cluster filesystem usage",
+ "type": "singlestat",
+ "valueFontSize": "50%",
+ "valueMaps": [
+ {
+ "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": "Prometheus",
+ "decimals": 2,
+ "editable": true,
+ "error": false,
+ "format": "none",
+ "gauge": {
+ "maxValue": 100,
+ "minValue": 0,
+ "show": false,
+ "thresholdLabels": false,
+ "thresholdMarkers": true
+ },
+ "gridPos": {
+ "h": 3,
+ "w": 4,
+ "x": 0,
+ "y": 6
+ },
+ "height": "1px",
+ "id": 11,
+ "interval": null,
+ "isNew": true,
+ "links": [],
+ "mappingType": 1,
+ "mappingTypes": [
+ {
+ "name": "value to text",
+ "value": 1
+ },
+ {
+ "name": "range to text",
+ "value": 2
+ }
+ ],
+ "maxDataPoints": 100,
+ "nullPointMode": "connected",
+ "nullText": null,
+ "options": {},
+ "postfix": " cores",
+ "postfixFontSize": "30%",
+ "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": "sum (rate (container_cpu_usage_seconds_total{id=\"/\",datacenter=\"$Site\",kubernetes_io_hostname=~\"^$Node$\"}[5m]))",
+ "format": "time_series",
+ "interval": "10s",
+ "intervalFactor": 1,
+ "refId": "A",
+ "step": 10
+ }
+ ],
+ "thresholds": "",
+ "title": "Used",
+ "type": "singlestat",
+ "valueFontSize": "50%",
+ "valueMaps": [
+ {
+ "op": "=",
+ "text": "N/A",
+ "value": "null"
+ }
+ ],
+ "valueName": "current"
+ },
+ {
+ "cacheTimeout": null,
+ "colorBackground": false,
+ "colorValue": false,
+ "colors": [
+ "rgba(50, 172, 45, 0.97)",
+ "rgba(237, 129, 40, 0.89)",
+ "rgba(245, 54, 54, 0.9)"
+ ],
+ "datasource": "Prometheus",
+ "decimals": 2,
+ "editable": true,
+ "error": false,
+ "format": "none",
+ "gauge": {
+ "maxValue": 100,
+ "minValue": 0,
+ "show": false,
+ "thresholdLabels": false,
+ "thresholdMarkers": true
+ },
+ "gridPos": {
+ "h": 3,
+ "w": 4,
+ "x": 4,
+ "y": 6
+ },
+ "height": "1px",
+ "id": 12,
+ "interval": null,
+ "isNew": true,
+ "links": [],
+ "mappingType": 1,
+ "mappingTypes": [
+ {
+ "name": "value to text",
+ "value": 1
+ },
+ {
+ "name": "range to text",
+ "value": 2
+ }
+ ],
+ "maxDataPoints": 100,
+ "nullPointMode": "connected",
+ "nullText": null,
+ "options": {},
+ "postfix": " cores",
+ "postfixFontSize": "30%",
+ "prefix": "",
+ "prefixFontSize": "50%",
+ "rangeMaps": [
+ {
+ "from": "null",
+ "text": "N/A",
+ "to": "null"
+ }
+ ],
+ "sparkline": {
+ "fillColor": "rgba(31, 118, 189, 0.18)",
+ "full": false,
+ "lineColor": "rgb(31, 120, 193)",
+ "show": false
+ },
+ "tableColumn": "",
+ "targets": [
+ {
+ "expr": "sum (machine_cpu_cores{datacenter=\"$Site\",kubernetes_io_hostname=~\"^$Node$\"})",
+ "interval": "10s",
+ "intervalFactor": 1,
+ "refId": "A",
+ "step": 10
+ }
+ ],
+ "thresholds": "",
+ "title": "Total",
+ "type": "singlestat",
+ "valueFontSize": "50%",
+ "valueMaps": [
+ {
+ "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": "Prometheus",
+ "decimals": 2,
+ "editable": true,
+ "error": false,
+ "format": "bytes",
+ "gauge": {
+ "maxValue": 100,
+ "minValue": 0,
+ "show": false,
+ "thresholdLabels": false,
+ "thresholdMarkers": true
+ },
+ "gridPos": {
+ "h": 3,
+ "w": 4,
+ "x": 8,
+ "y": 6
+ },
+ "height": "1px",
+ "id": 9,
+ "interval": null,
+ "isNew": true,
+ "links": [],
+ "mappingType": 1,
+ "mappingTypes": [
+ {
+ "name": "value to text",
+ "value": 1
+ },
+ {
+ "name": "range to text",
+ "value": 2
+ }
+ ],
+ "maxDataPoints": 100,
+ "nullPointMode": "connected",
+ "nullText": null,
+ "options": {},
+ "postfix": "",
+ "postfixFontSize": "20%",
+ "prefix": "",
+ "prefixFontSize": "20%",
+ "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": "sum (container_memory_working_set_bytes{id=\"/\",datacenter=\"$Site\",kubernetes_io_hostname=~\"^$Node$\"})",
+ "interval": "10s",
+ "intervalFactor": 1,
+ "refId": "A",
+ "step": 10
+ }
+ ],
+ "thresholds": "",
+ "title": "Used",
+ "type": "singlestat",
+ "valueFontSize": "50%",
+ "valueMaps": [
+ {
+ "op": "=",
+ "text": "N/A",
+ "value": "null"
+ }
+ ],
+ "valueName": "current"
+ },
+ {
+ "cacheTimeout": null,
+ "colorBackground": false,
+ "colorValue": false,
+ "colors": [
+ "rgba(50, 172, 45, 0.97)",
+ "rgba(237, 129, 40, 0.89)",
+ "rgba(245, 54, 54, 0.9)"
+ ],
+ "datasource": "Prometheus",
+ "decimals": 2,
+ "editable": true,
+ "error": false,
+ "format": "bytes",
+ "gauge": {
+ "maxValue": 100,
+ "minValue": 0,
+ "show": false,
