{
- "__inputs": [],
- "__requires": [
- {
- "type": "grafana",
- "id": "grafana",
- "name": "Grafana",
- "version": "5.3.2"
- },
- {
- "type": "panel",
- "id": "graph",
- "name": "Graph",
- "version": "5.0.0"
- },
- {
- "type": "panel",
- "id": "singlestat",
- "name": "Singlestat",
- "version": "5.0.0"
- }
- ],
- "annotations": {
- "list": [
+ "__inputs": [ ],
+ "__requires": [
{
- "builtIn": 1,
- "datasource": "-- Grafana --",
- "enable": true,
- "hide": true,
- "iconColor": "rgba(0, 211, 255, 1)",
- "name": "Annotations & Alerts",
- "type": "dashboard"
- }
- ]
- },
- "editable": false,
- "gnetId": null,
- "graphTooltip": 0,
- "id": null,
- "iteration": 1615564911000,
- "links": [],
- "panels": [
- {
- "gridPos": {
- "h": 1,
- "w": 24,
- "x": 0,
- "y": 0
- },
- "id": 16,
- "title": "$ceph_hosts System Overview",
- "type": "row"
- },
- {
- "cacheTimeout": null,
- "colorBackground": false,
- "colorValue": false,
- "colors": [
- "rgba(245, 54, 54, 0.9)",
- "rgba(237, 129, 40, 0.89)",
- "rgba(50, 172, 45, 0.97)"
- ],
- "datasource": "$datasource",
- "format": "none",
- "gauge": {
- "maxValue": 100,
- "minValue": 0,
- "show": false,
- "thresholdLabels": false,
- "thresholdMarkers": true
- },
- "gridPos": {
- "h": 5,
- "w": 3,
- "x": 0,
- "y": 1
- },
- "height": "160",
- "id": 1,
- "interval": null,
- "links": [],
- "mappingType": 1,
- "mappingTypes": [
- {
- "name": "value to text",
- "value": 1
- },
- {
- "name": "range to text",
- "value": 2
- }
- ],
- "maxDataPoints": "",
- "minSpan": 4,
- "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": [
- {
- "expr": "count(sum by (ceph_daemon) (ceph_osd_metadata{hostname='$ceph_hosts'}))",
- "format": "time_series",
- "intervalFactor": 2,
- "refId": "A",
- "step": 40,
- "textEditor": true
- }
- ],
- "thresholds": "",
- "title": "OSDs",
- "type": "singlestat",
- "valueFontSize": "80%",
- "valueMaps": [
- {
- "op": "=",
- "text": "N/A",
- "value": "null"
- }
- ],
- "valueName": "current"
- },
- {
- "aliasColors": {
- "interrupt": "#447EBC",
- "steal": "#6D1F62",
- "system": "#890F02",
- "user": "#3F6833",
- "wait": "#C15C17"
- },
- "bars": false,
- "dashLength": 10,
- "dashes": false,
- "datasource": "$datasource",
- "description": "Shows the CPU breakdown. When multiple servers are selected, only the first host's cpu data is shown",
- "fill": 1,
- "gridPos": {
- "h": 10,
- "w": 6,
- "x": 3,
- "y": 1
- },
- "id": 9,
- "legend": {
- "alignAsTable": false,
- "avg": false,
- "current": false,
- "max": false,
- "min": false,
- "show": true,
- "total": false,
- "values": false
- },
- "lines": true,
- "linewidth": 1,
- "links": [],
- "minSpan": 12,
- "nullPointMode": "null",
- "percentage": false,
- "pointradius": 5,
- "points": false,
- "renderer": "flot",
- "repeat": null,
- "seriesOverrides": [],
- "spaceLength": 10,
- "stack": true,
- "steppedLine": false,
- "targets": [
- {
- "expr": "sum by (mode) (\n irate(node_cpu{instance=~\"($ceph_hosts)([\\\\.:].*)?\", mode=~\"(irq|nice|softirq|steal|system|user|iowait)\"}[1m]) or\n irate(node_cpu_seconds_total{instance=~\"($ceph_hosts)([\\\\.:].*)?\", mode=~\"(irq|nice|softirq|steal|system|user|iowait)\"}[1m])\n) / scalar(\n sum(irate(node_cpu{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[1m]) or\n irate(node_cpu_seconds_total{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[1m]))\n) * 100",
- "format": "time_series",
- "intervalFactor": 2,
- "legendFormat": "{{mode}}",
- "refId": "A",
- "step": 10,
- "textEditor": true
- }
- ],
- "thresholds": [],
- "timeFrom": null,
- "timeShift": null,
- "title": "CPU Utilization",
- "tooltip": {
- "shared": true,
- "sort": 2,
- "value_type": "individual"
- },
- "type": "graph",
- "xaxis": {
- "buckets": null,
- "mode": "time",
- "name": null,
- "show": true,
- "values": []
- },
- "yaxes": [
- {
- "format": "percent",
- "label": "% Utilization",
- "logBase": 1,
- "max": "100",
- "min": null,
- "show": true
- },
- {
- "format": "short",
- "label": null,
- "logBase": 1,
- "max": null,
- "min": null,
- "show": false
- }
- ],
- "yaxis": {
- "align": false,
- "alignLevel": null
- }
- },
- {
- "aliasColors": {
- "Available": "#508642",
- "Free": "#508642",
- "Total": "#bf1b00",
- "Used": "#bf1b00",
- "total": "#bf1b00",
- "used": "#0a50a1"
- },
- "bars": false,
- "dashLength": 10,
- "dashes": false,
- "datasource": "$datasource",
- "fill": 1,
- "gridPos": {
- "h": 10,
- "w": 6,
- "x": 9,
- "y": 1
- },
- "id": 14,
- "legend": {
- "avg": false,
- "current": false,
- "max": false,
- "min": false,
- "show": true,
- "total": false,
- "values": false
- },
- "lines": true,
- "linewidth": 1,
- "links": [],
- "nullPointMode": "null",
- "percentage": false,
- "pointradius": 5,
- "points": false,
- "renderer": "flot",
- "seriesOverrides": [
- {
- "alias": "total",
- "color": "#bf1b00",
- "fill": 0,
- "linewidth": 2,
- "stack": false
- }
- ],
- "spaceLength": 10,
- "stack": true,
- "steppedLine": false,
- "targets": [
- {
- "expr": "(node_memory_MemTotal{instance=~\"$ceph_hosts([\\\\.:].*)?\"} or node_memory_MemTotal_bytes{instance=~\"$ceph_hosts([\\\\.:].*)?\"})- (\n (node_memory_MemFree{instance=~\"$ceph_hosts([\\\\.:].*)?\"} or node_memory_MemFree_bytes{instance=~\"$ceph_hosts([\\\\.:].*)?\"}) + \n (node_memory_Cached{instance=~\"$ceph_hosts([\\\\.:].*)?\"} or node_memory_Cached_bytes{instance=~\"$ceph_hosts([\\\\.:].*)?\"}) + \n (node_memory_Buffers{instance=~\"$ceph_hosts([\\\\.:].*)?\"} or node_memory_Buffers_bytes{instance=~\"$ceph_hosts([\\\\.:].*)?\"}) +\n (node_memory_Slab{instance=~\"$ceph_hosts([\\\\.:].*)?\"} or node_memory_Slab_bytes{instance=~\"$ceph_hosts([\\\\.:].*)?\"})\n )\n \n",
- "format": "time_series",
- "intervalFactor": 1,
- "legendFormat": "used",
- "refId": "D"
- },
- {
- "expr": "node_memory_MemFree{instance=~\"$ceph_hosts([\\\\.:].*)?\"} or node_memory_MemFree_bytes{instance=~\"$ceph_hosts([\\\\.:].*)?\"} ",
- "format": "time_series",
- "hide": false,
- "intervalFactor": 1,
- "legendFormat": "Free",
- "refId": "A"
- },
- {
- "expr": "(node_memory_Cached{instance=~\"$ceph_hosts([\\\\.:].*)?\"} or node_memory_Cached_bytes{instance=~\"$ceph_hosts([\\\\.:].*)?\"}) + \n(node_memory_Buffers{instance=~\"$ceph_hosts([\\\\.:].*)?\"} or node_memory_Buffers_bytes{instance=~\"$ceph_hosts([\\\\.:].*)?\"}) +\n(node_memory_Slab{instance=~\"$ceph_hosts([\\\\.:].*)?\"} or node_memory_Slab_bytes{instance=~\"$ceph_hosts([\\\\.:].*)?\"}) \n",
- "format": "time_series",
- "hide": false,
- "intervalFactor": 1,
- "legendFormat": "buffers/cache",
- "refId": "C"
- },
- {
- "expr": "node_memory_MemTotal{instance=~\"$ceph_hosts([\\\\.:].*)?\"} or node_memory_MemTotal_bytes{instance=~\"$ceph_hosts([\\\\.:].*)?\"} ",
- "format": "time_series",
- "hide": false,
- "intervalFactor": 1,
- "legendFormat": "total",
- "refId": "B"
- }
- ],
- "thresholds": [],
- "timeFrom": null,
- "timeShift": null,
- "title": "RAM Usage",
- "tooltip": {
- "shared": true,
- "sort": 2,
- "value_type": "individual"
+ "id": "grafana",
+ "name": "Grafana",
+ "type": "grafana",
+ "version": "5.3.2"
},
- "type": "graph",
- "xaxis": {
- "buckets": null,
- "mode": "time",
- "name": null,
- "show": true,
- "values": []
- },
- "yaxes": [
- {
- "format": "bytes",
- "label": "RAM used",
- "logBase": 1,
- "max": null,
- "min": "0",
- "show": true
- },
- {
- "format": "short",
- "label": null,
- "logBase": 1,
- "max": null,
- "min": null,
- "show": false
- }
- ],
- "yaxis": {
- "align": false,
- "alignLevel": null
- }
- },
- {
- "aliasColors": {},
- "bars": false,
- "dashLength": 10,
- "dashes": false,
- "datasource": "$datasource",
- "description": "Show the network load (rx,tx) across all interfaces (excluding loopback 'lo')",
- "fill": 0,
- "gridPos": {
- "h": 10,
- "w": 6,
- "x": 15,
- "y": 1
- },
- "id": 10,
- "legend": {
- "alignAsTable": false,
- "avg": false,
- "current": false,
- "hideZero": true,
- "max": false,
- "min": false,
- "rightSide": false,
- "show": true,
- "total": false,
- "values": false
- },
- "lines": true,
- "linewidth": 1,
- "links": [],
- "minSpan": 12,
- "nullPointMode": "null",
- "percentage": false,
- "pointradius": 5,
- "points": false,
- "renderer": "flot",
- "seriesOverrides": [
- {
- "alias": "/.*tx/",
- "transform": "negative-Y"
- }
- ],
- "spaceLength": 10,
- "stack": false,
- "steppedLine": false,
- "targets": [
- {
- "expr": "sum by (device) (\n irate(node_network_receive_bytes{instance=~\"($ceph_hosts)([\\\\.:].*)?\",device!=\"lo\"}[1m]) or \n irate(node_network_receive_bytes_total{instance=~\"($ceph_hosts)([\\\\.:].*)?\",device!=\"lo\"}[1m])\n)",
- "format": "time_series",
- "intervalFactor": 1,
- "legendFormat": "{{device}}.rx",
- "refId": "A",
- "step": 10,
- "textEditor": true
- },
- {
- "expr": "sum by (device) (\n irate(node_network_transmit_bytes{instance=~\"($ceph_hosts)([\\\\.:].*)?\",device!=\"lo\"}[1m]) or\n irate(node_network_transmit_bytes_total{instance=~\"($ceph_hosts)([\\\\.:].*)?\",device!=\"lo\"}[1m])\n)",
- "format": "time_series",
- "intervalFactor": 1,
- "legendFormat": "{{device}}.tx",
- "refId": "B",
- "step": 10
- }
- ],
- "thresholds": [],
- "timeFrom": null,
- "timeShift": null,
- "title": "Network Load",
- "tooltip": {
- "shared": true,
- "sort": 2,
- "value_type": "individual"
- },
- "type": "graph",
- "xaxis": {
- "buckets": null,
- "mode": "time",
- "name": null,
- "show": true,
- "values": []
- },
- "yaxes": [
- {
- "format": "decbytes",
- "label": "Send (-) / Receive (+)",
- "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": "$datasource",
- "fill": 0,
- "gridPos": {
- "h": 5,
- "w": 3,
- "x": 21,
- "y": 1
- },
- "hideTimeOverride": true,
- "id": 18,
- "legend": {
- "avg": false,
- "current": false,
- "max": false,
- "min": false,
- "show": false,
- "total": false,
- "values": false
- },
- "lines": true,
- "linewidth": 1,
- "links": [],
- "nullPointMode": "null",
- "percentage": false,
- "pointradius": 5,
- "points": false,
- "renderer": "flot",
- "seriesOverrides": [
- {
- "alias": "/.*tx/",
- "transform": "negative-Y"
- }
- ],
- "spaceLength": 10,
- "stack": false,
- "steppedLine": false,
- "targets": [
- {
- "expr": "irate(node_network_receive_drop{instance=~\"$ceph_hosts([\\\\.:].*)?\"}[1m]) or irate(node_network_receive_drop_total{instance=~\"$ceph_hosts([\\\\.:].*)?\"}[1m])",
- "format": "time_series",
- "instant": false,
- "intervalFactor": 1,
- "legendFormat": "{{device}}.rx",
- "refId": "A"
- },
- {
- "expr": "irate(node_network_transmit_drop{instance=~\"$ceph_hosts([\\\\.:].*)?\"}[1m]) or irate(node_network_transmit_drop_total{instance=~\"$ceph_hosts([\\\\.:].*)?\"}[1m])",
- "format": "time_series",
- "intervalFactor": 1,
- "legendFormat": "{{device}}.tx",
- "refId": "B"
- }
- ],
- "thresholds": [],
- "timeFrom": null,
- "timeShift": null,
- "title": "Network drop rate",
- "tooltip": {
- "shared": true,
- "sort": 0,
- "value_type": "individual"
- },
- "type": "graph",
- "xaxis": {
- "buckets": null,
- "mode": "time",
- "name": null,
- "show": true,
- "values": []
- },
- "yaxes": [
- {
- "format": "pps",
- "label": "Send (-) / Receive (+)",
- "logBase": 1,
- "max": null,
- "min": "0",
- "show": true
- },
- {
- "format": "short",
- "label": null,
- "logBase": 1,
- "max": null,
- "min": null,
- "show": false
- }
- ],
- "yaxis": {
- "align": false,
- "alignLevel": null
- }
- },
- {
- "cacheTimeout": null,
- "colorBackground": false,
- "colorValue": false,
- "colors": [
- "rgba(245, 54, 54, 0.9)",
- "rgba(237, 129, 40, 0.89)",
- "rgba(50, 172, 45, 0.97)"
- ],
- "datasource": "$datasource",
- "decimals": 0,
- "description": "Each OSD consists of a Journal/WAL partition and a data partition. The RAW Capacity shown is the sum of the data partitions across all OSDs on the selected OSD hosts.",
- "format": "bytes",
- "gauge": {
- "maxValue": 100,
- "minValue": 0,
- "show": false,
- "thresholdLabels": false,
- "thresholdMarkers": true
- },
- "gridPos": {
- "h": 5,
- "w": 3,
- "x": 0,
- "y": 6
- },
- "height": "160",
- "id": 2,
- "interval": null,
- "links": [],
- "mappingType": 1,
- "mappingTypes": [
- {
- "name": "value to text",
- "value": 1
- },
- {
- "name": "range to text",
- "value": 2
- }
- ],
- "maxDataPoints": "",
- "minSpan": 4,
- "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": [
- {
- "expr": "sum(ceph_osd_stat_bytes and on (ceph_daemon) ceph_disk_occupation{instance=~\"($ceph_hosts)([\\\\.:].*)?\"})",
- "format": "time_series",
- "intervalFactor": 2,
- "refId": "A",
- "step": 40,
- "textEditor": true
- }
- ],
- "thresholds": "",
- "title": "Raw Capacity",
- "type": "singlestat",
- "valueFontSize": "80%",
