429 lines
14 KiB
JSON
429 lines
14 KiB
JSON
{
|
|
"uid": "incus-capacity-planning",
|
|
"title": "Capacity Planning",
|
|
"description": "CPU/memory headroom per node, resource trends, disk space forecast with predict_linear, instance count trends.",
|
|
"tags": ["incus", "capacity", "operations"],
|
|
"timezone": "browser",
|
|
"schemaVersion": 39,
|
|
"refresh": "5m",
|
|
"time": { "from": "now-7d", "to": "now" },
|
|
"fiscalYearStartMonth": 0,
|
|
"liveNow": false,
|
|
"editable": true,
|
|
"graphTooltip": 1,
|
|
"links": [
|
|
{
|
|
"asDropdown": true,
|
|
"icon": "external link",
|
|
"includeVars": true,
|
|
"keepTime": true,
|
|
"tags": ["incus"],
|
|
"targetBlank": false,
|
|
"title": "Incus Dashboards",
|
|
"tooltip": "Navigate to other Incus dashboards",
|
|
"type": "dashboards"
|
|
}
|
|
],
|
|
"templating": { "list": [] },
|
|
"annotations": {
|
|
"list": [
|
|
{
|
|
"builtIn": 1,
|
|
"datasource": { "type": "grafana", "uid": "-- Grafana --" },
|
|
"enable": true,
|
|
"hide": true,
|
|
"iconColor": "rgba(0, 211, 255, 1)",
|
|
"name": "Annotations & Alerts",
|
|
"type": "dashboard"
|
|
}
|
|
]
|
|
},
|
|
"panels": [
|
|
{
|
|
"type": "row",
|
|
"title": "Current Headroom",
|
|
"collapsed": false,
|
|
"gridPos": { "h": 1, "w": 24, "x": 0, "y": 0 },
|
|
"id": 1,
|
|
"panels": []
|
|
},
|
|
{
|
|
"type": "bargauge",
|
|
"title": "CPU Headroom per Node",
|
|
"description": "Available (idle) CPU percentage per host node. Higher values mean more headroom for additional workloads.",
|
|
"gridPos": { "h": 6, "w": 12, "x": 0, "y": 1 },
|
|
"id": 2,
|
|
"datasource": { "type": "prometheus", "uid": "${DS_PROMETHEUS}" },
|
|
"targets": [
|
|
{
|
|
"expr": "avg by (instance) (rate(node_cpu_seconds_total{mode=\"idle\"}[$__rate_interval])) * 100",
|
|
"legendFormat": "{{instance}}",
|
|
"refId": "A"
|
|
}
|
|
],
|
|
"fieldConfig": {
|
|
"defaults": {
|
|
"thresholds": {
|
|
"mode": "absolute",
|
|
"steps": [
|
|
{ "color": "red", "value": null },
|
|
{ "color": "orange", "value": 20 },
|
|
{ "color": "yellow", "value": 40 },
|
|
{ "color": "green", "value": 60 }
|
|
]
|
|
},
|
|
"unit": "percent",
|
|
"min": 0,
|
|
"max": 100,
|
|
"color": { "mode": "thresholds" }
|
|
},
|
|
"overrides": []
|
|
},
|
|
"options": {
|
|
"reduceOptions": { "calcs": ["lastNotNull"], "fields": "", "values": false },
|
|
"orientation": "horizontal",
|
|
"displayMode": "gradient",
|
|
"showUnfilled": true,
|
|
"minVizWidth": 8,
|
|
"minVizHeight": 16,
|
|
"valueMode": "color"
|
|
}
|
|
},
|
|
{
|
|
"type": "bargauge",
|
|
"title": "Memory Headroom per Node",
|
|
"description": "Available memory percentage per host node. Higher values mean more headroom for additional workloads.",
|
|
"gridPos": { "h": 6, "w": 12, "x": 12, "y": 1 },
|
|
"id": 3,
|
|
"datasource": { "type": "prometheus", "uid": "${DS_PROMETHEUS}" },
|
|
"targets": [
|
|
{
|
|
"expr": "(node_memory_MemAvailable_bytes / node_memory_MemTotal_bytes) * 100",
|
|
"legendFormat": "{{instance}}",
|
|
"refId": "A"
|
|
}
|
|
],
|
|
"fieldConfig": {
|
|
"defaults": {
|
|
"thresholds": {
|
|
"mode": "absolute",