+ "thresholdLabels": false,
+ "thresholdMarkers": true
+ },
+ "gridPos": {
+ "h": 3,
+ "w": 4,
+ "x": 12,
+ "y": 6
+ },
+ "height": "1px",
+ "id": 10,
+ "interval": null,
+ "isNew": true,
+ "links": [],
+ "mappingType": 1,
+ "mappingTypes": [
+ {
+ "name": "value to text",
+ "value": 1
+ },
+ {
+ "name": "range to text",
+ "value": 2
+ }
+ ],
+ "maxDataPoints": 100,
+ "nullPointMode": "connected",
+ "nullText": null,
+ "options": {},
+ "postfix": "",
+ "postfixFontSize": "50%",
+ "prefix": "",
+ "prefixFontSize": "50%",
+ "rangeMaps": [
+ {
+ "from": "null",
+ "text": "N/A",
+ "to": "null"
+ }
+ ],
+ "sparkline": {
+ "fillColor": "rgba(31, 118, 189, 0.18)",
+ "full": false,
+ "lineColor": "rgb(31, 120, 193)",
+ "show": false
+ },
+ "tableColumn": "",
+ "targets": [
+ {
+ "expr": "sum (machine_memory_bytes{datacenter=\"$Site\",kubernetes_io_hostname=~\"^$Node$\"})",
+ "interval": "10s",
+ "intervalFactor": 1,
+ "refId": "A",
+ "step": 10
+ }
+ ],
+ "thresholds": "",
+ "title": "Total",
+ "type": "singlestat",
+ "valueFontSize": "50%",
+ "valueMaps": [
+ {
+ "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": "Prometheus",
+ "decimals": 2,
+ "editable": true,
+ "error": false,
+ "format": "bytes",
+ "gauge": {
+ "maxValue": 100,
+ "minValue": 0,
+ "show": false,
+ "thresholdLabels": false,
+ "thresholdMarkers": true
+ },
+ "gridPos": {
+ "h": 3,
+ "w": 4,
+ "x": 16,
+ "y": 6
+ },
+ "height": "1px",
+ "id": 13,
+ "interval": null,
+ "isNew": true,
+ "links": [],
+ "mappingType": 1,
+ "mappingTypes": [
+ {
+ "name": "value to text",
+ "value": 1
+ },
+ {
+ "name": "range to text",
+ "value": 2
+ }
+ ],
+ "maxDataPoints": 100,
+ "nullPointMode": "connected",
+ "nullText": null,
+ "options": {},
+ "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": "sum (container_fs_usage_bytes{device=~\"^/dev/[sv]d[a-z][1-9]$\",id=\"/\",datacenter=\"$Site\",kubernetes_io_hostname=~\"^$Node$\"})",
+ "format": "time_series",
+ "interval": "10s",
+ "intervalFactor": 1,
+ "refId": "A",
+ "step": 10
+ }
+ ],
+ "thresholds": "",
+ "title": "Used",
+ "type": "singlestat",
+ "valueFontSize": "50%",
+ "valueMaps": [
+ {
+ "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": "Prometheus",
+ "decimals": 2,
+ "editable": true,
+ "error": false,
+ "format": "bytes",
+ "gauge": {
+ "maxValue": 100,
+ "minValue": 0,
+ "show": false,
+ "thresholdLabels": false,
+ "thresholdMarkers": true
+ },
+ "gridPos": {
+ "h": 3,
+ "w": 4,
+ "x": 20,
+ "y": 6
+ },
+ "height": "1px",
+ "id": 14,
+ "interval": null,
+ "isNew": true,
+ "links": [],
+ "mappingType": 1,
+ "mappingTypes": [
+ {
+ "name": "value to text",
+ "value": 1
+ },
+ {
+ "name": "range to text",
+ "value": 2
+ }
+ ],
+ "maxDataPoints": 100,
+ "nullPointMode": "connected",
+ "nullText": null,
+ "options": {},
+ "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": "sum (container_fs_limit_bytes{device=~\"^overlay$\",id=\"/\",datacenter=\"$Site\",kubernetes_io_hostname=~\"^$Node$\"})",
+ "format": "time_series",
+ "interval": "10s",
+ "intervalFactor": 1,
+ "refId": "A",
+ "step": 10
+ }
+ ],
+ "thresholds": "",
+ "title": "Total",
+ "type": "singlestat",
+ "valueFontSize": "50%",
+ "valueMaps": [
+ {
+ "op": "=",
+ "text": "N/A",
+ "value": "null"
+ }
+ ],
+ "valueName": "current"
+ },
+ {
+ "collapsed": false,
+ "datasource": null,
+ "gridPos": {
+ "h": 1,
+ "w": 24,
+ "x": 0,
+ "y": 9
+ },
+ "id": 33,
+ "panels": [],
+ "title": "Usage total",
+ "type": "row"
+ },
+ {
+ "aliasColors": {},
+ "bars": false,
+ "dashLength": 10,
+ "dashes": false,
+ "datasource": "Prometheus",
+ "decimals": 2,
+ "editable": true,
+ "error": false,
+ "fill": 3,
+ "fillGradient": 0,
+ "grid": {},
+ "gridPos": {
+ "h": 7,
+ "w": 8,
+ "x": 0,
+ "y": 10
+ },
+ "height": "200px",
+ "id": 32,
+ "isNew": true,
+ "legend": {
+ "alignAsTable": false,
+ "avg": true,
+ "current": true,
+ "max": false,
+ "min": false,
+ "rightSide": false,
+ "show": false,
+ "sideWidth": 200,
+ "sort": "current",
+ "sortDesc": true,
+ "total": false,
+ "values": true
+ },
+ "lines": true,
+ "linewidth": 1,
+ "links": [],
+ "nullPointMode": "null",
+ "options": {
+ "dataLinks": []
+ },
+ "percentage": false,
+ "pointradius": 0.5,
+ "points": true,
+ "renderer": "flot",
+ "seriesOverrides": [],
+ "spaceLength": 10,