- "valueMaps": [
- {
- "op": "=",
- "text": "N/A",
- "value": "null"
- }
- ],
- "valueName": "current"
- },
- {
- "aliasColors": {},
- "bars": false,
- "dashLength": 10,
- "dashes": false,
- "datasource": "$datasource",
- "fill": 0,
- "gridPos": {
- "h": 5,
- "w": 3,
- "x": 21,
- "y": 6
- },
- "hideTimeOverride": true,
- "id": 19,
- "legend": {
- "avg": false,
- "current": false,
- "max": false,
- "min": false,
- "show": false,
- "total": false,
- "values": false
- },
- "lines": true,
- "linewidth": 1,
- "links": [],
- "nullPointMode": "null",
- "percentage": false,
- "pointradius": 5,
- "points": false,
- "renderer": "flot",
- "seriesOverrides": [
- {
- "alias": "/.*tx/",
- "transform": "negative-Y"
- }
- ],
- "spaceLength": 10,
- "stack": false,
- "steppedLine": false,
- "targets": [
- {
- "expr": "irate(node_network_receive_errs{instance=~\"$ceph_hosts([\\\\.:].*)?\"}[1m]) or irate(node_network_receive_errs_total{instance=~\"$ceph_hosts([\\\\.:].*)?\"}[1m])",
- "format": "time_series",
- "instant": false,
- "intervalFactor": 1,
- "legendFormat": "{{device}}.rx",
- "refId": "A"
- },
- {
- "expr": "irate(node_network_transmit_errs{instance=~\"$ceph_hosts([\\\\.:].*)?\"}[1m]) or irate(node_network_transmit_errs_total{instance=~\"$ceph_hosts([\\\\.:].*)?\"}[1m])",
- "format": "time_series",
- "intervalFactor": 1,
- "legendFormat": "{{device}}.tx",
- "refId": "B"
- }
- ],
- "thresholds": [],
- "timeFrom": null,
- "timeShift": null,
- "title": "Network error rate",
- "tooltip": {
- "shared": true,
- "sort": 0,
- "value_type": "individual"
- },
- "type": "graph",
- "xaxis": {
- "buckets": null,
- "mode": "time",
- "name": null,
- "show": true,
- "values": []
- },
- "yaxes": [
- {
- "format": "pps",
- "label": "Send (-) / Receive (+)",
- "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
- }
- },
- {
- "collapsed": false,
- "gridPos": {
- "h": 1,
- "w": 24,
- "x": 0,
- "y": 11
- },
- "id": 12,
- "panels": [],
- "repeat": null,
- "title": "OSD Disk Performance Statistics",
- "type": "row"
- },
- {
- "aliasColors": {},
- "bars": false,
- "dashLength": 10,
- "dashes": false,
- "datasource": "$datasource",
- "description": "For any OSD devices on the host, this chart shows the iops per physical device. Each device is shown by it's name and corresponding OSD id value",
- "fill": 1,
- "gridPos": {
- "h": 9,
- "w": 11,
- "x": 0,
- "y": 12
- },
- "id": 6,
- "legend": {
- "avg": false,
- "current": false,
- "max": false,
- "min": false,
- "show": true,
- "total": false,
- "values": false
- },
- "lines": true,
- "linewidth": 1,
- "links": [],
- "minSpan": 12,
- "nullPointMode": "connected",
- "percentage": false,
- "pointradius": 5,
- "points": false,
- "renderer": "flot",
- "seriesOverrides": [
- {
- "alias": "/.*reads/",
- "transform": "negative-Y"
- }
- ],
- "spaceLength": 10,
- "stack": false,
- "steppedLine": false,
- "targets": [
- {
- "expr": "label_replace(\n (\n irate(node_disk_writes_completed{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m]) or\n irate(node_disk_writes_completed_total{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m])\n ),\n \"instance\",\n \"$1\",\n \"instance\",\n \"([^:.]*).*\"\n)\n* on(instance, device, ceph_daemon) group_left\n label_replace(\n label_replace(\n ceph_disk_occupation,\n \"device\",\n \"$1\",\n \"device\",\n \"/dev/(.*)\"\n ),\n \"instance\",\n \"$1\",\n \"instance\",\n \"([^:.]*).*\"\n )",
- "format": "time_series",
- "intervalFactor": 1,
- "legendFormat": "{{device}}({{ceph_daemon}}) writes",
- "refId": "A",
- "step": 10,
- "textEditor": true
- },
- {
- "expr": "label_replace(\n (irate(node_disk_reads_completed{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m]) or irate(node_disk_reads_completed_total{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m])),\n \"instance\",\n \"$1\",\n \"instance\",\n \"([^:.]*).*\"\n)\n* on(instance, device, ceph_daemon) group_left\n label_replace(\n label_replace(\n ceph_disk_occupation,\n \"device\",\n \"$1\",\n \"device\",\n \"/dev/(.*)\"\n ),\n \"instance\",\n \"$1\",\n \"instance\",\n \"([^:.]*).*\"\n )",
- "format": "time_series",
- "hide": false,
- "intervalFactor": 1,
- "legendFormat": "{{device}}({{ceph_daemon}}) reads",
- "refId": "B"
- }
- ],
- "thresholds": [],
- "timeFrom": null,
- "timeShift": null,
- "title": "$ceph_hosts Disk IOPS",
- "tooltip": {
- "shared": true,
- "sort": 2,
- "value_type": "individual"
- },
- "type": "graph",
- "xaxis": {
- "buckets": null,
- "mode": "time",
- "name": null,
- "show": true,
- "values": []
+ {
+ "id": "graph",
+ "name": "Graph",
+ "type": "panel",
+ "version": "5.0.0"
},
- "yaxes": [
- {
- "format": "ops",
- "label": "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
+ {
+ "id": "singlestat",
+ "name": "Singlestat",
+ "type": "panel",
+ "version": "5.0.0"
}
- },
- {
- "aliasColors": {},
- "bars": false,
- "dashLength": 10,
- "dashes": false,
- "datasource": "$datasource",
- "description": "For OSD hosts, this chart shows the disk bandwidth (read bytes/sec + write bytes/sec) of the physical OSD device. Each device is shown by device name, and corresponding OSD id",
- "fill": 1,
- "gridPos": {
- "h": 9,
- "w": 11,
- "x": 12,
- "y": 12
- },
- "id": 8,
- "legend": {
- "avg": false,
- "current": false,
- "max": false,
- "min": false,
- "show": true,
- "total": false,
- "values": false
- },
- "lines": true,
- "linewidth": 1,
- "links": [],
- "minSpan": 12,
- "nullPointMode": "connected",
- "percentage": false,
- "pointradius": 5,
- "points": false,
- "renderer": "flot",
- "seriesOverrides": [
- {
- "alias": "/.*read/",
- "transform": "negative-Y"
- }
- ],
- "spaceLength": 10,
- "stack": false,
- "steppedLine": false,
- "targets": [
- {
- "expr": "label_replace((irate(node_disk_bytes_written{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m]) or irate(node_disk_written_bytes_total{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m])), \"instance\", \"$1\", \"instance\", \"([^:.]*).*\") * on(instance, device, ceph_daemon) group_left label_replace(label_replace(ceph_disk_occupation, \"device\", \"$1\", \"device\", \"/dev/(.*)\"), \"instance\", \"$1\", \"instance\", \"([^:.]*).*\")",
- "format": "time_series",
- "intervalFactor": 1,
- "legendFormat": "{{device}}({{ceph_daemon}}) write",
- "refId": "B"
- },
- {
- "expr": "label_replace((irate(node_disk_bytes_read{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m]) or irate(node_disk_read_bytes_total{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m])), \"instance\", \"$1\", \"instance\", \"([^:.]*).*\") * on(instance, device, ceph_daemon) group_left label_replace(label_replace(ceph_disk_occupation, \"device\", \"$1\", \"device\", \"/dev/(.*)\"), \"instance\", \"$1\", \"instance\", \"([^:.]*).*\")",
- "format": "time_series",
- "intervalFactor": 1,
- "legendFormat": "{{device}}({{ceph_daemon}}) read",
- "refId": "C"
- }
- ],
- "thresholds": [],
- "timeFrom": null,
- "timeShift": null,
- "title": "$ceph_hosts Throughput by Disk",
- "tooltip": {
- "shared": true,
- "sort": 2,
- "value_type": "individual"
+ ],
+ "annotations": {
+ "list": [
+ {
+ "builtIn": 1,
+ "datasource": "-- Grafana --",
+ "enable": true,
+ "hide": true,
+ "iconColor": "rgba(0, 211, 255, 1)",
+ "name": "Annotations & Alerts",
+ "showIn": 0,
+ "tags": [ ],
+ "type": "dashboard"
+ }
+ ]
+ },
+ "description": "",
+ "editable": false,
+ "gnetId": null,
+ "graphTooltip": 0,
+ "hideControls": false,
+ "id": null,
+ "links": [ ],
+ "panels": [
+ {
+ "collapse": false,
+ "collapsed": false,
+ "gridPos": {
+ "h": 1,
+ "w": 24,
+ "x": 0,
+ "y": 0
+ },
+ "id": 2,
+ "panels": [ ],
+ "repeat": null,
+ "repeatIteration": null,
+ "repeatRowId": null,
+ "showTitle": true,
+ "title": "$ceph_hosts System Overview",
+ "titleSize": "h6",
+ "type": "row"
},
- "type": "graph",
- "xaxis": {
- "buckets": null,
- "mode": "time",
- "name": null,
- "show": true,
- "values": []
+ {
+ "cacheTimeout": null,
+ "colorBackground": false,
+ "colorValue": false,
+ "colors": [
+ "#299c46",
+ "rgba(237, 129, 40, 0.89)",
+ "#d44a3a"
+ ],
+ "datasource": "$datasource",
+ "format": "none",
+ "gauge": {
+ "maxValue": 100,
+ "minValue": 0,
+ "show": false,
+ "thresholdLabels": false,
+ "thresholdMarkers": true
+ },
+ "gridPos": {
+ "h": 5,
+ "w": 3,
+ "x": 0,
+ "y": 1
+ },
+ "id": 3,
+ "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": [
+ {
+ "expr": "count(sum by (ceph_daemon) (ceph_osd_metadata{hostname='$ceph_hosts'}))",
+ "format": "time_series",
+ "intervalFactor": 1,
+ "legendFormat": "",
+ "refId": "A"
+ }
+ ],
+ "thresholds": "",
+ "title": "OSDs",
+ "type": "singlestat",
+ "valueFontSize": "80%",
+ "valueMaps": [
+ {
+ "op": "=",
+ "text": "N/A",
+ "value": "null"
+ }
+ ],
+ "valueName": "current"
},
- "yaxes": [
- {
- "format": "Bps",
- "label": "Read (-) / Write (+)",
- "logBase": 1,
- "max": null,
- "min": null,
- "show": true
- },
- {
- "format": "short",
- "label": null,
- "logBase": 1,
- "max": null,
- "min": null,
- "show": false
- }
- ],
- "yaxis": {
- "align": false,
- "alignLevel": null
- }
- },
- {
- "aliasColors": {},
- "bars": false,
- "dashLength": 10,
- "dashes": false,
- "datasource": "$datasource",
- "description": "For OSD hosts, this chart shows the latency at the physical drive. Each drive is shown by device name, with it's corresponding OSD id",
- "fill": 1,
- "gridPos": {
- "h": 9,
- "w": 11,
- "x": 0,
- "y": 21
+ {
+ "aliasColors": {
+ "interrupt": "#447EBC",
+ "steal": "#6D1F62",
+ "system": "#890F02",
+ "user": "#3F6833",
+ "wait": "#C15C17"
+ },
+ "bars": false,
+ "dashLength": 10,
+ "dashes": false,
+ "datasource": "$datasource",
+ "description": "Shows the CPU breakdown. When multiple servers are selected, only the first host's cpu data is shown",
+ "fill": 1,
+ "gridPos": {
+ "h": 10,
+ "w": 6,
+ "x": 3,
+ "y": 1
+ },
+ "id": 4,
+ "legend": {
+ "alignAsTable": false,
+ "avg": false,
+ "current": false,
+ "max": false,
+ "min": false,
+ "rightSide": false,
+ "show": true,
+ "sideWidth": null,
+ "total": false,
+ "values": false
+ },
+ "lines": true,
+ "linewidth": 1,
+ "links": [ ],
+ "nullPointMode": "null",
+ "percentage": false,
+ "pointradius": 5,
+ "points": false,
+ "renderer": "flot",
+ "repeat": null,
+ "seriesOverrides": [ ],
+ "spaceLength": 10,
+ "stack": false,
+ "steppedLine": false,
+ "targets": [
+ {
+ "expr": "sum by (mode) (\n irate(node_cpu{instance=~\"($ceph_hosts)([\\\\.:].*)?\", mode=~\"(irq|nice|softirq|steal|system|user|iowait)\"}[1m]) or\n irate(node_cpu_seconds_total{instance=~\"($ceph_hosts)([\\\\.:].*)?\", mode=~\"(irq|nice|softirq|steal|system|user|iowait)\"}[1m])\n) / scalar(\n sum(irate(node_cpu{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[1m]) or\n irate(node_cpu_seconds_total{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[1m]))\n) * 100",
+ "format": "time_series",
+ "intervalFactor": 1,
+ "legendFormat": "{{mode}}",
+ "refId": "A"
+ }
+ ],
+ "thresholds": [ ],
+ "timeFrom": null,
+ "timeShift": null,
+ "title": "CPU Utilization",
+ "tooltip": {
+ "shared": true,
+ "sort": 0,
+ "value_type": "individual"
+ },
+ "type": "graph",
+ "xaxis": {
+ "buckets": null,
+ "mode": "time",
+ "name": null,
+ "show": true,
+ "values": [ ]
+ },
+ "yaxes": [
+ {
+ "format": "percent",
+ "label": "% Utilization",
+ "logBase": 1,
+ "max": null,
+ "min": null,
+ "show": true
+ },
+ {
+ "format": "short",
+ "label": null,
+ "logBase": 1,
+ "max": null,
+ "min": null,
+ "show": true
+ }
+ ]
},
- "id": 7,
- "legend": {
- "avg": false,
- "current": false,
- "max": false,
- "min": false,
- "show": true,
- "total": false,
- "values": false
+ {
+ "aliasColors": {
+ "Available": "#508642",
+ "Free": "#508642",
+ "Total": "#bf1b00",
+ "Used": "#bf1b00",
+ "total": "#bf1b00",
+ "used": "#0a50a1"
+ },
+ "bars": false,
+ "dashLength": 10,
+ "dashes": false,
+ "datasource": "$datasource",
+ "description": "",
+ "fill": 1,
+ "gridPos": {
+ "h": 10,
+ "w": 6,
+ "x": 9,
+ "y": 1
+ },
+ "id": 5,
+ "legend": {
+ "alignAsTable": false,
+ "avg": false,
+ "current": false,
+ "max": false,
+ "min": false,
+ "rightSide": false,
+ "show": true,
+ "sideWidth": null,
+ "total": false,
+ "values": false
+ },
+ "lines": true,
+ "linewidth": 1,
+ "links": [ ],
+ "nullPointMode": "null",
+ "percentage": false,
+ "pointradius": 5,
+ "points": false,
+ "renderer": "flot",
+ "repeat": null,
+ "seriesOverrides": [
+ {
+ "alias": "total",
+ "color": "#bf1b00",
+ "fill": 0,
+ "linewidth": 2,
+ "stack": false
+ }
+ ],
+ "spaceLength": 10,
+ "stack": false,
+ "steppedLine": false,
+ "targets": [
+ {