|
|
"steps": [
|
|
{ "color": "red", "value": null },
|
|
{ "color": "orange", "value": 20 },
|
|
{ "color": "yellow", "value": 40 },
|
|
{ "color": "green", "value": 60 }
|
|
]
|
|
},
|
|
"unit": "percent",
|
|
"min": 0,
|
|
"max": 100,
|
|
"color": { "mode": "thresholds" }
|
|
},
|
|
"overrides": []
|
|
},
|
|
"options": {
|
|
"reduceOptions": { "calcs": ["lastNotNull"], "fields": "", "values": false },
|
|
"orientation": "horizontal",
|
|
"displayMode": "gradient",
|
|
"showUnfilled": true,
|
|
"minVizWidth": 8,
|
|
"minVizHeight": 16,
|
|
"valueMode": "color"
|
|
}
|
|
},
|
|
{
|
|
"type": "row",
|
|
"title": "Resource Trends",
|
|
"collapsed": false,
|
|
"gridPos": { "h": 1, "w": 24, "x": 0, "y": 7 },
|
|
"id": 4,
|
|
"panels": []
|
|
},
|
|
{
|
|
"type": "timeseries",
|
|
"title": "CPU Usage Trend (7-day)",
|
|
"description": "CPU utilization percentage per host node over the past 7 days. Helps identify sustained usage patterns and growth trends.",
|
|
"gridPos": { "h": 8, "w": 12, "x": 0, "y": 8 },
|
|
"id": 5,
|
|
"datasource": { "type": "prometheus", "uid": "${DS_PROMETHEUS}" },
|
|
"targets": [
|
|
{
|
|
"expr": "(1 - avg by (instance) (rate(node_cpu_seconds_total{mode=\"idle\"}[5m]))) * 100",
|
|
"legendFormat": "{{instance}}",
|
|
"refId": "A"
|
|
}
|
|
],
|
|
"fieldConfig": {
|
|
"defaults": {
|
|
"custom": {
|
|
"drawStyle": "line",
|
|
"lineInterpolation": "smooth",
|
|
"lineWidth": 2,
|
|
"fillOpacity": 25,
|
|
"gradientMode": "scheme",
|
|
"spanNulls": false,
|
|
"showPoints": "never",
|
|
"pointSize": 5,
|
|
"stacking": { "mode": "none", "group": "A" },
|
|
"axisPlacement": "auto",
|
|
"axisLabel": "",
|
|
"scaleDistribution": { "type": "linear" },
|
|
"thresholdsStyle": { "mode": "off" }
|
|
},
|
|
"unit": "percent",
|
|
"min": 0,
|
|
"max": 100,
|
|
"color": { "mode": "continuous-GrYlRd" },
|
|
"thresholds": {
|
|
"mode": "absolute",
|
|
"steps": [
|
|
{ "color": "green", "value": null },
|
|
{ "color": "yellow", "value": 70 },
|
|
{ "color": "red", "value": 90 }
|
|
]
|
|
}
|
|
},
|
|
"overrides": []
|
|
},
|
|
"options": {
|
|
"tooltip": { "mode": "multi", "sort": "desc" },
|
|
"legend": { "displayMode": "table", "placement": "right", "calcs": ["mean", "max", "lastNotNull"] }
|
|
}
|
|
},
|
|
{
|
|
"type": "timeseries",
|
|
"title": "Memory Usage Trend (7-day)",
|
|
"description": "Memory utilization percentage per host node over the past 7 days. Helps identify sustained usage patterns and growth trends.",
|
|
"gridPos": { "h": 8, "w": 12, "x": 12, "y": 8 },
|
|
"id": 6,
|
|
"datasource": { "type": "prometheus", "uid": "${DS_PROMETHEUS}" },
|
|
"targets": [
|
|
{
|
|
"expr": "(1 - node_memory_MemAvailable_bytes / node_memory_MemTotal_bytes) * 100",
|
|
"legendFormat": "{{instance}}",
|
|
"refId": "A"
|
|
}
|
|
],
|
|
"fieldConfig": {
|
|
"defaults": {
|
|
"custom": {
|
|
"drawStyle": "line",
|
|
"lineInterpolation": "smooth",
|
|
"lineWidth": 2,
|
|
"fillOpacity": 25,
|
|
"gradientMode": "scheme",
|
|
"spanNulls": false,
|
|
"showPoints": "never",
|
|
"pointSize": 5,
|
|
"stacking": { "mode": "none", "group": "A" },