+ "stack": false,
+ "steppedLine": false,
+ "targets": [
+ {
+ "expr": "sum (rate (container_network_receive_bytes_total{datacenter=\"$Site\",kubernetes_io_hostname=~\"^$Node$\"}[5m]))",
+ "format": "time_series",
+ "interval": "10s",
+ "intervalFactor": 4,
+ "legendFormat": "Received",
+ "metric": "network",
+ "refId": "A",
+ "step": 10
+ },
+ {
+ "expr": "- sum (rate (container_network_transmit_bytes_total{datacenter=\"$Site\",kubernetes_io_hostname=~\"^$Node$\"}[5m]))",
+ "format": "time_series",
+ "interval": "10s",
+ "intervalFactor": 4,
+ "legendFormat": "Sent",
+ "metric": "network",
+ "refId": "B",
+ "step": 10
+ }
+ ],
+ "thresholds": [],
+ "timeFrom": null,
+ "timeRegions": [],
+ "timeShift": null,
+ "title": "Network I/O pressure",
+ "tooltip": {
+ "msResolution": false,
+ "shared": true,
+ "sort": 0,
+ "value_type": "cumulative"
+ },
+ "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": "Bps",
+ "label": null,
+ "logBase": 1,
+ "max": null,
+ "min": null,
+ "show": false
+ }
+ ],
+ "yaxis": {
+ "align": false,
+ "alignLevel": null
+ }
+ },
+ {
+ "aliasColors": {},
+ "bars": false,
+ "dashLength": 10,
+ "dashes": false,
+ "datasource": "Prometheus",
+ "decimals": 2,
+ "editable": true,
+ "error": false,
+ "fill": 0,
+ "fillGradient": 0,
+ "grid": {},
+ "gridPos": {
+ "h": 7,
+ "w": 8,
+ "x": 8,
+ "y": 10
+ },
+ "height": "200px",
+ "id": 45,
+ "isNew": true,
+ "legend": {
+ "alignAsTable": false,
+ "avg": true,
+ "current": true,
+ "max": false,
+ "min": false,
+ "rightSide": false,
+ "show": false,
+ "sideWidth": 200,
+ "sort": "current",
+ "sortDesc": true,
+ "total": false,
+ "values": true
+ },
+ "lines": true,
+ "linewidth": 1,
+ "links": [],
+ "nullPointMode": "null",
+ "options": {
+ "dataLinks": []
+ },
+ "percentage": false,
+ "pointradius": 0.5,
+ "points": true,
+ "renderer": "flot",
+ "seriesOverrides": [],
+ "spaceLength": 10,
+ "stack": false,
+ "steppedLine": false,
+ "targets": [
+ {
+ "expr": "sum (rate (container_cpu_usage_seconds_total{image!=\"\",name=~\"^k8s_.*\",datacenter=\"$Site\",kubernetes_io_hostname=~\"^$Node$\",namespace=~\"$Namespace\"}[5m])) by (namespace)",
+ "format": "time_series",
+ "interval": "10s",
+ "intervalFactor": 4,
+ "legendFormat": "{{ namespace}}",
+ "metric": "network",
+ "refId": "A",
+ "step": 10
+ }
+ ],
+ "thresholds": [],
+ "timeFrom": null,
+ "timeRegions": [],
+ "timeShift": null,
+ "title": "CPU use by namespace",
+ "tooltip": {
+ "msResolution": false,
+ "shared": true,
+ "sort": 0,
+ "value_type": "cumulative"
+ },
+ "type": "graph",
+ "xaxis": {
+ "buckets": null,
+ "mode": "time",
+ "name": null,
+ "show": true,
+ "values": []
+ },
+ "yaxes": [
+ {
+ "format": "percentunit",
+ "label": null,
+ "logBase": 1,
+ "max": null,
+ "min": null,
+ "show": true
+ },
+ {
+ "format": "Bps",
+ "label": null,
+ "logBase": 1,
+ "max": null,
+ "min": null,
+ "show": false
+ }
+ ],
+ "yaxis": {
+ "align": false,
+ "alignLevel": null
+ }
+ },
+ {
+ "aliasColors": {},
+ "bars": false,
+ "dashLength": 10,
+ "dashes": false,
+ "datasource": "Prometheus",
+ "decimals": 2,
+ "editable": true,
+ "error": false,
+ "fill": 0,
+ "fillGradient": 0,
+ "grid": {},
+ "gridPos": {
+ "h": 7,
+ "w": 8,
+ "x": 16,
+ "y": 10
+ },
+ "height": "200px",
+ "id": 46,
+ "isNew": true,
+ "legend": {
+ "alignAsTable": false,
+ "avg": true,
+ "current": true,
+ "max": false,
+ "min": false,
+ "rightSide": false,
+ "show": false,
+ "sideWidth": 200,
+ "sort": "current",
+ "sortDesc": true,
+ "total": false,
+ "values": true
+ },
+ "lines": true,
+ "linewidth": 1,
+ "links": [],
+ "nullPointMode": "null",
+ "options": {
+ "dataLinks": []
+ },
+ "percentage": false,
+ "pointradius": 0.5,
+ "points": true,
+ "renderer": "flot",
+ "seriesOverrides": [],
+ "spaceLength": 10,
+ "stack": false,
+ "steppedLine": false,
+ "targets": [
+ {
+ "expr": "sum (container_memory_working_set_bytes{image!=\"\",name=~\"^k8s_.*\",datacenter=\"$Site\",kubernetes_io_hostname=~\"^$Node$\",namespace=~\"$Namespace\"}) by (namespace)",
+ "format": "time_series",
+ "interval": "10s",
+ "intervalFactor": 4,
+ "legendFormat": "{{ namespace}}",
+ "metric": "network",
+ "refId": "A",
+ "step": 10
+ }
+ ],
+ "thresholds": [],
+ "timeFrom": null,
+ "timeRegions": [],
+ "timeShift": null,