+ "expr": "node_memory_MemFree{instance=~\"$ceph_hosts([\\\\.:].*)?\"} or node_memory_MemFree_bytes{instance=~\"$ceph_hosts([\\\\.:].*)?\"} ",
+ "format": "time_series",
+ "intervalFactor": 1,
+ "legendFormat": "Free",
+ "refId": "A"
+ },
+ {
+ "expr": "node_memory_MemTotal{instance=~\"$ceph_hosts([\\\\.:].*)?\"} or node_memory_MemTotal_bytes{instance=~\"$ceph_hosts([\\\\.:].*)?\"} ",
+ "format": "time_series",
+ "intervalFactor": 1,
+ "legendFormat": "total",
+ "refId": "B"
+ },
+ {
+ "expr": "(node_memory_Cached{instance=~\"$ceph_hosts([\\\\.:].*)?\"} or node_memory_Cached_bytes{instance=~\"$ceph_hosts([\\\\.:].*)?\"}) + \n(node_memory_Buffers{instance=~\"$ceph_hosts([\\\\.:].*)?\"} or node_memory_Buffers_bytes{instance=~\"$ceph_hosts([\\\\.:].*)?\"}) +\n(node_memory_Slab{instance=~\"$ceph_hosts([\\\\.:].*)?\"} or node_memory_Slab_bytes{instance=~\"$ceph_hosts([\\\\.:].*)?\"}) \n",
+ "format": "time_series",
+ "intervalFactor": 1,
+ "legendFormat": "buffers/cache",
+ "refId": "C"
+ },
+ {
+ "expr": "(node_memory_MemTotal{instance=~\"$ceph_hosts([\\\\.:].*)?\"} or node_memory_MemTotal_bytes{instance=~\"$ceph_hosts([\\\\.:].*)?\"})- (\n (node_memory_MemFree{instance=~\"$ceph_hosts([\\\\.:].*)?\"} or node_memory_MemFree_bytes{instance=~\"$ceph_hosts([\\\\.:].*)?\"}) + \n (node_memory_Cached{instance=~\"$ceph_hosts([\\\\.:].*)?\"} or node_memory_Cached_bytes{instance=~\"$ceph_hosts([\\\\.:].*)?\"}) + \n (node_memory_Buffers{instance=~\"$ceph_hosts([\\\\.:].*)?\"} or node_memory_Buffers_bytes{instance=~\"$ceph_hosts([\\\\.:].*)?\"}) +\n (node_memory_Slab{instance=~\"$ceph_hosts([\\\\.:].*)?\"} or node_memory_Slab_bytes{instance=~\"$ceph_hosts([\\\\.:].*)?\"})\n )\n \n",
+ "format": "time_series",
+ "intervalFactor": 1,
+ "legendFormat": "used",
+ "refId": "D"
+ }
+ ],
+ "thresholds": [ ],
+ "timeFrom": null,
+ "timeShift": null,
+ "title": "RAM Usage",
+ "tooltip": {
+ "shared": true,
+ "sort": 0,
+ "value_type": "individual"
+ },
+ "type": "graph",
+ "xaxis": {
+ "buckets": null,
+ "mode": "time",
+ "name": null,
+ "show": true,
+ "values": [ ]
+ },
+ "yaxes": [
+ {
+ "format": "bytes",
+ "label": "RAM used",
+ "logBase": 1,
+ "max": null,
+ "min": null,
+ "show": true
+ },
+ {
+ "format": "short",
+ "label": null,
+ "logBase": 1,
+ "max": null,
+ "min": null,
+ "show": true
+ }
+ ]
},
- "lines": true,
- "linewidth": 1,
- "links": [],
- "minSpan": 12,
- "nullPointMode": "null as zero",
- "percentage": false,
- "pointradius": 5,
- "points": false,
- "renderer": "flot",
- "seriesOverrides": [],
- "spaceLength": 10,
- "stack": false,
- "steppedLine": false,
- "targets": [
- {
- "expr": "max by(instance,device) (label_replace((irate(node_disk_write_time_seconds_total{ instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m]) ) / clamp_min(irate(node_disk_writes_completed_total{ instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m]), 0.001) or (irate(node_disk_read_time_seconds_total{ instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m]) ) / clamp_min(irate(node_disk_reads_completed_total{ instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m]), 0.001), \"instance\", \"$1\", \"instance\", \"([^:.]*).*\")) * on(instance, device, ceph_daemon) group_left label_replace(label_replace(ceph_disk_occupation{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}, \"device\", \"$1\", \"device\", \"/dev/(.*)\"), \"instance\", \"$1\", \"instance\", \"([^:.]*).*\")",
- "format": "time_series",
- "hide": false,
- "intervalFactor": 1,
- "legendFormat": "{{device}}({{ceph_daemon}})",
- "refId": "D"
- }
- ],
- "thresholds": [],
- "timeFrom": null,
- "timeShift": null,
- "title": "$ceph_hosts Disk Latency",
- "tooltip": {
- "shared": true,
- "sort": 2,
- "value_type": "individual"
+ {
+ "aliasColors": { },
+ "bars": false,
+ "dashLength": 10,
+ "dashes": false,
+ "datasource": "$datasource",
+ "description": "Show the network load (rx,tx) across all interfaces (excluding loopback 'lo')",
+ "fill": 1,
+ "gridPos": {
+ "h": 10,
+ "w": 6,
+ "x": 15,
+ "y": 1
+ },
+ "id": 6,
+ "legend": {
+ "alignAsTable": false,
+ "avg": false,
+ "current": false,
+ "max": false,
+ "min": false,
+ "rightSide": false,
+ "show": true,
+ "sideWidth": null,
+ "total": false,
+ "values": false
+ },
+ "lines": true,
+ "linewidth": 1,
+ "links": [ ],
+ "nullPointMode": "null",
+ "percentage": false,
+ "pointradius": 5,
+ "points": false,
+ "renderer": "flot",
+ "repeat": null,
+ "seriesOverrides": [
+ {
+ "alias": "/.*tx/",
+ "transform": "negative-Y"
+ }
+ ],
+ "spaceLength": 10,
+ "stack": false,
+ "steppedLine": false,
+ "targets": [
+ {
+ "expr": "sum by (device) (\n irate(node_network_receive_bytes{instance=~\"($ceph_hosts)([\\\\.:].*)?\",device!=\"lo\"}[1m]) or \n irate(node_network_receive_bytes_total{instance=~\"($ceph_hosts)([\\\\.:].*)?\",device!=\"lo\"}[1m])\n)",
+ "format": "time_series",
+ "intervalFactor": 1,
+ "legendFormat": "{{device}}.rx",
+ "refId": "A"
+ },
+ {
+ "expr": "sum by (device) (\n irate(node_network_transmit_bytes{instance=~\"($ceph_hosts)([\\\\.:].*)?\",device!=\"lo\"}[1m]) or\n irate(node_network_transmit_bytes_total{instance=~\"($ceph_hosts)([\\\\.:].*)?\",device!=\"lo\"}[1m])\n)",
+ "format": "time_series",
+ "intervalFactor": 1,
+ "legendFormat": "{{device}}.tx",
+ "refId": "B"
+ }
+ ],
+ "thresholds": [ ],
+ "timeFrom": null,
+ "timeShift": null,
+ "title": "Network Load",
+ "tooltip": {
+ "shared": true,
+ "sort": 0,
+ "value_type": "individual"
+ },
+ "type": "graph",
+ "xaxis": {
+ "buckets": null,
+ "mode": "time",
+ "name": null,
+ "show": true,
+ "values": [ ]
+ },
+ "yaxes": [
+ {
+ "format": "decbytes",
+ "label": "Send (-) / Receive (+)",
+ "logBase": 1,
+ "max": null,
+ "min": null,
+ "show": true
+ },
+ {
+ "format": "short",
+ "label": null,
+ "logBase": 1,
+ "max": null,
+ "min": null,
+ "show": true
+ }
+ ]
},
- "type": "graph",
- "xaxis": {
- "buckets": null,
- "mode": "time",
- "name": null,
- "show": true,
- "values": []
+ {
+ "aliasColors": { },
+ "bars": false,
+ "dashLength": 10,
+ "dashes": false,
+ "datasource": "$datasource",
+ "description": "",
+ "fill": 1,
+ "gridPos": {
+ "h": 5,
+ "w": 3,
+ "x": 21,
+ "y": 1
+ },
+ "id": 7,
+ "legend": {
+ "alignAsTable": false,
+ "avg": false,
+ "current": false,
+ "max": false,
+ "min": false,
+ "rightSide": false,
+ "show": true,
+ "sideWidth": null,
+ "total": false,
+ "values": false
+ },
+ "lines": true,
+ "linewidth": 1,
+ "links": [ ],
+ "nullPointMode": "null",
+ "percentage": false,
+ "pointradius": 5,
+ "points": false,
+ "renderer": "flot",
+ "repeat": null,
+ "seriesOverrides": [
+ {
+ "alias": "/.*tx/",
+ "transform": "negative-Y"
+ }
+ ],
+ "spaceLength": 10,
+ "stack": false,
+ "steppedLine": false,
+ "targets": [
+ {
+ "expr": "irate(node_network_receive_drop{instance=~\"$ceph_hosts([\\\\.:].*)?\"}[1m]) or irate(node_network_receive_drop_total{instance=~\"$ceph_hosts([\\\\.:].*)?\"}[1m])",
+ "format": "time_series",
+ "intervalFactor": 1,
+ "legendFormat": "{{device}}.rx",
+ "refId": "A"
+ },
+ {
+ "expr": "irate(node_network_transmit_drop{instance=~\"$ceph_hosts([\\\\.:].*)?\"}[1m]) or irate(node_network_transmit_drop_total{instance=~\"$ceph_hosts([\\\\.:].*)?\"}[1m])",
+ "format": "time_series",
+ "intervalFactor": 1,
+ "legendFormat": "{{device}}.tx",
+ "refId": "B"
+ }
+ ],
+ "thresholds": [ ],
+ "timeFrom": null,
+ "timeShift": null,
+ "title": "Network drop rate",
+ "tooltip": {
+ "shared": true,
+ "sort": 0,
+ "value_type": "individual"
+ },
+ "type": "graph",
+ "xaxis": {
+ "buckets": null,
+ "mode": "time",
+ "name": null,
+ "show": true,
+ "values": [ ]
+ },
+ "yaxes": [
+ {
+ "format": "pps",
+ "label": "Send (-) / Receive (+)",
+ "logBase": 1,
+ "max": null,
+ "min": null,
+ "show": true
+ },
+ {
+ "format": "short",
+ "label": null,
+ "logBase": 1,
+ "max": null,
+ "min": null,
+ "show": true
+ }
+ ]
},
- "yaxes": [
- {
- "format": "s",
- "label": "",
- "logBase": 1,
- "max": null,
- "min": "0",
- "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": "$datasource",
- "description": "Show disk utilization % (util) of any OSD devices on the host by the physical device name and associated OSD id.",
- "fill": 1,
- "fillGradient": 0,
- "gridPos": {
- "h": 9,
- "w": 11,
- "x": 12,
- "y": 21
+ {
+ "cacheTimeout": null,
+ "colorBackground": false,
+ "colorValue": false,
+ "colors": [
+ "#299c46",
+ "rgba(237, 129, 40, 0.89)",
+ "#d44a3a"
+ ],
+ "datasource": "$datasource",
+ "description": "Each OSD consists of a Journal/WAL partition and a data partition. The RAW Capacity shown is the sum of the data partitions across all OSDs on the selected OSD hosts.",
+ "format": "bytes",
+ "gauge": {
+ "maxValue": 100,
+ "minValue": 0,
+ "show": false,
+ "thresholdLabels": false,
+ "thresholdMarkers": true
+ },
+ "gridPos": {
+ "h": 5,
+ "w": 3,
+ "x": 0,
+ "y": 6
+ },
+ "id": 8,
+ "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": [
+ {
+ "expr": "sum(ceph_osd_stat_bytes and on (ceph_daemon) ceph_disk_occupation{instance=~\"($ceph_hosts)([\\\\.:].*)?\"})",
+ "format": "time_series",
+ "intervalFactor": 1,
+ "legendFormat": "",
+ "refId": "A"
+ }
+ ],
+ "thresholds": "",
+ "title": "Raw Capacity",
+ "type": "singlestat",
+ "valueFontSize": "80%",
+ "valueMaps": [
+ {
+ "op": "=",
+ "text": "N/A",
+ "value": "null"
+ }
+ ],
+ "valueName": "current"
},
- "id": 5,
- "legend": {
- "avg": false,
- "current": false,
- "max": false,
- "min": false,
- "show": true,
- "total": false,
- "values": false
+ {
+ "aliasColors": { },
+ "bars": false,
+ "dashLength": 10,
+ "dashes": false,
+ "datasource": "$datasource",
+ "description": "",
+ "fill": 1,
+ "gridPos": {
+ "h": 5,
+ "w": 3,
+ "x": 21,
+ "y": 6
+ },
+ "id": 9,
+ "legend": {
+ "alignAsTable": false,
+ "avg": false,
+ "current": false,
+ "max": false,
+ "min": false,
+ "rightSide": false,
+ "show": true,
+ "sideWidth": null,
+ "total": false,
+ "values": false
+ },
+ "lines": true,
+ "linewidth": 1,
+ "links": [ ],
+ "nullPointMode": "null",
+ "percentage": false,
+ "pointradius": 5,
+ "points": false,
+ "renderer": "flot",
+ "repeat": null,
+ "seriesOverrides": [
+ {
+ "alias": "/.*tx/",
+ "transform": "negative-Y"
+ }
+ ],
+ "spaceLength": 10,
+ "stack": false,
+ "steppedLine": false,
+ "targets": [
+ {
+ "expr": "irate(node_network_receive_errs{instance=~\"$ceph_hosts([\\\\.:].*)?\"}[1m]) or irate(node_network_receive_errs_total{instance=~\"$ceph_hosts([\\\\.:].*)?\"}[1m])",
+ "format": "time_series",
+ "intervalFactor": 1,
+ "legendFormat": "{{device}}.rx",
+ "refId": "A"
+ },
+ {
+ "expr": "irate(node_network_transmit_errs{instance=~\"$ceph_hosts([\\\\.:].*)?\"}[1m]) or irate(node_network_transmit_errs_total{instance=~\"$ceph_hosts([\\\\.:].*)?\"}[1m])",
+ "format": "time_series",
+ "intervalFactor": 1,
+ "legendFormat": "{{device}}.tx",
+ "refId": "B"
+ }
+ ],
+ "thresholds": [ ],
+ "timeFrom": null,
+ "timeShift": null,
+ "title": "Network error rate",
+ "tooltip": {
+ "shared": true,
+ "sort": 0,
+ "value_type": "individual"
+ },
+ "type": "graph",
+ "xaxis": {
+ "buckets": null,
+ "mode": "time",
+ "name": null,
+ "show": true,
+ "values": [ ]
+ },
+ "yaxes": [
+ {
+ "format": "pps",
+ "label": "Send (-) / Receive (+)",
+ "logBase": 1,
+ "max": null,
+ "min": null,
+ "show": true
+ },
+ {
+ "format": "short",
+ "label": null,
+ "logBase": 1,
+ "max": null,
+ "min": null,
+ "show": true
+ }
+ ]
},
- "lines": true,
- "linewidth": 1,
- "links": [],
- "maxPerRow": 2,
- "nullPointMode": "connected",
- "options": {
- "dataLinks": []
+ {
+ "collapse": false,
+ "collapsed": false,
+ "gridPos": {
+ "h": 1,
+ "w": 24,
+ "x": 0,
+ "y": 11
+ },
+ "id": 10,
+ "panels": [ ],
+ "repeat": null,
+ "repeatIteration": null,
+ "repeatRowId": null,
+ "showTitle": true,
+ "title": "OSD Disk Performance Statistics",
+ "titleSize": "h6",
+ "type": "row"
},
- "percentage": false,
- "pointradius": 5,
- "points": false,
- "renderer": "flot",
- "seriesOverrides": [],
- "spaceLength": 10,
- "stack": false,
- "steppedLine": false,
- "targets": [
- {
- "expr": "label_replace(((irate(node_disk_io_time_ms{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m]) / 10 ) or irate(node_disk_io_time_seconds_total{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m]) * 100), \"instance\", \"$1\", \"instance\", \"([^:.]*).*\") * on(instance, device, ceph_daemon) group_left label_replace(label_replace(ceph_disk_occupation{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}, \"device\", \"$1\", \"device\", \"/dev/(.*)\"), \"instance\", \"$1\", \"instance\", \"([^:.]*).*\")",