|
|
"axisPlacement": "auto",
|
|
"axisLabel": "",
|
|
"scaleDistribution": { "type": "linear" },
|
|
"thresholdsStyle": { "mode": "off" }
|
|
},
|
|
"unit": "percent",
|
|
"min": 0,
|
|
"max": 100,
|
|
"color": { "mode": "continuous-GrYlRd" },
|
|
"thresholds": {
|
|
"mode": "absolute",
|
|
"steps": [
|
|
{ "color": "green", "value": null },
|
|
{ "color": "yellow", "value": 70 },
|
|
{ "color": "red", "value": 90 }
|
|
]
|
|
}
|
|
},
|
|
"overrides": []
|
|
},
|
|
"options": {
|
|
"tooltip": { "mode": "multi", "sort": "desc" },
|
|
"legend": { "displayMode": "table", "placement": "right", "calcs": ["mean", "max", "lastNotNull"] }
|
|
}
|
|
},
|
|
{
|
|
"type": "row",
|
|
"title": "Disk Space Forecast",
|
|
"collapsed": false,
|
|
"gridPos": { "h": 1, "w": 24, "x": 0, "y": 16 },
|
|
"id": 7,
|
|
"panels": []
|
|
},
|
|
{
|
|
"type": "timeseries",
|
|
"title": "Disk Space Forecast (30-day projection)",
|
|
"description": "Current available disk space on root filesystem with a 30-day linear projection based on the past 7 days of data. Dashed lines show the forecast; if a forecast line drops below zero, the disk will fill up within 30 days at current consumption rates.",
|
|
"gridPos": { "h": 8, "w": 24, "x": 0, "y": 17 },
|
|
"id": 8,
|
|
"datasource": { "type": "prometheus", "uid": "${DS_PROMETHEUS}" },
|
|
"targets": [
|
|
{
|
|
"expr": "node_filesystem_avail_bytes{mountpoint=\"/\"}",
|
|
"legendFormat": "{{instance}} actual",
|
|
"refId": "A"
|
|
},
|
|
{
|
|
"expr": "predict_linear(node_filesystem_avail_bytes{mountpoint=\"/\"}[7d], 30*24*3600)",
|
|
"legendFormat": "{{instance}} forecast",
|
|
"refId": "B"
|
|
}
|
|
],
|
|
"fieldConfig": {
|
|
"defaults": {
|
|
"custom": {
|
|
"drawStyle": "line",
|
|
"lineInterpolation": "smooth",
|
|
"lineWidth": 2,
|
|
"fillOpacity": 15,
|
|
"gradientMode": "none",
|
|
"spanNulls": false,
|
|
"showPoints": "never",
|
|
"pointSize": 5,
|
|
"stacking": { "mode": "none", "group": "A" },
|
|
"axisPlacement": "auto",
|
|
"axisLabel": "",
|
|
"scaleDistribution": { "type": "linear" },
|
|
"thresholdsStyle": { "mode": "off" }
|
|
},
|
|
"unit": "bytes",
|
|
"color": { "mode": "palette-classic" },
|
|
"thresholds": {
|
|
"mode": "absolute",
|
|
"steps": [{ "color": "green", "value": null }]
|
|
}
|
|
},
|
|
"overrides": [
|
|
{
|
|
"matcher": { "id": "byRegexp", "options": ".*forecast$" },
|
|
"properties": [
|
|
{ "id": "custom.lineStyle", "value": { "fill": "dash", "dash": [10, 10] } },
|
|
{ "id": "custom.fillOpacity", "value": 0 }
|
|
]
|
|
}
|
|
]
|
|
},
|
|
"options": {
|
|
"tooltip": { "mode": "multi", "sort": "desc" },
|
|
"legend": { "displayMode": "table", "placement": "right", "calcs": ["mean", "max", "lastNotNull"] }
|
|
}
|
|
},
|
|
{
|
|
"type": "row",
|
|
"title": "Instance Trends",
|
|
"collapsed": false,
|
|
"gridPos": { "h": 1, "w": 24, "x": 0, "y": 25 },
|
|
"id": 9,
|
|
"panels": []
|
|
},
|
|
{
|
|
"type": "timeseries",
|
|
"title": "Instance Count Over Time",
|
|
"description": "Total number of running Incus instances over time. Tracks growth and churn in the cluster workload count.",