+ "title": "Memory use by namespace",
+ "tooltip": {
+ "msResolution": false,
+ "shared": true,
+ "sort": 0,
+ "value_type": "cumulative"
+ },
+ "type": "graph",
+ "xaxis": {
+ "buckets": null,
+ "mode": "time",
+ "name": null,
+ "show": true,
+ "values": []
+ },
+ "yaxes": [
+ {
+ "format": "decbytes",
+ "label": null,
+ "logBase": 1,
+ "max": null,
+ "min": null,
+ "show": true
+ },
+ {
+ "format": "Bps",
+ "label": null,
+ "logBase": 1,
+ "max": null,
+ "min": null,
+ "show": false
+ }
+ ],
+ "yaxis": {
+ "align": false,
+ "alignLevel": null
+ }
+ },
+ {
+ "collapsed": false,
+ "datasource": null,
+ "gridPos": {
+ "h": 1,
+ "w": 24,
+ "x": 0,
+ "y": 17
+ },
+ "id": 35,
+ "panels": [],
+ "title": "Pods CPU usage",
+ "type": "row"
+ },
+ {
+ "aliasColors": {},
+ "bars": false,
+ "dashLength": 10,
+ "dashes": false,
+ "datasource": "Prometheus",
+ "decimals": 3,
+ "editable": true,
+ "error": false,
+ "fill": 1,
+ "fillGradient": 0,
+ "grid": {},
+ "gridPos": {
+ "h": 7,
+ "w": 24,
+ "x": 0,
+ "y": 18
+ },
+ "height": "",
+ "id": 17,
+ "isNew": true,
+ "legend": {
+ "alignAsTable": true,
+ "avg": true,
+ "current": true,
+ "max": false,
+ "min": false,
+ "rightSide": true,
+ "show": true,
+ "sort": "current",
+ "sortDesc": true,
+ "total": false,
+ "values": true
+ },
+ "lines": true,
+ "linewidth": 0,
+ "links": [],
+ "nullPointMode": "null",
+ "options": {
+ "dataLinks": []
+ },
+ "percentage": false,
+ "pointradius": 0.5,
+ "points": true,
+ "renderer": "flot",
+ "seriesOverrides": [],
+ "spaceLength": 10,
+ "stack": false,
+ "steppedLine": true,
+ "targets": [
+ {
+ "expr": "sum (rate (container_cpu_usage_seconds_total{image!=\"\",name=~\"^k8s_.*\",datacenter=\"$Site\",kubernetes_io_hostname=~\"^$Node$\",namespace=~\"$Namespace\"}[5m])) by (pod_name)",
+ "format": "time_series",
+ "interval": "10s",
+ "intervalFactor": 5,
+ "legendFormat": "{{ pod_name }}",
+ "metric": "container_cpu",
+ "refId": "A",
+ "step": 10
+ }
+ ],
+ "thresholds": [],
+ "timeFrom": null,
+ "timeRegions": [],
+ "timeShift": null,
+ "title": "Pods CPU usage (5m avg)",
+ "tooltip": {
+ "msResolution": true,
+ "shared": false,
+ "sort": 2,
+ "value_type": "cumulative"
+ },
+ "type": "graph",
+ "xaxis": {
+ "buckets": null,
+ "mode": "time",
+ "name": null,
+ "show": true,
+ "values": []
+ },
+ "yaxes": [
+ {
+ "format": "none",
+ "label": "cores",
+ "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
+ }
+ },
+ {
+ "collapsed": true,
+ "datasource": null,
+ "gridPos": {
+ "h": 1,
+ "w": 24,
+ "x": 0,
+ "y": 25
+ },
+ "id": 37,
+ "panels": [
+ {
+ "aliasColors": {},
+ "bars": false,
+ "dashLength": 10,
+ "dashes": false,
+ "datasource": "Prometheus",
+ "decimals": 3,
+ "editable": true,
+ "error": false,
+ "fill": 1,
+ "fillGradient": 0,
+ "grid": {},
+ "gridPos": {
+ "h": 7,
+ "w": 24,
+ "x": 0,
+ "y": 26
+ },
+ "height": "",
+ "id": 24,
+ "isNew": true,
+ "legend": {
+ "alignAsTable": true,
+ "avg": true,
+ "current": true,
+ "hideEmpty": false,
+ "hideZero": false,
+ "max": false,
+ "min": false,
+ "rightSide": true,
+ "show": true,
+ "sideWidth": null,
+ "sort": "current",
+ "sortDesc": true,
+ "total": false,
+ "values": true
+ },
+ "lines": true,
+ "linewidth": 1,
+ "links": [],
+ "nullPointMode": "null",
+ "options": {
+ "dataLinks": []
+ },
+ "percentage": false,
+ "pointradius": 0.5,
+ "points": true,
+ "renderer": "flot",
+ "seriesOverrides": [],
+ "spaceLength": 10,
+ "stack": false,
+ "steppedLine": false,
+ "targets": [
+ {
+ "expr": "sum (rate (container_cpu_usage_seconds_total{image!=\"\",name=~\"^k8s_.*\",container_name!=\"POD\",datacenter=\"$Site\",kubernetes_io_hostname=~\"^$Node$\",namespace=~\"$Namespace\"}[5m])) by (container_name, pod_name)",
+ "format": "time_series",
+ "hide": false,
+ "interval": "10s",
+ "intervalFactor": 5,
+ "legendFormat": "pod: {{ pod_name }} | {{ container_name }}",
+ "metric": "container_cpu",
+ "refId": "A",
+ "step": 10
+ },
+ {
+ "expr": "sum (rate (container_cpu_usage_seconds_total{image!=\"\",name!~\"^k8s_.*\",datacenter=\"$Site\",kubernetes_io_hostname=~\"^$Node$\",namespace=~\"$Namespace\"}[5m])) by (kubernetes_io_hostname, name, image)",
+ "format": "time_series",
+ "hide": false,
+ "interval": "10s",
+ "intervalFactor": 5,