- "format": "time_series",
- "hide": false,
- "intervalFactor": 1,
- "legendFormat": "{{device}}({{ceph_daemon}})",
- "refId": "A"
- }
- ],
- "thresholds": [],
- "timeFrom": null,
- "timeRegions": [],
- "timeShift": null,
- "title": "$ceph_hosts Disk utilization",
- "tooltip": {
- "shared": true,
- "sort": 2,
- "value_type": "individual"
+ {
+ "aliasColors": { },
+ "bars": false,
+ "dashLength": 10,
+ "dashes": false,
+ "datasource": "$datasource",
+ "description": "For any OSD devices on the host, this chart shows the iops per physical device. Each device is shown by it's name and corresponding OSD id value",
+ "fill": 1,
+ "gridPos": {
+ "h": 9,
+ "w": 11,
+ "x": 0,
+ "y": 12
+ },
+ "id": 11,
+ "legend": {
+ "alignAsTable": false,
+ "avg": false,
+ "current": false,
+ "max": false,
+ "min": false,
+ "rightSide": false,
+ "show": true,
+ "sideWidth": null,
+ "total": false,
+ "values": false
+ },
+ "lines": true,
+ "linewidth": 1,
+ "links": [ ],
+ "nullPointMode": "connected",
+ "percentage": false,
+ "pointradius": 5,
+ "points": false,
+ "renderer": "flot",
+ "repeat": null,
+ "seriesOverrides": [
+ {
+ "alias": "/.*reads/",
+ "transform": "negative-Y"
+ }
+ ],
+ "spaceLength": 10,
+ "stack": false,
+ "steppedLine": false,
+ "targets": [
+ {
+ "expr": "label_replace(\n (\n irate(node_disk_writes_completed{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m]) or\n irate(node_disk_writes_completed_total{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m])\n ),\n \"instance\",\n \"$1\",\n \"instance\",\n \"([^:.]*).*\"\n)\n* on(instance, device, ceph_daemon) group_left\n label_replace(\n label_replace(\n ceph_disk_occupation,\n \"device\",\n \"$1\",\n \"device\",\n \"/dev/(.*)\"\n ),\n \"instance\",\n \"$1\",\n \"instance\",\n \"([^:.]*).*\"\n )",
+ "format": "time_series",
+ "intervalFactor": 1,
+ "legendFormat": "{{device}}({{ceph_daemon}}) writes",
+ "refId": "A"
+ },
+ {
+ "expr": "label_replace(\n (irate(node_disk_reads_completed{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m]) or irate(node_disk_reads_completed_total{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m])),\n \"instance\",\n \"$1\",\n \"instance\",\n \"([^:.]*).*\"\n)\n* on(instance, device, ceph_daemon) group_left\n label_replace(\n label_replace(\n ceph_disk_occupation,\n \"device\",\n \"$1\",\n \"device\",\n \"/dev/(.*)\"\n ),\n \"instance\",\n \"$1\",\n \"instance\",\n \"([^:.]*).*\"\n )",
+ "format": "time_series",
+ "intervalFactor": 1,
+ "legendFormat": "{{device}}({{ceph_daemon}}) reads",
+ "refId": "B"
+ }
+ ],
+ "thresholds": [ ],
+ "timeFrom": null,
+ "timeShift": null,
+ "title": "$ceph_hosts Disk IOPS",
+ "tooltip": {
+ "shared": true,
+ "sort": 0,
+ "value_type": "individual"
+ },
+ "type": "graph",
+ "xaxis": {
+ "buckets": null,
+ "mode": "time",
+ "name": null,
+ "show": true,
+ "values": [ ]
+ },
+ "yaxes": [
+ {
+ "format": "ops",
+ "label": "Read (-) / Write (+)",
+ "logBase": 1,
+ "max": null,
+ "min": null,
+ "show": true
+ },
+ {
+ "format": "short",
+ "label": null,
+ "logBase": 1,
+ "max": null,
+ "min": null,
+ "show": true
+ }
+ ]
},
- "type": "graph",
- "xaxis": {
- "buckets": null,
- "mode": "time",
- "name": null,
- "show": true,
- "values": []
+ {
+ "aliasColors": { },
+ "bars": false,
+ "dashLength": 10,
+ "dashes": false,
+ "datasource": "$datasource",
+ "description": "For OSD hosts, this chart shows the disk bandwidth (read bytes/sec + write bytes/sec) of the physical OSD device. Each device is shown by device name, and corresponding OSD id",
+ "fill": 1,
+ "gridPos": {
+ "h": 9,
+ "w": 11,
+ "x": 12,
+ "y": 12
+ },
+ "id": 12,
+ "legend": {
+ "alignAsTable": false,
+ "avg": false,
+ "current": false,
+ "max": false,
+ "min": false,
+ "rightSide": false,
+ "show": true,
+ "sideWidth": null,
+ "total": false,
+ "values": false
+ },
+ "lines": true,
+ "linewidth": 1,
+ "links": [ ],
+ "nullPointMode": "connected",
+ "percentage": false,
+ "pointradius": 5,
+ "points": false,
+ "renderer": "flot",
+ "repeat": null,
+ "seriesOverrides": [
+ {
+ "alias": "/.*read/",
+ "transform": "negative-Y"
+ }
+ ],
+ "spaceLength": 10,
+ "stack": false,
+ "steppedLine": false,
+ "targets": [
+ {
+ "expr": "label_replace((irate(node_disk_bytes_written{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m]) or irate(node_disk_written_bytes_total{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m])), \"instance\", \"$1\", \"instance\", \"([^:.]*).*\") * on(instance, device, ceph_daemon) group_left label_replace(label_replace(ceph_disk_occupation, \"device\", \"$1\", \"device\", \"/dev/(.*)\"), \"instance\", \"$1\", \"instance\", \"([^:.]*).*\")",
+ "format": "time_series",
+ "intervalFactor": 1,
+ "legendFormat": "{{device}}({{ceph_daemon}}) write",
+ "refId": "A"
+ },
+ {
+ "expr": "label_replace((irate(node_disk_bytes_read{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m]) or irate(node_disk_read_bytes_total{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m])), \"instance\", \"$1\", \"instance\", \"([^:.]*).*\") * on(instance, device, ceph_daemon) group_left label_replace(label_replace(ceph_disk_occupation, \"device\", \"$1\", \"device\", \"/dev/(.*)\"), \"instance\", \"$1\", \"instance\", \"([^:.]*).*\")",
+ "format": "time_series",
+ "intervalFactor": 1,
+ "legendFormat": "{{device}}({{ceph_daemon}}) read",
+ "refId": "B"
+ }
+ ],
+ "thresholds": [ ],
+ "timeFrom": null,
+ "timeShift": null,
+ "title": "$ceph_hosts Throughput by Disk",
+ "tooltip": {
+ "shared": true,
+ "sort": 0,
+ "value_type": "individual"
+ },
+ "type": "graph",
+ "xaxis": {
+ "buckets": null,
+ "mode": "time",
+ "name": null,
+ "show": true,
+ "values": [ ]
+ },
+ "yaxes": [
+ {
+ "format": "Bps",
+ "label": "Read (-) / Write (+)",
+ "logBase": 1,
+ "max": null,
+ "min": null,
+ "show": true
+ },
+ {
+ "format": "short",
+ "label": null,
+ "logBase": 1,
+ "max": null,
+ "min": null,
+ "show": true
+ }
+ ]
},
- "yaxes": [
- {
- "format": "percent",
- "label": "%Util",
- "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
- }
- }
- ],
- "refresh": "10s",
- "schemaVersion": 16,
- "style": "dark",
- "tags": [
- "overview"
- ],
- "templating": {
- "list": [
{
- "current": {
- "text": "default",
- "value": "default"
- },
- "hide": 0,
- "label": "Data Source",
- "name": "datasource",
- "options": [],
- "query": "prometheus",
- "refresh": 1,
- "regex": "",
- "skipUrlSync": false,
- "type": "datasource"
+ "aliasColors": { },
+ "bars": false,
+ "dashLength": 10,
+ "dashes": false,
+ "datasource": "$datasource",
+ "description": "For OSD hosts, this chart shows the latency at the physical drive. Each drive is shown by device name, with it's corresponding OSD id",
+ "fill": 1,
+ "gridPos": {
+ "h": 9,
+ "w": 11,
+ "x": 0,
+ "y": 21
+ },
+ "id": 13,
+ "legend": {
+ "alignAsTable": false,
+ "avg": false,
+ "current": false,
+ "max": false,
+ "min": false,
+ "rightSide": false,
+ "show": true,
+ "sideWidth": null,
+ "total": false,
+ "values": false
+ },
+ "lines": true,
+ "linewidth": 1,
+ "links": [ ],
+ "nullPointMode": "null as zero",
+ "percentage": false,
+ "pointradius": 5,
+ "points": false,
+ "renderer": "flot",
+ "repeat": null,
+ "seriesOverrides": [ ],
+ "spaceLength": 10,
+ "stack": false,
+ "steppedLine": false,
+ "targets": [
+ {
+ "expr": "max by(instance,device) (label_replace((irate(node_disk_write_time_seconds_total{ instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m]) ) / clamp_min(irate(node_disk_writes_completed_total{ instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m]), 0.001) or (irate(node_disk_read_time_seconds_total{ instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m]) ) / clamp_min(irate(node_disk_reads_completed_total{ instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m]), 0.001), \"instance\", \"$1\", \"instance\", \"([^:.]*).*\")) * on(instance, device, ceph_daemon) group_left label_replace(label_replace(ceph_disk_occupation{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}, \"device\", \"$1\", \"device\", \"/dev/(.*)\"), \"instance\", \"$1\", \"instance\", \"([^:.]*).*\")",
+ "format": "time_series",
+ "intervalFactor": 1,
+ "legendFormat": "{{device}}({{ceph_daemon}})",
+ "refId": "A"
+ }
+ ],
+ "thresholds": [ ],
+ "timeFrom": null,
+ "timeShift": null,
+ "title": "$ceph_hosts Disk Latency",
+ "tooltip": {
+ "shared": true,
+ "sort": 0,
+ "value_type": "individual"
+ },
+ "type": "graph",
+ "xaxis": {
+ "buckets": null,
+ "mode": "time",
+ "name": null,
+ "show": true,
+ "values": [ ]
+ },
+ "yaxes": [
+ {
+ "format": "s",
+ "label": "",
+ "logBase": 1,
+ "max": null,
+ "min": null,
+ "show": true
+ },
+ {
+ "format": "short",
+ "label": null,
+ "logBase": 1,
+ "max": null,
+ "min": null,
+ "show": true
+ }
+ ]
},
{
- "allValue": null,
- "current": {},
- "datasource": "$datasource",
- "hide": 0,
- "includeAll": false,
- "label": "Hostname",
- "multi": false,
- "name": "ceph_hosts",
- "options": [],
- "query": "label_values(node_scrape_collector_success, instance) ",
- "refresh": 1,
- "regex": "([^.:]*).*",
- "skipUrlSync": false,
- "sort": 3,
- "tagValuesQuery": "",
- "tags": [],
- "tagsQuery": "",
- "type": "query",
- "useTags": false
+ "aliasColors": { },
+ "bars": false,
+ "dashLength": 10,
+ "dashes": false,
+ "datasource": "$datasource",
+ "description": "Show disk utilization % (util) of any OSD devices on the host by the physical device name and associated OSD id.",
+ "fill": 1,
+ "gridPos": {
+ "h": 9,
+ "w": 11,
+ "x": 12,
+ "y": 21
+ },
+ "id": 14,
+ "legend": {
+ "alignAsTable": false,
+ "avg": false,
+ "current": false,
+ "max": false,
+ "min": false,
+ "rightSide": false,
+ "show": true,
+ "sideWidth": null,
+ "total": false,
+ "values": false
+ },
+ "lines": true,
+ "linewidth": 1,
+ "links": [ ],
+ "nullPointMode": "connected",
+ "percentage": false,
+ "pointradius": 5,
+ "points": false,
+ "renderer": "flot",
+ "repeat": null,
+ "seriesOverrides": [ ],
+ "spaceLength": 10,
+ "stack": false,
+ "steppedLine": false,
+ "targets": [
+ {
+ "expr": "label_replace(((irate(node_disk_io_time_ms{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m]) / 10 ) or irate(node_disk_io_time_seconds_total{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m]) * 100), \"instance\", \"$1\", \"instance\", \"([^:.]*).*\") * on(instance, device, ceph_daemon) group_left label_replace(label_replace(ceph_disk_occupation{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}, \"device\", \"$1\", \"device\", \"/dev/(.*)\"), \"instance\", \"$1\", \"instance\", \"([^:.]*).*\")",
+ "format": "time_series",
+ "intervalFactor": 1,
+ "legendFormat": "{{device}}({{ceph_daemon}})",
+ "refId": "A"
+ }
+ ],
+ "thresholds": [ ],
+ "timeFrom": null,
+ "timeShift": null,
+ "title": "$ceph_hosts Disk utilization",
+ "tooltip": {
+ "shared": true,
+ "sort": 0,
+ "value_type": "individual"
+ },
+ "type": "graph",
+ "xaxis": {
+ "buckets": null,
+ "mode": "time",
+ "name": null,
+ "show": true,
+ "values": [ ]
+ },
+ "yaxes": [
+ {
+ "format": "percent",
+ "label": "%Util",
+ "logBase": 1,
+ "max": null,
+ "min": null,
+ "show": true
+ },
+ {
+ "format": "short",
+ "label": null,
+ "logBase": 1,
+ "max": null,
+ "min": null,
+ "show": true
+ }
+ ]
}
- ]
- },
- "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": "",
- "title": "Host Details",
- "uid": "rtOg0AiWz",
- "version": 4
+ ],
+ "refresh": "10s",
+ "rows": [ ],
+ "schemaVersion": 16,
+ "style": "dark",
+ "tags": [
+ "overview"
+ ],
+ "templating": {
+ "list": [
+ {
+ "current": {
+ "text": "default",
+ "value": "default"
+ },
+ "hide": 0,
+ "label": "Data Source",
+ "name": "datasource",
+ "options": [ ],
+ "query": "prometheus",
+ "refresh": 1,
+ "regex": "",
+ "type": "datasource"
+ },
+ {
+ "allValue": null,
+ "current": { },
+ "datasource": "$datasource",
+ "hide": 0,
+ "includeAll": false,
+ "label": "Hostname",
+ "multi": false,
+ "name": "ceph_hosts",
+ "options": [ ],
+ "query": "label_values(node_scrape_collector_success, instance) ",
+ "refresh": 1,
+ "regex": "([^.:]*).*",
+ "sort": 3,
+ "tagValuesQuery": "",
+ "tags": [ ],
+ "tagsQuery": "",
+ "type": "query",
+ "useTags": false
+ }
+ ]
+ },
+ "time": {
+ "from": "now-1h",
+ "to": "now"
+ },
+ "timepicker": {
+ "refresh_intervals": [
+ "5s",
+ "10s",
+ "30s",
+ "1m",
+ "5m",
+ "15m",
+ "30m",
+ "1h",
+ "2h",
+ "1d"
+ ],
+ "time_options": [
+ "5m",
+ "15m",
+ "1h",
+ "6h",
+ "12h",
+ "24h",
+ "2d",
+ "7d",
+ "30d"
+ ]
+ },
+ "timezone": "",
+ "title": "Host Details",
+ "uid": "rtOg0AiWz",
+ "version": 0
}
{
- "__inputs": [],