|
|
"gridPos": { "h": 8, "w": 12, "x": 0, "y": 26 },
|
|
"id": 10,
|
|
"datasource": { "type": "prometheus", "uid": "${DS_PROMETHEUS}" },
|
|
"targets": [
|
|
{
|
|
"expr": "count(incus_uptime_seconds)",
|
|
"legendFormat": "Instances",
|
|
"refId": "A"
|
|
}
|
|
],
|
|
"fieldConfig": {
|
|
"defaults": {
|
|
"custom": {
|
|
"drawStyle": "line",
|
|
"lineInterpolation": "smooth",
|
|
"lineWidth": 2,
|
|
"fillOpacity": 25,
|
|
"gradientMode": "none",
|
|
"spanNulls": false,
|
|
"showPoints": "never",
|
|
"pointSize": 5,
|
|
"stacking": { "mode": "none", "group": "A" },
|
|
"axisPlacement": "auto",
|
|
"axisLabel": "",
|
|
"scaleDistribution": { "type": "linear" },
|
|
"thresholdsStyle": { "mode": "off" }
|
|
},
|
|
"unit": "short",
|
|
"color": { "mode": "palette-classic" },
|
|
"thresholds": {
|
|
"mode": "absolute",
|
|
"steps": [{ "color": "green", "value": null }]
|
|
}
|
|
},
|
|
"overrides": []
|
|
},
|
|
"options": {
|
|
"tooltip": { "mode": "multi", "sort": "desc" },
|
|
"legend": { "displayMode": "table", "placement": "right", "calcs": ["mean", "max", "lastNotNull"] }
|
|
}
|
|
},
|
|
{
|
|
"type": "timeseries",
|
|
"title": "Memory Allocation vs Usage per Instance",
|
|
"description": "Compares total allocated memory (MemTotal) against actual resident memory (RSS) for each Incus instance. A large gap between allocated and used indicates overprovisioning; converging lines indicate an instance may need more memory.",
|
|
"gridPos": { "h": 8, "w": 12, "x": 12, "y": 26 },
|
|
"id": 11,
|
|
"datasource": { "type": "prometheus", "uid": "${DS_PROMETHEUS}" },
|
|
"targets": [
|
|
{
|
|
"expr": "incus_memory_MemTotal_bytes",
|
|
"legendFormat": "{{name}} allocated",
|
|
"refId": "A"
|
|
},
|
|
{
|
|
"expr": "incus_memory_RSS_bytes",
|
|
"legendFormat": "{{name}} used (RSS)",
|
|
"refId": "B"
|
|
}
|
|
],
|
|
"fieldConfig": {
|
|
"defaults": {
|
|
"custom": {
|
|
"drawStyle": "line",
|
|
"lineInterpolation": "smooth",
|
|
"lineWidth": 1,
|
|
"fillOpacity": 0,
|
|
"gradientMode": "none",
|
|
"spanNulls": false,
|
|
"showPoints": "never",
|
|
"pointSize": 5,
|
|
"stacking": { "mode": "none", "group": "A" },
|
|
"axisPlacement": "auto",
|
|
"axisLabel": "",
|
|
"scaleDistribution": { "type": "linear" },
|
|
"thresholdsStyle": { "mode": "off" }
|
|
},
|
|
"unit": "bytes",
|
|
"color": { "mode": "palette-classic" },
|
|
"thresholds": {
|
|
"mode": "absolute",
|
|
"steps": [{ "color": "green", "value": null }]
|
|
}
|
|
},
|
|
"overrides": [
|
|
{
|
|
"matcher": { "id": "byRegexp", "options": ".*allocated$" },
|
|
"properties": [
|
|
{ "id": "custom.lineStyle", "value": { "fill": "dash", "dash": [10, 10] } },
|
|
{ "id": "custom.lineWidth", "value": 2 }
|
|
]
|
|
},
|
|
{
|
|
"matcher": { "id": "byRegexp", "options": ".*used \\(RSS\\)$" },
|
|
"properties": [
|
|
{ "id": "custom.fillOpacity", "value": 15 }
|
|
]
|
|
}
|
|
]
|
|
},
|
|
"options": {
|
|
"tooltip": { "mode": "multi", "sort": "desc" },
|
|
"legend": { "displayMode": "table", "placement": "right", "calcs": ["mean", "max", "lastNotNull"] }
|
|
}
|
|
}
|
|
]
|
|
}
|