+ "legendFormat": "docker: {{ kubernetes_io_hostname }} | {{ image }} ({{ name }})",
+ "metric": "container_cpu",
+ "refId": "B",
+ "step": 10
+ },
+ {
+ "expr": "sum (rate (container_cpu_usage_seconds_total{rkt_container_name!=\"\",datacenter=\"$Site\",kubernetes_io_hostname=~\"^$Node$\",namespace=~\"$Namespace\"}[5m])) by (kubernetes_io_hostname, rkt_container_name)",
+ "format": "time_series",
+ "interval": "10s",
+ "intervalFactor": 5,
+ "legendFormat": "rkt: {{ kubernetes_io_hostname }} | {{ rkt_container_name }}",
+ "metric": "container_cpu",
+ "refId": "C",
+ "step": 10
+ }
+ ],
+ "thresholds": [],
+ "timeFrom": null,
+ "timeRegions": [],
+ "timeShift": null,
+ "title": "Containers CPU usage (5m avg)",
+ "tooltip": {
+ "msResolution": true,
+ "shared": false,
+ "sort": 2,
+ "value_type": "cumulative"
+ },
+ "type": "graph",
+ "xaxis": {
+ "buckets": null,
+ "mode": "time",
+ "name": null,
+ "show": true,
+ "values": []
+ },
+ "yaxes": [
+ {
+ "format": "none",
+ "label": "cores",
+ "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": "Containers CPU usage",
+ "type": "row"
+ },
+ {
+ "collapsed": true,
+ "datasource": null,
+ "gridPos": {
+ "h": 1,
+ "w": 24,
+ "x": 0,
+ "y": 26
+ },
+ "id": 38,
+ "panels": [
+ {
+ "aliasColors": {},
+ "bars": false,
+ "dashLength": 10,
+ "dashes": false,
+ "datasource": "Prometheus",
+ "decimals": 3,
+ "editable": true,
+ "error": false,
+ "fill": 0,
+ "fillGradient": 0,
+ "grid": {},
+ "gridPos": {
+ "h": 13,
+ "w": 24,
+ "x": 0,
+ "y": 27
+ },
+ "id": 20,
+ "isNew": true,
+ "legend": {
+ "alignAsTable": true,
+ "avg": true,
+ "current": true,
+ "max": false,
+ "min": false,
+ "rightSide": false,
+ "show": true,
+ "sort": "current",
+ "sortDesc": true,
+ "total": false,
+ "values": true
+ },
+ "lines": true,
+ "linewidth": 1,
+ "links": [],
+ "nullPointMode": "null",
+ "options": {
+ "dataLinks": []
+ },
+ "percentage": false,
+ "pointradius": 0.5,
+ "points": true,
+ "renderer": "flot",
+ "seriesOverrides": [],
+ "spaceLength": 10,
+ "stack": false,
+ "steppedLine": false,
+ "targets": [
+ {
+ "expr": "sum (rate (container_cpu_usage_seconds_total{id!=\"/\",datacenter=\"$Site\",kubernetes_io_hostname=~\"^$Node$\",namespace=~\"$Namespace\"}[5m])) by (id)",
+ "format": "time_series",
+ "hide": false,
+ "interval": "10s",
+ "intervalFactor": 4,
+ "legendFormat": "{{ id }}",
+ "metric": "container_cpu",
+ "refId": "A",
+ "step": 10
+ }
+ ],
+ "thresholds": [],
+ "timeFrom": null,
+ "timeRegions": [],
+ "timeShift": null,
+ "title": "All processes CPU usage (5m avg)",
+ "tooltip": {
+ "msResolution": true,
+ "shared": false,
+ "sort": 2,
+ "value_type": "cumulative"
+ },
+ "type": "graph",
+ "xaxis": {
+ "buckets": null,
+ "mode": "time",
+ "name": null,
+ "show": true,
+ "values": []
+ },
+ "yaxes": [
+ {
+ "format": "none",
+ "label": "cores",
+ "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": "All processes CPU usage",
+ "type": "row"
+ },
+ {
+ "collapsed": false,
+ "datasource": null,
+ "gridPos": {
+ "h": 1,
+ "w": 24,
+ "x": 0,
+ "y": 27
+ },
+ "id": 39,
+ "panels": [],
+ "title": "Pods memory usage",
+ "type": "row"
+ },
+ {
+ "aliasColors": {},
+ "bars": false,
+ "dashLength": 10,
+ "dashes": false,
+ "datasource": "Prometheus",
+ "decimals": 2,
+ "editable": true,
+ "error": false,
+ "fill": 1,
+ "fillGradient": 0,
+ "grid": {},
+ "gridPos": {
+ "h": 7,
+ "w": 24,
+ "x": 0,
+ "y": 28
+ },
+ "id": 25,
+ "isNew": true,
+ "legend": {
+ "alignAsTable": true,
+ "avg": true,
+ "current": true,
+ "max": false,
+ "min": false,
+ "rightSide": true,
+ "show": true,
+ "sideWidth": null,
+ "sort": "current",
+ "sortDesc": true,
+ "total": false,
+ "values": true
+ },
+ "lines": true,
+ "linewidth": 0,
+ "links": [],
+ "nullPointMode": "null",
+ "options": {
+ "dataLinks": []
+ },
+ "percentage": false,
+ "pointradius": 1,
+ "points": true,
+ "renderer": "flot",
+ "seriesOverrides": [],
+ "spaceLength": 10,
+ "stack": false,
+ "steppedLine": true,
+ "targets": [
+ {
+ "expr": "sum (container_memory_working_set_bytes{image!=\"\",name=~\"^k8s_.*\",datacenter=\"$Site\",kubernetes_io_hostname=~\"^$Node$\",namespace=~\"$Namespace\"}) by (pod_name)",
+ "format": "time_series",
+ "interval": "10s",
+ "intervalFactor": 5,
+ "legendFormat": "{{ pod_name }}",
+ "metric": "container_memory_usage:sort_desc",