- "__requires": [
- {
- "type": "grafana",
- "id": "grafana",
- "name": "Grafana",
- "version": "5.3.2"
- },
- {
- "type": "panel",
- "id": "graph",
- "name": "Graph",
- "version": "5.0.0"
- },
- {
- "type": "panel",
- "id": "singlestat",
- "name": "Singlestat",
- "version": "5.0.0"
- }
- ],
- "annotations": {
- "list": [
+ "__inputs": [ ],
+ "__requires": [
{
- "builtIn": 1,
- "datasource": "-- Grafana --",
- "enable": true,
- "hide": true,
- "iconColor": "rgba(0, 211, 255, 1)",
- "name": "Annotations & Alerts",
- "type": "dashboard"
- }
- ]
- },
- "editable": false,
- "gnetId": null,
- "graphTooltip": 0,
- "id": null,
- "iteration": 1557393917915,
- "links": [],
- "panels": [
- {
- "cacheTimeout": null,
- "colorBackground": false,
- "colorValue": false,
- "colors": [
- "#299c46",
- "rgba(237, 129, 40, 0.89)",
- "#d44a3a"
- ],
- "datasource": "$datasource",
- "format": "none",
- "gauge": {
- "maxValue": 100,
- "minValue": 0,
- "show": false,
- "thresholdLabels": false,
- "thresholdMarkers": true
- },
- "gridPos": {
- "h": 5,
- "w": 4,
- "x": 0,
- "y": 0
- },
- "id": 5,
- "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": [
- {
- "expr": "count(sum by (hostname) (ceph_osd_metadata))",
- "format": "time_series",
- "instant": true,
- "intervalFactor": 1,
- "refId": "A"
- }
- ],
- "thresholds": "",
- "title": "OSD Hosts",
- "type": "singlestat",
- "valueFontSize": "80%",
- "valueMaps": [
- {
- "op": "=",
- "text": "N/A",
- "value": "null"
- }
- ],
- "valueName": "current"
- },
- {
- "cacheTimeout": null,
- "colorBackground": false,
- "colorValue": false,
- "colors": [
- "#299c46",
- "rgba(237, 129, 40, 0.89)",
- "#d44a3a"
- ],
- "datasource": "$datasource",
- "description": "Average CPU busy across all hosts (OSD, RGW, MON etc) within the cluster",
- "decimals": 2,
- "format": "percentunit",
- "gauge": {
- "maxValue": 100,
- "minValue": 0,
- "show": false,
- "thresholdLabels": false,
- "thresholdMarkers": true
- },
- "gridPos": {
- "h": 5,
- "w": 4,
- "x": 4,
- "y": 0
- },
- "id": 6,
- "interval": null,
- "links": [],
- "mappingType": 1,
- "mappingTypes": [
- {
- "name": "value to text",
- "value": 1
- },
- {
- "name": "range to text",
- "value": 2
- }
- ],
- "maxDataPoints": 100,
- "nullPointMode": "connected",
- "nullText": null,
- "postfix": "",
- "postfixFontSize": "50%",
- "prefix": "",
- "prefixFontSize": "50%",
- "rangeMaps": [
- {
- "from": "null",
- "text": "N/A",
- "to": "null"
- }
- ],
- "sparkline": {
- "fillColor": "rgba(31, 118, 189, 0.18)",
- "full": false,
- "lineColor": "rgb(31, 120, 193)",
- "show": false
- },
- "tableColumn": "",
- "targets": [
- {
- "expr": "avg(\n 1 - (\n avg by(instance) \n (irate(node_cpu_seconds_total{mode='idle',instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"}[1m]) or\n irate(node_cpu{mode='idle',instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"}[1m]))\n )\n )",
- "format": "time_series",
- "instant": true,
- "intervalFactor": 1,
- "refId": "A"
- }
- ],
- "thresholds": "",
- "title": "AVG CPU Busy",
- "type": "singlestat",
- "valueFontSize": "80%",
- "valueMaps": [
- {
- "op": "=",
- "text": "N/A",
- "value": "null"
- }
- ],
- "valueName": "current"
- },
- {
- "cacheTimeout": null,
- "colorBackground": false,
- "colorValue": false,
- "colors": [
- "#299c46",
- "rgba(237, 129, 40, 0.89)",
- "#d44a3a"
- ],
- "datasource": "$datasource",
- "description": "Average Memory Usage across all hosts in the cluster (excludes buffer/cache usage)",
- "decimals": 2,
- "format": "percentunit",
- "gauge": {
- "maxValue": 100,
- "minValue": 0,
- "show": false,
- "thresholdLabels": false,
- "thresholdMarkers": true
- },
- "gridPos": {
- "h": 5,
- "w": 4,
- "x": 8,
- "y": 0
- },
- "id": 9,
- "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": [
- {
- "expr": "avg (((node_memory_MemTotal{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"} or node_memory_MemTotal_bytes{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"})- (\n (node_memory_MemFree{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"} or node_memory_MemFree_bytes{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"}) + \n (node_memory_Cached{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"} or node_memory_Cached_bytes{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"}) + \n (node_memory_Buffers{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"} or node_memory_Buffers_bytes{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"}) +\n (node_memory_Slab{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"} or node_memory_Slab_bytes{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"})\n )) /\n (node_memory_MemTotal{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"} or node_memory_MemTotal_bytes{instance=~\"($osd_hosts|$rgw_hosts|$mon_hosts|$mds_hosts).*\"} ))",
- "format": "time_series",
- "instant": true,
- "intervalFactor": 1,
- "refId": "A"
- }
- ],
- "thresholds": "",
- "title": "AVG RAM Utilization",
- "type": "singlestat",
- "valueFontSize": "80%",
- "valueMaps": [
- {
- "op": "=",
- "text": "N/A",
- "value": "null"
- }
- ],
- "valueName": "current"
- },
- {
- "cacheTimeout": null,
- "colorBackground": false,
- "colorValue": false,
- "colors": [
- "#299c46",
- "rgba(237, 129, 40, 0.89)",
- "#d44a3a"
- ],
- "datasource": "$datasource",
- "description": "IOPS Load at the device as reported by the OS on all OSD hosts",
- "format": "none",
- "gauge": {
- "maxValue": 100,
- "minValue": 0,
- "show": false,
- "thresholdLabels": false,
- "thresholdMarkers": true
- },
- "gridPos": {
- "h": 5,
- "w": 4,
- "x": 12,
- "y": 0
- },
- "id": 2,
- "interval": null,
- "links": [],
- "mappingType": 1,
- "mappingTypes": [
- {
- "name": "value to text",
- "value": 1
- },
- {
- "name": "range to text",
- "value": 2
- }
- ],
- "maxDataPoints": 100,
- "nullPointMode": "connected",
- "nullText": null,
- "postfix": "",
- "postfixFontSize": "50%",
- "prefix": "",
- "prefixFontSize": "50%",
- "rangeMaps": [
- {
- "from": "null",
- "text": "N/A",
- "to": "null"
- }
- ],
- "sparkline": {
- "fillColor": "rgba(31, 118, 189, 0.18)",
- "full": false,
- "lineColor": "rgb(31, 120, 193)",
- "show": false
- },
- "tableColumn": "",
- "targets": [
- {
- "expr": "sum ((irate(node_disk_reads_completed{instance=~\"($osd_hosts).*\"}[5m]) or irate(node_disk_reads_completed_total{instance=~\"($osd_hosts).*\"}[5m]) ) + \n(irate(node_disk_writes_completed{instance=~\"($osd_hosts).*\"}[5m]) or irate(node_disk_writes_completed_total{instance=~\"($osd_hosts).*\"}[5m])))",
- "format": "time_series",
- "instant": true,
- "intervalFactor": 1,
- "refId": "A"
- }
- ],
- "thresholds": "",
- "title": "Physical IOPS",
- "type": "singlestat",
- "valueFontSize": "80%",
- "valueMaps": [
- {
- "op": "=",
- "text": "N/A",
- "value": "null"
- }
- ],
- "valueName": "current"
- },
- {
- "cacheTimeout": null,
- "colorBackground": false,
- "colorValue": false,
- "colors": [
- "#299c46",
- "rgba(237, 129, 40, 0.89)",
- "#d44a3a"
- ],
- "datasource": "$datasource",
- "description": "Average Disk utilization for all OSD data devices (i.e. excludes journal/WAL)",
- "format": "percent",
- "gauge": {
- "maxValue": 100,
- "minValue": 0,
- "show": false,
- "thresholdLabels": false,
- "thresholdMarkers": true
- },
- "gridPos": {
- "h": 5,
- "w": 4,
- "x": 16,
- "y": 0
- },
- "id": 20,
- "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": [
- {
- "expr" : "avg (\n label_replace((irate(node_disk_io_time_ms[5m]) / 10 ) or\n (irate(node_disk_io_time_seconds_total[5m]) * 100), \"instance\", \"$1\", \"instance\", \"([^.:]*).*\"\n ) *\n on(instance, device, ceph_daemon) label_replace(label_replace(ceph_disk_occupation{instance=~\"($osd_hosts).*\"}, \"device\", \"$1\", \"device\", \"/dev/(.*)\"), \"instance\", \"$1\", \"instance\", \"([^.:]*).*\")\n)",
- "format": "time_series",
- "instant": true,
- "intervalFactor": 1,
- "refId": "A"
- }
- ],
- "thresholds": "",
- "title": "AVG Disk Utilization",
- "type": "singlestat",
- "valueFontSize": "80%",
- "valueMaps": [
- {
- "op": "=",
- "text": "N/A",
- "value": "null"
- }
- ],
- "valueName": "current"
- },
- {
- "cacheTimeout": null,
- "colorBackground": false,
- "colorValue": false,
- "colors": [
- "#299c46",
- "rgba(237, 129, 40, 0.89)",
- "#d44a3a"
- ],
- "datasource": "$datasource",
- "decimals": 0,
- "description": "Total send/receive network load across all hosts in the ceph cluster",
- "format": "bytes",
- "gauge": {
- "maxValue": 100,
- "minValue": 0,
- "show": false,
- "thresholdLabels": false,
- "thresholdMarkers": true
- },
- "gridPos": {
- "h": 5,
- "w": 4,
- "x": 20,
- "y": 0
+ "id": "grafana",
+ "name": "Grafana",
+ "type": "grafana",
+ "version": "5.3.2"
},
- "id": 18,
- "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": [
- {
- "expr": "sum (\n irate(node_network_receive_bytes{instance=~\"($osd_hosts|mon_hosts|mds_hosts|rgw_hosts).*\",device!=\"lo\"}[1m]) or\n irate(node_network_receive_bytes_total{instance=~\"($osd_hosts|mon_hosts|mds_hosts|rgw_hosts).*\",device!=\"lo\"}[1m])\n ) +\nsum (\n irate(node_network_transmit_bytes{instance=~\"($osd_hosts|mon_hosts|mds_hosts|rgw_hosts).*\",device!=\"lo\"}[1m]) or\n irate(node_network_transmit_bytes_total{instance=~\"($osd_hosts|mon_hosts|mds_hosts|rgw_hosts).*\",device!=\"lo\"}[1m])\n )",
- "format": "time_series",
- "instant": true,
- "intervalFactor": 1,
- "refId": "A"
- }
- ],
- "thresholds": "",
- "title": "Network Load",
- "type": "singlestat",
- "valueFontSize": "80%",
- "valueMaps": [
- {
- "op": "=",
- "text": "N/A",
- "value": "null"
- }
- ],
- "valueName": "current"
- },
- {
- "aliasColors": {},
- "bars": false,
- "dashLength": 10,
- "dashes": false,
- "datasource": "$datasource",
- "description": "Show the top 10 busiest hosts by cpu",
- "fill": 0,
- "gridPos": {
- "h": 9,
- "w": 12,
- "x": 0,
- "y": 5
- },
- "id": 13,
- "legend": {
- "avg": false,
- "current": false,
- "max": false,
- "min": false,
- "show": false,
- "total": false,
- "values": false
- },
- "lines": true,
- "linewidth": 1,
- "links": [],
- "nullPointMode": "null",
- "percentage": false,
- "pointradius": 5,
- "points": false,
- "renderer": "flot",
- "seriesOverrides": [],
- "spaceLength": 10,
- "stack": false,
- "steppedLine": false,
- "targets": [
- {
- "expr": "topk(10,100 * ( 1 - (\n avg by(instance) \n (irate(node_cpu_seconds_total{mode='idle',instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"}[1m]) or\n irate(node_cpu{mode='idle',instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"}[1m]))\n )\n )\n)",
- "format": "time_series",
- "intervalFactor": 1,
- "legendFormat": "{{instance}}",
- "refId": "A"
- }
- ],
- "thresholds": [],
- "timeFrom": null,
- "timeRegions": [],
- "timeShift": null,
- "title": "CPU Busy - Top 10 Hosts",
- "tooltip": {
- "shared": true,
- "sort": 2,
- "value_type": "individual"
- },
- "type": "graph",
- "xaxis": {
- "buckets": null,
- "mode": "time",
- "name": null,
- "show": true,
- "values": []
+ {
+ "id": "graph",
+ "name": "Graph",
+ "type": "panel",
+ "version": "5.0.0"
},
- "yaxes": [
- {
- "decimals": 1,
- "format": "percent",
- "label": null,
- "logBase": 1,
- "max": "100",
- "min": "0",
- "show": true
- },
- {
- "format": "short",
- "label": null,
- "logBase": 1,
- "max": null,
- "min": null,
- "show": false
- }
- ],
- "yaxis": {
- "align": false,
- "alignLevel": null
+ {
+ "id": "singlestat",
+ "name": "Singlestat",
+ "type": "panel",
+ "version": "5.0.0"
}
- },
- {
- "aliasColors": {},
- "bars": false,
- "dashLength": 10,
- "dashes": false,
- "datasource": "$datasource",
- "description": "Top 10 hosts by network load",
- "fill": 0,
- "gridPos": {
- "h": 9,
- "w": 12,
- "x": 12,
- "y": 5
- },
- "id": 19,
- "legend": {
- "avg": false,
- "current": false,
- "max": false,
- "min": false,
- "show": false,
- "total": false,
- "values": false
+ ],
+ "annotations": {
+ "list": [
+ {
+ "builtIn": 1,
+ "datasource": "-- Grafana --",
+ "enable": true,
+ "hide": true,
+ "iconColor": "rgba(0, 211, 255, 1)",
+ "name": "Annotations & Alerts",
+ "showIn": 0,
+ "tags": [ ],
+ "type": "dashboard"
+ }
+ ]
+ },
+ "description": "",
+ "editable": false,
+ "gnetId": null,
+ "graphTooltip": 0,
+ "hideControls": false,
+ "id": null,
+ "links": [ ],
+ "panels": [
+ {
+ "cacheTimeout": null,
+ "colorBackground": false,
+ "colorValue": false,
+ "colors": [
+ "#299c46",
+ "rgba(237, 129, 40, 0.89)",
+ "#d44a3a"
+ ],
+ "datasource": "$datasource",
+ "format": "none",
+ "gauge": {
+ "maxValue": 100,
+ "minValue": 0,
+ "show": false,
+ "thresholdLabels": false,
+ "thresholdMarkers": true
+ },
+ "gridPos": {
+ "h": 5,
+ "w": 4,
+ "x": 0,
+ "y": 0
+ },
+ "id": 2,
+ "interval": null,
+ "links": [ ],
+ "mappingType": 1,
+ "mappingTypes": [
+ {
+ "name": "value to text",
+ "value": 1
+ },
+ {
+ "name": "range to text",
+ "value": 2
+ }
+ ],
+ "maxDataPoints": 100,
+ "nullPointMode": "connected",
+ "nullText": null,
+ "postfix": "",
+ "postfixFontSize": "50%",
+ "prefix": "",
+ "prefixFontSize": "50%",
+ "rangeMaps": [
+ {
+ "from": "null",
+ "text": "N/A",
+ "to": "null"
+ }
+ ],
+ "sparkline": {
+ "fillColor": "rgba(31, 118, 189, 0.18)",
+ "full": false,
+ "lineColor": "rgb(31, 120, 193)",
+ "show": false
+ },
+ "tableColumn": "",
+ "targets": [
+ {
+ "expr": "count(sum by (hostname) (ceph_osd_metadata))",
+ "format": "time_series",
+ "intervalFactor": 1,
+ "legendFormat": "",
+ "refId": "A"
+ }
+ ],
+ "thresholds": "",
+ "title": "OSD Hosts",
+ "type": "singlestat",
+ "valueFontSize": "80%",
+ "valueMaps": [
+ {
+ "op": "=",
+ "text": "N/A",
+ "value": "null"
+ }
+ ],
+ "valueName": "current"
},
- "lines": true,
- "linewidth": 1,
- "links": [],
- "nullPointMode": "null",
- "percentage": false,
- "pointradius": 5,
- "points": false,
- "renderer": "flot",
- "seriesOverrides": [],
- "spaceLength": 10,
- "stack": false,
- "steppedLine": false,
- "targets": [
- {
- "expr": "topk(10, (sum by(instance) (\n (\n irate(node_network_receive_bytes{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\",device!=\"lo\"}[1m]) or\n irate(node_network_receive_bytes_total{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\",device!=\"lo\"}[1m])\n ) +\n (\n irate(node_network_transmit_bytes{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\",device!=\"lo\"}[1m]) or\n irate(node_network_transmit_bytes_total{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\",device!=\"lo\"}[1m])\n ))\n )\n)",
- "format": "time_series",
- "intervalFactor": 1,
- "legendFormat": "{{instance}}",
- "refId": "A"
- }
- ],
- "thresholds": [],
- "timeFrom": null,
- "timeRegions": [],
- "timeShift": null,
- "title": "Network Load - Top 10 Hosts",
- "tooltip": {
- "shared": true,
- "sort": 2,
- "value_type": "individual"
+ {
+ "cacheTimeout": null,
+ "colorBackground": false,
+ "colorValue": false,
+ "colors": [
+ "#299c46",
+ "rgba(237, 129, 40, 0.89)",
+ "#d44a3a"
+ ],
+ "datasource": "$datasource",
+ "description": "Average CPU busy across all hosts (OSD, RGW, MON etc) within the cluster",
+ "format": "percentunit",
+ "gauge": {
+ "maxValue": 100,
+ "minValue": 0,
+ "show": false,
+ "thresholdLabels": false,
+ "thresholdMarkers": true
+ },
+ "gridPos": {
+ "h": 5,
+ "w": 4,
+ "x": 4,
+ "y": 0
+ },
+ "id": 3,
+ "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": [
+ {
+ "expr": "avg(\n 1 - (\n avg by(instance) \n (irate(node_cpu_seconds_total{mode='idle',instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"}[1m]) or\n irate(node_cpu{mode='idle',instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"}[1m]))\n )\n )",
+ "format": "time_series",
+ "intervalFactor": 1,
+ "legendFormat": "",
+ "refId": "A"
+ }
+ ],
+ "thresholds": "",
+ "title": "AVG CPU Busy",
+ "type": "singlestat",
+ "valueFontSize": "80%",
+ "valueMaps": [
+ {
+ "op": "=",
+ "text": "N/A",
+ "value": "null"
+ }
+ ],
+ "valueName": "current"
},
- "type": "graph",
- "xaxis": {
- "buckets": null,
- "mode": "time",
- "name": null,
- "show": true,
- "values": []
+ {
+ "cacheTimeout": null,
+ "colorBackground": false,
+ "colorValue": false,
+ "colors": [
+ "#299c46",
+ "rgba(237, 129, 40, 0.89)",
+ "#d44a3a"
+ ],
+ "datasource": "$datasource",
+ "description": "Average Memory Usage across all hosts in the cluster (excludes buffer/cache usage)",
+ "format": "percentunit",
+ "gauge": {
+ "maxValue": 100,
+ "minValue": 0,
+ "show": false,
+ "thresholdLabels": false,
+ "thresholdMarkers": true
+ },
+ "gridPos": {
+ "h": 5,
+ "w": 4,
+ "x": 8,
+ "y": 0
+ },
+ "id": 4,
+ "interval": null,
+ "links": [ ],
+ "mappingType": 1,
+ "mappingTypes": [
+ {
+ "name": "value to text",
+ "value": 1
+ },
+ {
+ "name": "range to text",
+ "value": 2
+ }
+ ],
+ "maxDataPoints": 100,
+ "nullPointMode": "connected",
+ "nullText": null,
+ "postfix": "",
+ "postfixFontSize": "50%",
+ "prefix": "",
+ "prefixFontSize": "50%",
+ "rangeMaps": [
+ {
+ "from": "null",
+ "text": "N/A",
+ "to": "null"
+ }
+ ],
+ "sparkline": {
+ "fillColor": "rgba(31, 118, 189, 0.18)",
+ "full": false,
+ "lineColor": "rgb(31, 120, 193)",
+ "show": false
+ },
+ "tableColumn": "",
+ "targets": [
+ {
+ "expr": "avg (((node_memory_MemTotal{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"} or node_memory_MemTotal_bytes{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"})- (\n (node_memory_MemFree{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"} or node_memory_MemFree_bytes{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"}) + \n (node_memory_Cached{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"} or node_memory_Cached_bytes{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"}) + \n (node_memory_Buffers{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"} or node_memory_Buffers_bytes{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"}) +\n (node_memory_Slab{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"} or node_memory_Slab_bytes{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"})\n )) /\n (node_memory_MemTotal{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"} or node_memory_MemTotal_bytes{instance=~\"($osd_hosts|$rgw_hosts|$mon_hosts|$mds_hosts).*\"} ))",
+ "format": "time_series",
+ "intervalFactor": 1,
+ "legendFormat": "",
+ "refId": "A"
+ }
+ ],
+ "thresholds": "",
+ "title": "AVG RAM Utilization",
+ "type": "singlestat",
+ "valueFontSize": "80%",
+ "valueMaps": [
+ {
+ "op": "=",
+ "text": "N/A",
+ "value": "null"
+ }
+ ],
+ "valueName": "current"
},
- "yaxes": [
- {
- "decimals": 1,
- "format": "Bps",
- "label": null,
- "logBase": 1,
- "max": null,
- "min": "0",
- "show": true
- },
- {
- "format": "short",
- "label": null,
- "logBase": 1,
- "max": null,
- "min": null,
- "show": false
- }
- ],
- "yaxis": {
- "align": false,
- "alignLevel": null
- }
- }
- ],
- "refresh": "10s",
- "schemaVersion": 16,
- "style": "dark",
- "tags": [],
- "templating": {
- "list": [
{
- "current": {
- "text": "default",
- "value": "default"
- },
- "hide": 0,
- "label": "Data Source",
- "name": "datasource",
- "options": [],
- "query": "prometheus",
- "refresh": 1,
- "regex": "",
- "skipUrlSync": false,
- "type": "datasource"
+ "cacheTimeout": null,
+ "colorBackground": false,
+ "colorValue": false,
+ "colors": [
+ "#299c46",
+ "rgba(237, 129, 40, 0.89)",
+ "#d44a3a"
+ ],
+ "datasource": "$datasource",
+ "description": "IOPS Load at the device as reported by the OS on all OSD hosts",
+ "format": "none",
+ "gauge": {
+ "maxValue": 100,
+ "minValue": 0,
+ "show": false,
+ "thresholdLabels": false,
+ "thresholdMarkers": true
+ },
+ "gridPos": {
+ "h": 5,
+ "w": 4,
+ "x": 12,
+ "y": 0
+ },
+ "id": 5,
+ "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": [
+ {
+ "expr": "sum ((irate(node_disk_reads_completed{instance=~\"($osd_hosts).*\"}[5m]) or irate(node_disk_reads_completed_total{instance=~\"($osd_hosts).*\"}[5m]) ) + \n(irate(node_disk_writes_completed{instance=~\"($osd_hosts).*\"}[5m]) or irate(node_disk_writes_completed_total{instance=~\"($osd_hosts).*\"}[5m])))",
+ "format": "time_series",
+ "intervalFactor": 1,
+ "legendFormat": "",
+ "refId": "A"
+ }
+ ],
+ "thresholds": "",
+ "title": "Physical IOPS",
+ "type": "singlestat",
+ "valueFontSize": "80%",
+ "valueMaps": [
+ {
+ "op": "=",
+ "text": "N/A",
+ "value": "null"
+ }
+ ],
+ "valueName": "current"
},
{
- "allValue": "",
- "current": {},
- "datasource": "$datasource",
- "hide": 2,
- "includeAll": true,
- "label": null,
- "multi": false,
- "name": "osd_hosts",
- "options": [],
- "query": "label_values(ceph_disk_occupation, exported_instance)",
- "refresh": 1,
- "regex": "([^.]*).*",
- "skipUrlSync": false,
- "sort": 1,
- "tagValuesQuery": "",
- "tags": [],
- "tagsQuery": "ceph",
- "type": "query",
- "useTags": false
+ "cacheTimeout": null,
+ "colorBackground": false,
+ "colorValue": false,
+ "colors": [
+ "#299c46",
+ "rgba(237, 129, 40, 0.89)",
+ "#d44a3a"
+ ],
+ "datasource": "$datasource",
+ "description": "Average Disk utilization for all OSD data devices (i.e. excludes journal/WAL)",
+ "format": "percent",
+ "gauge": {
+ "maxValue": 100,
+ "minValue": 0,
+ "show": false,
+ "thresholdLabels": false,
+ "thresholdMarkers": true
+ },
+ "gridPos": {
+ "h": 5,
+ "w": 4,
+ "x": 16,
+ "y": 0
+ },
+ "id": 6,
+ "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": [
+ {
+ "expr": "avg (\n label_replace((irate(node_disk_io_time_ms[5m]) / 10 ) or\n (irate(node_disk_io_time_seconds_total[5m]) * 100), \"instance\", \"$1\", \"instance\", \"([^.:]*).*\"\n ) *\n on(instance, device, ceph_daemon) label_replace(label_replace(ceph_disk_occupation{instance=~\"($osd_hosts).*\"}, \"device\", \"$1\", \"device\", \"/dev/(.*)\"), \"instance\", \"$1\", \"instance\", \"([^.:]*).*\")\n)",
+ "format": "time_series",
+ "intervalFactor": 1,
+ "legendFormat": "",
+ "refId": "A"
+ }
+ ],
+ "thresholds": "",
+ "title": "AVG Disk Utilization",
+ "type": "singlestat",
+ "valueFontSize": "80%",
+ "valueMaps": [
+ {
+ "op": "=",
+ "text": "N/A",
+ "value": "null"
+ }
+ ],
+ "valueName": "current"
},
{
- "allValue": null,
- "current": {},
- "datasource": "$datasource",
- "hide": 2,
- "includeAll": true,
- "label": null,
- "multi": false,
- "name": "mon_hosts",
- "options": [],
- "query": "label_values(ceph_mon_metadata, ceph_daemon)",
- "refresh": 1,
- "regex": "mon.(.*)",
- "skipUrlSync": false,
- "sort": 0,
- "tagValuesQuery": "",
- "tags": [],
- "tagsQuery": "",
- "type": "query",
- "useTags": false
+ "cacheTimeout": null,
+ "colorBackground": false,
+ "colorValue": false,
+ "colors": [
+ "#299c46",
+ "rgba(237, 129, 40, 0.89)",
+ "#d44a3a"
+ ],
+ "datasource": "$datasource",
+ "description": "Total send/receive network load across all hosts in the ceph cluster",
+ "format": "bytes",
+ "gauge": {
+ "maxValue": 100,
+ "minValue": 0,
+ "show": false,
+ "thresholdLabels": false,
+ "thresholdMarkers": true
+ },
+ "gridPos": {
+ "h": 5,
+ "w": 4,
+ "x": 20,
+ "y": 0
+ },
+ "id": 7,
+ "interval": null,
+ "links": [ ],
+ "mappingType": 1,
+ "mappingTypes": [
+ {
+ "name": "value to text",
+ "value": 1
+ },
+ {
+ "name": "range to text",
+ "value": 2
+ }
+ ],
+ "maxDataPoints": 100,
+ "nullPointMode": "connected",
+ "nullText": null,
+ "postfix": "",
+ "postfixFontSize": "50%",
+ "prefix": "",
+ "prefixFontSize": "50%",
+ "rangeMaps": [
+ {
+ "from": "null",
+ "text": "N/A",
+ "to": "null"
+ }
+ ],
+ "sparkline": {
+ "fillColor": "rgba(31, 118, 189, 0.18)",
+ "full": false,
+ "lineColor": "rgb(31, 120, 193)",
+ "show": false
+ },
+ "tableColumn": "",
+ "targets": [
+ {
+ "expr": "sum (\n irate(node_network_receive_bytes{instance=~\"($osd_hosts|mon_hosts|mds_hosts|rgw_hosts).*\",device!=\"lo\"}[1m]) or\n irate(node_network_receive_bytes_total{instance=~\"($osd_hosts|mon_hosts|mds_hosts|rgw_hosts).*\",device!=\"lo\"}[1m])\n ) +\nsum (\n irate(node_network_transmit_bytes{instance=~\"($osd_hosts|mon_hosts|mds_hosts|rgw_hosts).*\",device!=\"lo\"}[1m]) or\n irate(node_network_transmit_bytes_total{instance=~\"($osd_hosts|mon_hosts|mds_hosts|rgw_hosts).*\",device!=\"lo\"}[1m])\n )",
+ "format": "time_series",
+ "intervalFactor": 1,
+ "legendFormat": "",
+ "refId": "A"
+ }
+ ],
+ "thresholds": "",
+ "title": "Network Load",
+ "type": "singlestat",
+ "valueFontSize": "80%",
+ "valueMaps": [
+ {
+ "op": "=",
+ "text": "N/A",
+ "value": "null"
+ }
+ ],
+ "valueName": "current"
},
{
- "allValue": null,
- "current": {},
- "datasource": "$datasource",
- "hide": 2,
- "includeAll": true,
- "label": null,
- "multi": false,
- "name": "mds_hosts",
- "options": [],
- "query": "label_values(ceph_mds_inodes, ceph_daemon)",
- "refresh": 1,
- "regex": "mds.(.*)",
- "skipUrlSync": false,
- "sort": 0,
- "tagValuesQuery": "",
- "tags": [],
- "tagsQuery": "",
- "type": "query",
- "useTags": false
+ "aliasColors": { },
+ "bars": false,
+ "dashLength": 10,
+ "dashes": false,
+ "datasource": "$datasource",
+ "description": "Show the top 10 busiest hosts by cpu",
+ "fill": 1,
+ "gridPos": {
+ "h": 9,
+ "w": 12,
+ "x": 0,
+ "y": 5
+ },
+ "id": 8,
+ "legend": {
+ "alignAsTable": false,
+ "avg": false,
+ "current": false,
+ "max": false,
+ "min": false,
+ "rightSide": false,
+ "show": true,
+ "sideWidth": null,
+ "total": false,
+ "values": false
+ },