+ "refId": "A",
+ "step": 10
+ }
+ ],
+ "thresholds": [],
+ "timeFrom": null,
+ "timeRegions": [],
+ "timeShift": null,
+ "title": "Pods memory usage",
+ "tooltip": {
+ "msResolution": false,
+ "shared": false,
+ "sort": 2,
+ "value_type": "cumulative"
+ },
+ "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
+ }
+ },
+ {
+ "collapsed": true,
+ "datasource": null,
+ "gridPos": {
+ "h": 1,
+ "w": 24,
+ "x": 0,
+ "y": 35
+ },
+ "id": 41,
+ "panels": [
+ {
+ "aliasColors": {},
+ "bars": false,
+ "dashLength": 10,
+ "dashes": false,
+ "datasource": "Prometheus",
+ "decimals": 2,
+ "editable": true,
+ "error": false,
+ "fill": 1,
+ "fillGradient": 0,
+ "grid": {},
+ "gridPos": {
+ "h": 7,
+ "w": 24,
+ "x": 0,
+ "y": 36
+ },
+ "id": 27,
+ "isNew": true,
+ "legend": {
+ "alignAsTable": true,
+ "avg": true,
+ "current": true,
+ "max": false,
+ "min": false,
+ "rightSide": true,
+ "show": true,
+ "sideWidth": null,
+ "sort": "current",
+ "sortDesc": true,
+ "total": false,
+ "values": true
+ },
+ "lines": true,
+ "linewidth": 1,
+ "links": [],
+ "nullPointMode": "null",
+ "options": {
+ "dataLinks": []
+ },
+ "percentage": false,
+ "pointradius": 0.5,
+ "points": true,
+ "renderer": "flot",
+ "seriesOverrides": [],
+ "spaceLength": 10,
+ "stack": false,
+ "steppedLine": true,
+ "targets": [
+ {
+ "expr": "sum (container_memory_working_set_bytes{image!=\"\",name=~\"^k8s_.*\",container_name!=\"POD\",datacenter=\"$Site\",kubernetes_io_hostname=~\"^$Node$\",namespace=~\"$Namespace\"}) by (container_name, pod_name)",
+ "format": "time_series",
+ "interval": "10s",
+ "intervalFactor": 5,
+ "legendFormat": "pod: {{ pod_name }} | {{ container_name }}",
+ "metric": "container_memory_usage:sort_desc",
+ "refId": "A",
+ "step": 10
+ },
+ {
+ "expr": "sum (container_memory_working_set_bytes{image!=\"\",name!~\"^k8s_.*\",datacenter=\"$Site\",kubernetes_io_hostname=~\"^$Node$\",namespace=~\"$Namespace\"}) by (kubernetes_io_hostname, name, image)",
+ "format": "time_series",
+ "interval": "10s",
+ "intervalFactor": 5,
+ "legendFormat": "docker: {{ kubernetes_io_hostname }} | {{ image }} ({{ name }})",
+ "metric": "container_memory_usage:sort_desc",
+ "refId": "B",
+ "step": 10
+ },
+ {
+ "expr": "sum (container_memory_working_set_bytes{rkt_container_name!=\"\",datacenter=\"$Site\",kubernetes_io_hostname=~\"^$Node$\",namespace=~\"$Namespace\"}) by (kubernetes_io_hostname, rkt_container_name)",
+ "format": "time_series",
+ "interval": "10s",
+ "intervalFactor": 5,
+ "legendFormat": "rkt: {{ kubernetes_io_hostname }} | {{ rkt_container_name }}",
+ "metric": "container_memory_usage:sort_desc",
+ "refId": "C",
+ "step": 10
+ }
+ ],
+ "thresholds": [],
+ "timeFrom": null,
+ "timeRegions": [],
+ "timeShift": null,
+ "title": "Containers memory usage",
+ "tooltip": {
+ "msResolution": false,
+ "shared": false,
+ "sort": 2,
+ "value_type": "cumulative"
+ },
+ "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
+ }
+ }
+ ],
+ "title": "Containers memory usage",
+ "type": "row"
+ },
+ {
+ "collapsed": true,
+ "datasource": null,
+ "gridPos": {
+ "h": 1,
+ "w": 24,
+ "x": 0,
+ "y": 36
+ },
+ "id": 42,
+ "panels": [
+ {
+ "aliasColors": {},
+ "bars": false,
+ "dashLength": 10,
+ "dashes": false,
+ "datasource": "Prometheus",
+ "decimals": 2,
+ "editable": true,
+ "error": false,
+ "fill": 0,
+ "fillGradient": 0,
+ "grid": {},
+ "gridPos": {
+ "h": 13,
+ "w": 24,
+ "x": 0,
+ "y": 37
+ },
+ "id": 28,
+ "isNew": true,
+ "legend": {
+ "alignAsTable": true,
+ "avg": true,
+ "current": true,
+ "max": false,
+ "min": false,
+ "rightSide": false,
+ "show": true,
+ "sideWidth": 200,
+ "sort": "current",
+ "sortDesc": true,
+ "total": false,
+ "values": true
+ },
+ "lines": true,
+ "linewidth": 1,
+ "links": [],
+ "nullPointMode": "null",
+ "options": {
+ "dataLinks": []
+ },
+ "percentage": false,
+ "pointradius": 0.5,
+ "points": true,
+ "renderer": "flot",
+ "seriesOverrides": [],
+ "spaceLength": 10,
+ "stack": false,
+ "steppedLine": false,
+ "targets": [
+ {
+ "expr": "sum (container_memory_working_set_bytes{id!=\"/\",datacenter=\"$Site\",kubernetes_io_hostname=~\"^$Node$\",namespace=~\"$Namespace\"}) by (id)",
+ "format": "time_series",
+ "interval": "10s",
+ "intervalFactor": 4,
+ "legendFormat": "{{ id }}",