+ "lines": true,
+ "linewidth": 1,
+ "links": [ ],
+ "nullPointMode": "null",
+ "percentage": false,
+ "pointradius": 5,
+ "points": false,
+ "renderer": "flot",
+ "repeat": null,
+ "seriesOverrides": [ ],
+ "spaceLength": 10,
+ "stack": false,
+ "steppedLine": false,
+ "targets": [
+ {
+ "expr": "topk(10,100 * ( 1 - (\n avg by(instance) \n (irate(node_cpu_seconds_total{mode='idle',instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"}[1m]) or\n irate(node_cpu{mode='idle',instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"}[1m]))\n )\n )\n)",
+ "format": "time_series",
+ "intervalFactor": 1,
+ "legendFormat": "{{instance}}",
+ "refId": "A"
+ }
+ ],
+ "thresholds": [ ],
+ "timeFrom": null,
+ "timeShift": null,
+ "title": "CPU Busy - Top 10 Hosts",
+ "tooltip": {
+ "shared": true,
+ "sort": 0,
+ "value_type": "individual"
+ },
+ "type": "graph",
+ "xaxis": {
+ "buckets": null,
+ "mode": "time",
+ "name": null,
+ "show": true,
+ "values": [ ]
+ },
+ "yaxes": [
+ {
+ "format": "percent",
+ "label": null,
+ "logBase": 1,
+ "max": null,
+ "min": 0,
+ "show": true
+ },
+ {
+ "format": "short",
+ "label": null,
+ "logBase": 1,
+ "max": null,
+ "min": 0,
+ "show": true
+ }
+ ]
},
{
- "allValue": null,
- "current": {},
- "datasource": "$datasource",
- "hide": 2,
- "includeAll": true,
- "label": null,
- "multi": false,
- "name": "rgw_hosts",
- "options": [],
- "query": "label_values(ceph_rgw_qlen, ceph_daemon)",
- "refresh": 1,
- "regex": "rgw.(.*)",
- "skipUrlSync": false,
- "sort": 0,
- "tagValuesQuery": "",
- "tags": [],
- "tagsQuery": "",
- "type": "query",
- "useTags": false
+ "aliasColors": { },
+ "bars": false,
+ "dashLength": 10,
+ "dashes": false,
+ "datasource": "$datasource",
+ "description": "Top 10 hosts by network load",
+ "fill": 1,
+ "gridPos": {
+ "h": 9,
+ "w": 12,
+ "x": 12,
+ "y": 5
+ },
+ "id": 9,
+ "legend": {
+ "alignAsTable": false,
+ "avg": false,
+ "current": false,
+ "max": false,
+ "min": false,
+ "rightSide": false,
+ "show": true,
+ "sideWidth": null,
+ "total": false,
+ "values": false
+ },
+ "lines": true,
+ "linewidth": 1,
+ "links": [ ],
+ "nullPointMode": "null",
+ "percentage": false,
+ "pointradius": 5,
+ "points": false,
+ "renderer": "flot",
+ "repeat": null,
+ "seriesOverrides": [ ],
+ "spaceLength": 10,
+ "stack": false,
+ "steppedLine": false,
+ "targets": [
+ {
+ "expr": "topk(10, (sum by(instance) (\n (\n irate(node_network_receive_bytes{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\",device!=\"lo\"}[1m]) or\n irate(node_network_receive_bytes_total{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\",device!=\"lo\"}[1m])\n ) +\n (\n irate(node_network_transmit_bytes{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\",device!=\"lo\"}[1m]) or\n irate(node_network_transmit_bytes_total{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\",device!=\"lo\"}[1m])\n ))\n )\n)",
+ "format": "time_series",
+ "intervalFactor": 1,
+ "legendFormat": "{{instance}}",
+ "refId": "A"
+ }
+ ],
+ "thresholds": [ ],
+ "timeFrom": null,
+ "timeShift": null,
+ "title": "Network Load - Top 10 Hosts",
+ "tooltip": {
+ "shared": true,
+ "sort": 0,
+ "value_type": "individual"
+ },
+ "type": "graph",
+ "xaxis": {
+ "buckets": null,
+ "mode": "time",
+ "name": null,
+ "show": true,
+ "values": [ ]
+ },
+ "yaxes": [
+ {
+ "format": "Bps",
+ "label": null,
+ "logBase": 1,
+ "max": null,
+ "min": 0,
+ "show": true
+ },
+ {
+ "format": "short",
+ "label": null,
+ "logBase": 1,
+ "max": null,
+ "min": 0,
+ "show": true
+ }
+ ]
}
- ]
- },
- "time": {
- "from": "now-1h",
- "to": "now"
- },
- "timepicker": {
- "refresh_intervals": [
- "5s",
- "10s",
- "30s",
- "1m",
- "5m",
- "15m",
- "30m",
- "1h",
- "2h",
- "1d"
- ],
- "time_options": [
- "5m",
- "15m",
- "1h",
- "6h",
- "12h",
- "24h",
- "2d",
- "7d",
- "30d"
- ]
- },
- "timezone": "",
- "title": "Host Overview",
- "uid": "y0KGL0iZz",
- "version": 3
+ ],
+ "refresh": "10s",
+ "rows": [ ],
+ "schemaVersion": 16,
+ "style": "dark",
+ "tags": [ ],
+ "templating": {
+ "list": [
+ {
+ "current": {
+ "text": "default",
+ "value": "default"
+ },
+ "hide": 0,
+ "label": "Data Source",
+ "name": "datasource",
+ "options": [ ],
+ "query": "prometheus",
+ "refresh": 1,
+ "regex": "",
+ "type": "datasource"
+ },
+ {
+ "allValue": null,
+ "current": { },
+ "datasource": "$datasource",
+ "hide": 0,
+ "includeAll": true,
+ "label": null,
+ "multi": false,
+ "name": "osd_hosts",
+ "options": [ ],
+ "query": "label_values(ceph_disk_occupation, exported_instance)",
+ "refresh": 1,
+ "regex": "([^.]*).*",
+ "sort": 1,
+ "tagValuesQuery": "",
+ "tags": [ ],
+ "tagsQuery": "",
+ "type": "query",
+ "useTags": false
+ },
+ {
+ "allValue": null,
+ "current": { },
+ "datasource": "$datasource",
+ "hide": 0,
+ "includeAll": true,
+ "label": null,
+ "multi": false,
+ "name": "mon_hosts",
+ "options": [ ],
+ "query": "label_values(ceph_mon_metadata, ceph_daemon)",
+ "refresh": 1,
+ "regex": "mon.(.*)",
+ "sort": 1,
+ "tagValuesQuery": "",
+ "tags": [ ],
+ "tagsQuery": "",
+ "type": "query",
+ "useTags": false
+ },
+ {
+ "allValue": null,
+ "current": { },
+ "datasource": "$datasource",
+ "hide": 0,
+ "includeAll": true,
+ "label": null,
+ "multi": false,
+ "name": "mds_hosts",
+ "options": [ ],
+ "query": "label_values(ceph_mds_inodes, ceph_daemon)",
+ "refresh": 1,
+ "regex": "mds.(.*)",
+ "sort": 1,
+ "tagValuesQuery": "",
+ "tags": [ ],
+ "tagsQuery": "",
+ "type": "query",
+ "useTags": false
+ },
+ {
+ "allValue": null,
+ "current": { },
+ "datasource": "$datasource",
+ "hide": 0,
+ "includeAll": true,
+ "label": null,
+ "multi": false,
+ "name": "rgw_hosts",
+ "options": [ ],
+ "query": "label_values(ceph_rgw_qlen, ceph_daemon)",
+ "refresh": 1,
+ "regex": "rgw.(.*)",
+ "sort": 1,
+ "tagValuesQuery": "",
+ "tags": [ ],
+ "tagsQuery": "",
+ "type": "query",
+ "useTags": false
+ }
+ ]
+ },
+ "time": {
+ "from": "now-1h",
+ "to": "now"
+ },
+ "timepicker": {
+ "refresh_intervals": [
+ "5s",
+ "10s",
+ "30s",
+ "1m",
+ "5m",
+ "15m",
+ "30m",
+ "1h",
+ "2h",
+ "1d"
+ ],
+ "time_options": [
+ "5m",
+ "15m",
+ "1h",
+ "6h",
+ "12h",
+ "24h",
+ "2d",
+ "7d",
+ "30d"
+ ]
+ },
+ "timezone": "",
+ "title": "Host Overview",
+ "uid": "y0KGL0iZz",
+ "version": 0
}
local addStyle(alias, colorMode, colors, dateFormat, decimals, mappingType, pattern, thresholds, type, unit, valueMaps) =
{'alias': alias, 'colorMode': colorMode, 'colors':colors, 'dateFormat':dateFormat, 'decimals':decimals, 'mappingType':mappingType, 'pattern':pattern, 'thresholds':thresholds, 'type':type, 'unit':unit, 'valueMaps':valueMaps};
+{
+ "hosts-overview.json":
+ local HostsOverviewSingleStatPanel(format, title, description, valueName, expr, targetFormat, x, y, w, h) =
+ addSingelStatSchema('$datasource', format, title, description, valueName, false, 100, false, false, '')
+ .addTarget(addTargetSchema(expr, 1, targetFormat, '')) + {gridPos: {x: x, y: y, w: w, h: h}};
+
+ local HostsOverviewGraphPanel(title, description, formatY1, expr, legendFormat, x, y, w, h) =
+ graphPanelSchema({}, title, description, 'null', false, formatY1, 'short', null, null, 0, 1, '$datasource')
+ .addTargets(
+ [addTargetSchema(expr, 1, 'time_series', legendFormat)]) + {gridPos: {x: x, y: y, w: w, h: h}};
+
+ dashboardSchema(
+ 'Host Overview', '', 'y0KGL0iZz', 'now-1h', '10s', 16, [], '', {refresh_intervals:['5s','10s','30s','1m','5m','15m','30m','1h','2h','1d'],time_options:['5m','15m','1h','6h','12h','24h','2d','7d','30d']}
+ )
+ .addRequired(
+ type='grafana', id='grafana', name='Grafana', version='5.3.2'
+ )
+ .addRequired(
+ type='panel', id='graph', name='Graph', version='5.0.0'
+ )
+ .addRequired(
+ type='panel', id='singlestat', name='Singlestat', version='5.0.0'
+ )
+ .addAnnotation(
+ addAnnotationSchema(
+ 1, '-- Grafana --', true, true, 'rgba(0, 211, 255, 1)', 'Annotations & Alerts', 'dashboard')
+ )
+ .addTemplate(
+ g.template.datasource('datasource', 'prometheus', 'default', label='Data Source')
+ )
+ .addTemplate(
+ addTemplateSchema('osd_hosts', '$datasource', 'label_values(ceph_disk_occupation, exported_instance)', 1, true, 1, null, '([^.]*).*')
+ )
+ .addTemplate(
+ addTemplateSchema('mon_hosts', '$datasource', 'label_values(ceph_mon_metadata, ceph_daemon)', 1, true, 1, null, 'mon.(.*)')
+ )
+ .addTemplate(
+ addTemplateSchema('mds_hosts', '$datasource', 'label_values(ceph_mds_inodes, ceph_daemon)', 1, true, 1, null, 'mds.(.*)')
+ )
+ .addTemplate(
+ addTemplateSchema('rgw_hosts', '$datasource', 'label_values(ceph_rgw_qlen, ceph_daemon)', 1, true, 1, null, 'rgw.(.*)')
+ )
+ .addPanels([
+ HostsOverviewSingleStatPanel(
+ 'none', 'OSD Hosts', '', 'current', 'count(sum by (hostname) (ceph_osd_metadata))', 'time_series', 0, 0, 4, 5),
+ HostsOverviewSingleStatPanel(
+ 'percentunit', 'AVG CPU Busy', 'Average CPU busy across all hosts (OSD, RGW, MON etc) within the cluster', 'current', 'avg(\n 1 - (\n avg by(instance) \n (irate(node_cpu_seconds_total{mode=\'idle\',instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"}[1m]) or\n irate(node_cpu{mode=\'idle\',instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"}[1m]))\n )\n )', 'time_series', 4, 0, 4, 5),
+ HostsOverviewSingleStatPanel(
+ 'percentunit', 'AVG RAM Utilization', 'Average Memory Usage across all hosts in the cluster (excludes buffer/cache usage)', 'current', 'avg (((node_memory_MemTotal{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"} or node_memory_MemTotal_bytes{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"})- (\n (node_memory_MemFree{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"} or node_memory_MemFree_bytes{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"}) + \n (node_memory_Cached{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"} or node_memory_Cached_bytes{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"}) + \n (node_memory_Buffers{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"} or node_memory_Buffers_bytes{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"}) +\n (node_memory_Slab{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"} or node_memory_Slab_bytes{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"})\n )) /\n (node_memory_MemTotal{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"} or node_memory_MemTotal_bytes{instance=~\"($osd_hosts|$rgw_hosts|$mon_hosts|$mds_hosts).*\"} ))', 'time_series', 8, 0, 4, 5),
+ HostsOverviewSingleStatPanel(
+ 'none', 'Physical IOPS', 'IOPS Load at the device as reported by the OS on all OSD hosts', 'current', 'sum ((irate(node_disk_reads_completed{instance=~\"($osd_hosts).*\"}[5m]) or irate(node_disk_reads_completed_total{instance=~\"($osd_hosts).*\"}[5m]) ) + \n(irate(node_disk_writes_completed{instance=~\"($osd_hosts).*\"}[5m]) or irate(node_disk_writes_completed_total{instance=~\"($osd_hosts).*\"}[5m])))', 'time_series', 12, 0, 4, 5),
+ HostsOverviewSingleStatPanel(
+ 'percent', 'AVG Disk Utilization', 'Average Disk utilization for all OSD data devices (i.e. excludes journal/WAL)', 'current', 'avg (\n label_replace((irate(node_disk_io_time_ms[5m]) / 10 ) or\n (irate(node_disk_io_time_seconds_total[5m]) * 100), \"instance\", \"$1\", \"instance\", \"([^.:]*).*\"\n ) *\n on(instance, device, ceph_daemon) label_replace(label_replace(ceph_disk_occupation{instance=~\"($osd_hosts).*\"}, \"device\", \"$1\", \"device\", \"/dev/(.*)\"), \"instance\", \"$1\", \"instance\", \"([^.:]*).*\")\n)', 'time_series', 16, 0, 4, 5),
+ HostsOverviewSingleStatPanel(
+ 'bytes', 'Network Load', 'Total send/receive network load across all hosts in the ceph cluster', 'current', 'sum (\n irate(node_network_receive_bytes{instance=~\"($osd_hosts|mon_hosts|mds_hosts|rgw_hosts).*\",device!=\"lo\"}[1m]) or\n irate(node_network_receive_bytes_total{instance=~\"($osd_hosts|mon_hosts|mds_hosts|rgw_hosts).*\",device!=\"lo\"}[1m])\n ) +\nsum (\n irate(node_network_transmit_bytes{instance=~\"($osd_hosts|mon_hosts|mds_hosts|rgw_hosts).*\",device!=\"lo\"}[1m]) or\n irate(node_network_transmit_bytes_total{instance=~\"($osd_hosts|mon_hosts|mds_hosts|rgw_hosts).*\",device!=\"lo\"}[1m])\n )', 'time_series', 20, 0, 4, 5),
+ HostsOverviewGraphPanel(
+ 'CPU Busy - Top 10 Hosts', 'Show the top 10 busiest hosts by cpu', 'percent', 'topk(10,100 * ( 1 - (\n avg by(instance) \n (irate(node_cpu_seconds_total{mode=\'idle\',instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"}[1m]) or\n irate(node_cpu{mode=\'idle\',instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\"}[1m]))\n )\n )\n)', '{{instance}}', 0, 5, 12, 9),
+ HostsOverviewGraphPanel(
+ 'Network Load - Top 10 Hosts', 'Top 10 hosts by network load', 'Bps', 'topk(10, (sum by(instance) (\n (\n irate(node_network_receive_bytes{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\",device!=\"lo\"}[1m]) or\n irate(node_network_receive_bytes_total{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\",device!=\"lo\"}[1m])\n ) +\n (\n irate(node_network_transmit_bytes{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\",device!=\"lo\"}[1m]) or\n irate(node_network_transmit_bytes_total{instance=~\"($osd_hosts|$mon_hosts|$mds_hosts|$rgw_hosts).*\",device!=\"lo\"}[1m])\n ))\n )\n)', '{{instance}}', 12, 5, 12, 9),
+ ])
+}
+{
+ "host-details.json":
+ local HostDetailsSingleStatPanel(format, title, description, valueName, expr, targetFormat, x, y, w, h) =
+ addSingelStatSchema('$datasource', format, title, description, valueName, false, 100, false, false, '')
+ .addTarget(addTargetSchema(expr, 1, targetFormat, '')) + {gridPos: {x: x, y: y, w: w, h: h}};
+
+ local HostDetailsGraphPanel(alias, title, description, nullPointMode, formatY1, labelY1, expr, legendFormat, x, y, w, h) =
+ graphPanelSchema(alias, title, description, nullPointMode, false, formatY1, 'short', labelY1, null, null, 1, '$datasource')
+ .addTargets(
+ [addTargetSchema(expr, 1, 'time_series', legendFormat)]) + {gridPos: {x: x, y: y, w: w, h: h}};
+
+ dashboardSchema(
+ 'Host Details', '', 'rtOg0AiWz', 'now-1h', '10s', 16, ['overview'], '', {refresh_intervals:['5s','10s','30s','1m','5m','15m','30m','1h','2h','1d'],time_options:['5m','15m','1h','6h','12h','24h','2d','7d','30d']}
+ )
+ .addRequired(
+ type='grafana', id='grafana', name='Grafana', version='5.3.2'
+ )
+ .addRequired(
+ type='panel', id='graph', name='Graph', version='5.0.0'
+ )
+ .addRequired(
+ type='panel', id='singlestat', name='Singlestat', version='5.0.0'
+ )
+ .addAnnotation(
+ addAnnotationSchema(
+ 1, '-- Grafana --', true, true, 'rgba(0, 211, 255, 1)', 'Annotations & Alerts', 'dashboard')
+ )
+ .addTemplate(
+ g.template.datasource('datasource', 'prometheus', 'default', label='Data Source')
+ )
+ .addTemplate(
+ addTemplateSchema('ceph_hosts', '$datasource', 'label_values(node_scrape_collector_success, instance) ', 1, false, 3, 'Hostname', '([^.:]*).*')
+ )
+ .addPanels([
+ addRowSchema(false, true, '$ceph_hosts System Overview') + {gridPos: {x: 0, y: 0, w: 24, h: 1}},
+ HostDetailsSingleStatPanel(
+ 'none', 'OSDs', '', 'current', 'count(sum by (ceph_daemon) (ceph_osd_metadata{hostname=\'$ceph_hosts\'}))', 'time_series', 0, 1, 3, 5
+ ),
+ HostDetailsGraphPanel(
+ {"interrupt": "#447EBC","steal": "#6D1F62","system": "#890F02","user": "#3F6833","wait": "#C15C17"},'CPU Utilization', 'Shows the CPU breakdown. When multiple servers are selected, only the first host\'s cpu data is shown', 'null', 'percent', '% Utilization', 'sum by (mode) (\n irate(node_cpu{instance=~\"($ceph_hosts)([\\\\.:].*)?\", mode=~\"(irq|nice|softirq|steal|system|user|iowait)\"}[1m]) or\n irate(node_cpu_seconds_total{instance=~\"($ceph_hosts)([\\\\.:].*)?\", mode=~\"(irq|nice|softirq|steal|system|user|iowait)\"}[1m])\n) / scalar(\n sum(irate(node_cpu{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[1m]) or\n irate(node_cpu_seconds_total{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[1m]))\n) * 100', '{{mode}}', 3, 1, 6, 10
+ ),
+ HostDetailsGraphPanel(
+ {"Available": "#508642","Free": "#508642","Total": "#bf1b00","Used": "#bf1b00","total": "#bf1b00","used": "#0a50a1"},'RAM Usage', '', 'null', 'bytes', 'RAM used', 'node_memory_MemFree{instance=~\"$ceph_hosts([\\\\.:].*)?\"} or node_memory_MemFree_bytes{instance=~\"$ceph_hosts([\\\\.:].*)?\"} ', 'Free', 9, 1, 6, 10)
+ .addTargets(
+ [addTargetSchema('node_memory_MemTotal{instance=~\"$ceph_hosts([\\\\.:].*)?\"} or node_memory_MemTotal_bytes{instance=~\"$ceph_hosts([\\\\.:].*)?\"} ', 1, 'time_series', 'total'),
+ addTargetSchema('(node_memory_Cached{instance=~\"$ceph_hosts([\\\\.:].*)?\"} or node_memory_Cached_bytes{instance=~\"$ceph_hosts([\\\\.:].*)?\"}) + \n(node_memory_Buffers{instance=~\"$ceph_hosts([\\\\.:].*)?\"} or node_memory_Buffers_bytes{instance=~\"$ceph_hosts([\\\\.:].*)?\"}) +\n(node_memory_Slab{instance=~\"$ceph_hosts([\\\\.:].*)?\"} or node_memory_Slab_bytes{instance=~\"$ceph_hosts([\\\\.:].*)?\"}) \n', 1, 'time_series', 'buffers/cache'),
+ addTargetSchema('(node_memory_MemTotal{instance=~\"$ceph_hosts([\\\\.:].*)?\"} or node_memory_MemTotal_bytes{instance=~\"$ceph_hosts([\\\\.:].*)?\"})- (\n (node_memory_MemFree{instance=~\"$ceph_hosts([\\\\.:].*)?\"} or node_memory_MemFree_bytes{instance=~\"$ceph_hosts([\\\\.:].*)?\"}) + \n (node_memory_Cached{instance=~\"$ceph_hosts([\\\\.:].*)?\"} or node_memory_Cached_bytes{instance=~\"$ceph_hosts([\\\\.:].*)?\"}) + \n (node_memory_Buffers{instance=~\"$ceph_hosts([\\\\.:].*)?\"} or node_memory_Buffers_bytes{instance=~\"$ceph_hosts([\\\\.:].*)?\"}) +\n (node_memory_Slab{instance=~\"$ceph_hosts([\\\\.:].*)?\"} or node_memory_Slab_bytes{instance=~\"$ceph_hosts([\\\\.:].*)?\"})\n )\n \n', 1, 'time_series', 'used')])
+ .addSeriesOverride({"alias": "total","color": "#bf1b00","fill": 0,"linewidth": 2,"stack": false}
+ ),
+ HostDetailsGraphPanel(
+ {},'Network Load', 'Show the network load (rx,tx) across all interfaces (excluding loopback \'lo\')', 'null', 'decbytes', 'Send (-) / Receive (+)', 'sum by (device) (\n irate(node_network_receive_bytes{instance=~\"($ceph_hosts)([\\\\.:].*)?\",device!=\"lo\"}[1m]) or \n irate(node_network_receive_bytes_total{instance=~\"($ceph_hosts)([\\\\.:].*)?\",device!=\"lo\"}[1m])\n)', '{{device}}.rx', 15, 1, 6, 10)
+ .addTargets(
+ [addTargetSchema('sum by (device) (\n irate(node_network_transmit_bytes{instance=~\"($ceph_hosts)([\\\\.:].*)?\",device!=\"lo\"}[1m]) or\n irate(node_network_transmit_bytes_total{instance=~\"($ceph_hosts)([\\\\.:].*)?\",device!=\"lo\"}[1m])\n)', 1, 'time_series', '{{device}}.tx')])
+ .addSeriesOverride({"alias": "/.*tx/","transform": "negative-Y"}
+ ),
+ HostDetailsGraphPanel(
+ {},'Network drop rate', '', 'null', 'pps', 'Send (-) / Receive (+)', 'irate(node_network_receive_drop{instance=~\"$ceph_hosts([\\\\.:].*)?\"}[1m]) or irate(node_network_receive_drop_total{instance=~\"$ceph_hosts([\\\\.:].*)?\"}[1m])', '{{device}}.rx', 21, 1, 3, 5)
+ .addTargets(
+ [addTargetSchema('irate(node_network_transmit_drop{instance=~\"$ceph_hosts([\\\\.:].*)?\"}[1m]) or irate(node_network_transmit_drop_total{instance=~\"$ceph_hosts([\\\\.:].*)?\"}[1m])', 1, 'time_series', '{{device}}.tx')])
+ .addSeriesOverride({"alias": "/.*tx/","transform": "negative-Y"}
+ ),
+ HostDetailsSingleStatPanel(
+ 'bytes', 'Raw Capacity', 'Each OSD consists of a Journal/WAL partition and a data partition. The RAW Capacity shown is the sum of the data partitions across all OSDs on the selected OSD hosts.', 'current', 'sum(ceph_osd_stat_bytes and on (ceph_daemon) ceph_disk_occupation{instance=~\"($ceph_hosts)([\\\\.:].*)?\"})', 'time_series', 0, 6, 3, 5
+ ),
+ HostDetailsGraphPanel(
+ {},'Network error rate', '', 'null', 'pps', 'Send (-) / Receive (+)', 'irate(node_network_receive_errs{instance=~\"$ceph_hosts([\\\\.:].*)?\"}[1m]) or irate(node_network_receive_errs_total{instance=~\"$ceph_hosts([\\\\.:].*)?\"}[1m])', '{{device}}.rx', 21, 6, 3, 5)
+ .addTargets(
+ [addTargetSchema('irate(node_network_transmit_errs{instance=~\"$ceph_hosts([\\\\.:].*)?\"}[1m]) or irate(node_network_transmit_errs_total{instance=~\"$ceph_hosts([\\\\.:].*)?\"}[1m])', 1, 'time_series', '{{device}}.tx')])
+ .addSeriesOverride({"alias": "/.*tx/","transform": "negative-Y"}
+ ),
+ addRowSchema(false, true, 'OSD Disk Performance Statistics') + {gridPos: {x: 0, y: 11, w: 24, h: 1}},
+ HostDetailsGraphPanel(
+ {},'$ceph_hosts Disk IOPS', 'For any OSD devices on the host, this chart shows the iops per physical device. Each device is shown by it\'s name and corresponding OSD id value', 'connected', 'ops', 'Read (-) / Write (+)', 'label_replace(\n (\n irate(node_disk_writes_completed{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m]) or\n irate(node_disk_writes_completed_total{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m])\n ),\n \"instance\",\n \"$1\",\n \"instance\",\n \"([^:.]*).*\"\n)\n* on(instance, device, ceph_daemon) group_left\n label_replace(\n label_replace(\n ceph_disk_occupation,\n \"device\",\n \"$1\",\n \"device\",\n \"/dev/(.*)\"\n ),\n \"instance\",\n \"$1\",\n \"instance\",\n \"([^:.]*).*\"\n )', '{{device}}({{ceph_daemon}}) writes', 0, 12, 11, 9)
+ .addTargets(
+ [addTargetSchema('label_replace(\n (irate(node_disk_reads_completed{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m]) or irate(node_disk_reads_completed_total{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m])),\n \"instance\",\n \"$1\",\n \"instance\",\n \"([^:.]*).*\"\n)\n* on(instance, device, ceph_daemon) group_left\n label_replace(\n label_replace(\n ceph_disk_occupation,\n \"device\",\n \"$1\",\n \"device\",\n \"/dev/(.*)\"\n ),\n \"instance\",\n \"$1\",\n \"instance\",\n \"([^:.]*).*\"\n )', 1, 'time_series', '{{device}}({{ceph_daemon}}) reads')])
+ .addSeriesOverride({"alias": "/.*reads/","transform": "negative-Y"}
+ ),
+ HostDetailsGraphPanel(
+ {},'$ceph_hosts Throughput by Disk', 'For OSD hosts, this chart shows the disk bandwidth (read bytes/sec + write bytes/sec) of the physical OSD device. Each device is shown by device name, and corresponding OSD id', 'connected', 'Bps', 'Read (-) / Write (+)', 'label_replace((irate(node_disk_bytes_written{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m]) or irate(node_disk_written_bytes_total{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m])), \"instance\", \"$1\", \"instance\", \"([^:.]*).*\") * on(instance, device, ceph_daemon) group_left label_replace(label_replace(ceph_disk_occupation, \"device\", \"$1\", \"device\", \"/dev/(.*)\"), \"instance\", \"$1\", \"instance\", \"([^:.]*).*\")', '{{device}}({{ceph_daemon}}) write', 12, 12, 11, 9)
+ .addTargets(
+ [addTargetSchema('label_replace((irate(node_disk_bytes_read{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m]) or irate(node_disk_read_bytes_total{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m])), \"instance\", \"$1\", \"instance\", \"([^:.]*).*\") * on(instance, device, ceph_daemon) group_left label_replace(label_replace(ceph_disk_occupation, \"device\", \"$1\", \"device\", \"/dev/(.*)\"), \"instance\", \"$1\", \"instance\", \"([^:.]*).*\")', 1, 'time_series', '{{device}}({{ceph_daemon}}) read')])
+ .addSeriesOverride({"alias": "/.*read/","transform": "negative-Y"}
+ ),
+ HostDetailsGraphPanel(
+ {},'$ceph_hosts Disk Latency', 'For OSD hosts, this chart shows the latency at the physical drive. Each drive is shown by device name, with it\'s corresponding OSD id', 'null as zero', 's', '', 'max by(instance,device) (label_replace((irate(node_disk_write_time_seconds_total{ instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m]) ) / clamp_min(irate(node_disk_writes_completed_total{ instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m]), 0.001) or (irate(node_disk_read_time_seconds_total{ instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m]) ) / clamp_min(irate(node_disk_reads_completed_total{ instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m]), 0.001), \"instance\", \"$1\", \"instance\", \"([^:.]*).*\")) * on(instance, device, ceph_daemon) group_left label_replace(label_replace(ceph_disk_occupation{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}, \"device\", \"$1\", \"device\", \"/dev/(.*)\"), \"instance\", \"$1\", \"instance\", \"([^:.]*).*\")', '{{device}}({{ceph_daemon}})', 0, 21, 11, 9
+ ),
+ HostDetailsGraphPanel(
+ {},'$ceph_hosts Disk utilization', 'Show disk utilization % (util) of any OSD devices on the host by the physical device name and associated OSD id.', 'connected', 'percent', '%Util', 'label_replace(((irate(node_disk_io_time_ms{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m]) / 10 ) or irate(node_disk_io_time_seconds_total{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}[5m]) * 100), \"instance\", \"$1\", \"instance\", \"([^:.]*).*\") * on(instance, device, ceph_daemon) group_left label_replace(label_replace(ceph_disk_occupation{instance=~\"($ceph_hosts)([\\\\.:].*)?\"}, \"device\", \"$1\", \"device\", \"/dev/(.*)\"), \"instance\", \"$1\", \"instance\", \"([^:.]*).*\")', '{{device}}({{ceph_daemon}})', 12, 21, 11, 9
+ )
+ ])
+}
{
"radosgw-sync-overview.json":
local RgwSyncOverviewPanel(title, formatY1, labelY1, rgwMetric, x, y, w, h) =
projects / ceph.git / commitdiff