+ "metric": "container_memory_usage:sort_desc",
+ "refId": "A",
+ "step": 10
+ }
+ ],
+ "thresholds": [],
+ "timeFrom": null,
+ "timeRegions": [],
+ "timeShift": null,
+ "title": "All processes memory usage",
+ "tooltip": {
+ "msResolution": false,
+ "shared": false,
+ "sort": 2,
+ "value_type": "cumulative"
+ },
+ "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
+ }
+ }
+ ],
+ "title": "All processes memory usage",
+ "type": "row"
+ },
+ {
+ "collapsed": false,
+ "datasource": null,
+ "gridPos": {
+ "h": 1,
+ "w": 24,
+ "x": 0,
+ "y": 37
+ },
+ "id": 43,
+ "panels": [],
+ "title": "Pods network I/O",
+ "type": "row"
+ },
+ {
+ "aliasColors": {},
+ "bars": false,
+ "dashLength": 10,
+ "dashes": false,
+ "datasource": "Prometheus",
+ "decimals": 2,
+ "editable": true,
+ "error": false,
+ "fill": 1,
+ "fillGradient": 0,
+ "grid": {},
+ "gridPos": {
+ "h": 7,
+ "w": 24,
+ "x": 0,
+ "y": 38
+ },
+ "id": 16,
+ "isNew": true,
+ "legend": {
+ "alignAsTable": true,
+ "avg": true,
+ "current": true,
+ "max": false,
+ "min": false,
+ "rightSide": true,
+ "show": true,
+ "sideWidth": null,
+ "sort": "current",
+ "sortDesc": true,
+ "total": false,
+ "values": true
+ },
+ "lines": true,
+ "linewidth": 1,
+ "links": [],
+ "nullPointMode": "null as zero",
+ "options": {
+ "dataLinks": []
+ },
+ "percentage": false,
+ "pointradius": 1,
+ "points": false,
+ "renderer": "flot",
+ "seriesOverrides": [],
+ "spaceLength": 10,
+ "stack": false,
+ "steppedLine": true,
+ "targets": [
+ {
+ "expr": "sum (rate (container_network_receive_bytes_total{image!=\"\",name=~\"^k8s_.*\",datacenter=\"$Site\",kubernetes_io_hostname=~\"^$Node$\",namespace=~\"$Namespace\"}[5m])) by (pod_name)",
+ "format": "time_series",
+ "interval": "10s",
+ "intervalFactor": 5,
+ "legendFormat": "-> {{ pod_name }}",
+ "metric": "network",
+ "refId": "A",
+ "step": 10
+ },
+ {
+ "expr": "- sum (rate (container_network_transmit_bytes_total{image!=\"\",name=~\"^k8s_.*\",datacenter=\"$Site\",kubernetes_io_hostname=~\"^$Node$\",namespace=~\"$Namespace\"}[5m])) by (pod_name)",
+ "format": "time_series",
+ "interval": "10s",
+ "intervalFactor": 5,
+ "legendFormat": "<- {{ pod_name }}",
+ "metric": "network",
+ "refId": "B",
+ "step": 10
+ }
+ ],
+ "thresholds": [],
+ "timeFrom": null,
+ "timeRegions": [],
+ "timeShift": null,
+ "title": "Pods network I/O (5m avg)",
+ "tooltip": {
+ "msResolution": false,
+ "shared": false,
+ "sort": 2,
+ "value_type": "cumulative"
+ },
+ "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": false
+ }
+ ],
+ "yaxis": {
+ "align": false,
+ "alignLevel": null
+ }
+ },
+ {
+ "collapsed": true,
+ "datasource": null,
+ "gridPos": {
+ "h": 1,
+ "w": 24,
+ "x": 0,
+ "y": 45
+ },
+ "id": 44,
+ "panels": [
+ {
+ "aliasColors": {},
+ "bars": false,
+ "dashLength": 10,
+ "dashes": false,
+ "datasource": "Prometheus",
+ "decimals": 2,
+ "editable": true,
+ "error": false,
+ "fill": 1,
+ "fillGradient": 0,
+ "grid": {},
+ "gridPos": {
+ "h": 7,
+ "w": 24,
+ "x": 0,
+ "y": 46
+ },
+ "id": 30,
+ "isNew": true,
+ "legend": {
+ "alignAsTable": true,
+ "avg": true,
+ "current": true,
+ "max": false,
+ "min": false,
+ "rightSide": true,
+ "show": true,
+ "sideWidth": null,
+ "sort": "current",
+ "sortDesc": true,
+ "total": false,
+ "values": true
+ },
+ "lines": true,
+ "linewidth": 1,
+ "links": [],
+ "nullPointMode": "null as zero",
+ "options": {
+ "dataLinks": []
+ },
+ "percentage": false,
+ "pointradius": 5,
+ "points": false,
+ "renderer": "flot",
+ "seriesOverrides": [],
+ "spaceLength": 10,
+ "stack": false,
+ "steppedLine": true,
+ "targets": [
+ {
+ "expr": "sum (rate (container_network_receive_bytes_total{image!=\"\",name=~\"^k8s_.*\",datacenter=\"$Site\",kubernetes_io_hostname=~\"^$Node$\",namespace=~\"$Namespace\"}[5m])) by (container_name, pod_name)",
+ "format": "time_series",
+ "hide": false,
+ "interval": "10s",
+ "intervalFactor": 5,
+ "legendFormat": "-> pod: {{ pod_name }} | {{ container_name }}",
+ "metric": "network",
+ "refId": "B",
+ "step": 10
+ },
+ {
+ "expr": "- sum (rate (container_network_transmit_bytes_total{image!=\"\",name=~\"^k8s_.*\",datacenter=\"$Site\",kubernetes_io_hostname=~\"^$Node$\",namespace=~\"$Namespace\"}[5m])) by (container_name, pod_name)",
+ "format": "time_series",
+ "hide": false,
+ "interval": "10s",
+ "intervalFactor": 5,
+ "legendFormat": "<- pod: {{ pod_name }} | {{ container_name }}",
+ "metric": "network",
+ "refId": "D",
+ "step": 10
+ },
+ {
+ "expr": "sum (rate (container_network_receive_bytes_total{image!=\"\",name!~\"^k8s_.*\",datacenter=\"$Site\",kubernetes_io_hostname=~\"^$Node$\",namespace=~\"$Namespace\"}[5m])) by (kubernetes_io_hostname, name, image)",
+ "format": "time_series",
+ "hide": false,
+ "interval": "10s",
+ "intervalFactor": 5,
+ "legendFormat": "-> docker: {{ kubernetes_io_hostname }} | {{ image }} ({{ name }})",
+ "metric": "network",
+ "refId": "A",
+ "step": 10
+ },
+ {
+ "expr": "- sum (rate (container_network_transmit_bytes_total{image!=\"\",name!~\"^k8s_.*\",datacenter=\"$Site\",kubernetes_io_hostname=~\"^$Node$\",namespace=~\"$Namespace\"}[5m])) by (kubernetes_io_hostname, name, image)",
+ "format": "time_series",
+ "hide": false,
+ "interval": "10s",
+ "intervalFactor": 5,
+ "legendFormat": "<- docker: {{ kubernetes_io_hostname }} | {{ image }} ({{ name }})",
+ "metric": "network",
+ "refId": "C",
+ "step": 10
+ },
+ {
+ "expr": "sum (rate (container_network_transmit_bytes_total{rkt_container_name!=\"\",datacenter=\"$Site\",kubernetes_io_hostname=~\"^$Node$\",namespace=~\"$Namespace\"}[5m])) by (kubernetes_io_hostname, rkt_container_name)",
+ "format": "time_series",
+ "hide": false,
+ "interval": "10s",
+ "intervalFactor": 5,
+ "legendFormat": "-> rkt: {{ kubernetes_io_hostname }} | {{ rkt_container_name }}",
+ "metric": "network",
+ "refId": "E",
+ "step": 10
+ },
+ {
+ "expr": "- sum (rate (container_network_transmit_bytes_total{rkt_container_name!=\"\",datacenter=\"$Site\",kubernetes_io_hostname=~\"^$Node$\",namespace=~\"$Namespace\"}[5m])) by (kubernetes_io_hostname, rkt_container_name)",
+ "format": "time_series",
+ "hide": false,
+ "interval": "10s",
+ "intervalFactor": 5,
+ "legendFormat": "<- rkt: {{ kubernetes_io_hostname }} | {{ rkt_container_name }}",
+ "metric": "network",
+ "refId": "F",
+ "step": 10
+ }
+ ],
+ "thresholds": [],
+ "timeFrom": null,
+ "timeRegions": [],
+ "timeShift": null,
+ "title": "Containers network I/O (5m avg)",
+ "tooltip": {
+ "msResolution": false,
+ "shared": false,
+ "sort": 2,
+ "value_type": "cumulative"
+ },
+ "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": false
+ }
+ ],
+ "yaxis": {
+ "align": false,
+ "alignLevel": null
+ }
+ }
+ ],
+ "title": "Containers network I/O",
+ "type": "row"
+ }
+ ],
+ "refresh": "1m",
+ "schemaVersion": 20,
+ "style": "dark",
+ "tags": [
+ "aether"
+ ],
+ "templating": {
+ "list": [
+ {
+ "allValue": null,
+ "datasource": "Prometheus",
+ "definition": "label_values(kube_pod_info, datacenter)",
+ "hide": 0,
+ "includeAll": false,
+ "label": "Site",
+ "multi": false,
+ "name": "Site",
+ "query": "label_values(kube_pod_info, datacenter)",
+ "refresh": 2,
+ "regex": "",
+ "skipUrlSync": false,
+ "sort": 0,
+ "tagValuesQuery": "",
+ "tags": [],
+ "tagsQuery": "",
+ "type": "query",
+ "useTags": false
+ },
+ {
+ "allValue": ".*",
+ "current": {
+ "text": "All",
+ "value": "$__all"
+ },
+ "datasource": "Prometheus",
+ "definition": "label_values(node_load1{datacenter=\"$Site\"}, kubernetes_node)",
+ "hide": 0,
+ "includeAll": true,
+ "label": null,
+ "multi": false,
+ "name": "Node",
+ "options": [],
+ "query": "label_values(node_load1{datacenter=\"$Site\"}, kubernetes_node)",
+ "refresh": 2,
+ "regex": "",
+ "skipUrlSync": false,
+ "sort": 0,
+ "tagValuesQuery": "",
+ "tags": [],
+ "tagsQuery": "",
+ "type": "query",
+ "useTags": false
+ },
+ {
+ "allValue": ".*",
+ "current": {
+ "text": "All",
+ "value": "$__all"
+ },
+ "datasource": "Prometheus",
+ "definition": "label_values(kube_pod_info{datacenter=\"$Site\"}, namespace)",
+ "hide": 0,
+ "includeAll": true,
+ "label": null,
+ "multi": true,
+ "name": "Namespace",
+ "options": [],
+ "query": "label_values(kube_pod_info{datacenter=\"$Site\"}, namespace)",
+ "refresh": 2,
+ "regex": "",
+ "skipUrlSync": false,
+ "sort": 0,
+ "tagValuesQuery": "",
+ "tags": [],
+ "tagsQuery": "",
+ "type": "query",
+ "useTags": false
+ }
+ ]
+ },
+ "time": {
+ "from": "now-3h",
+ "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": "Multi-Site K8S Cluster Monitoring",
+ "uid": "1bI4P-xiz",
+ "version": 1
+}