scott
87015b19b1
Single-pane dashboard combining app status (UP/DOWN, response time, 7-day uptime, SSL expiry), pod health (status counts, restarts, not ready pods, namespace distribution), cluster resources (CPU/memory/disk gauges, node/pod/PVC counts), and a firing alerts table. Auto-deploys via the existing grafana-dashboards ArgoCD app. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
567 lines
16 KiB
JSON
567 lines
16 KiB
JSON
{
|
|
"description": "Single-pane view of application health, pod status, and cluster resource utilization",
|
|
"editable": true,
|
|
"graphTooltip": 1,
|
|
"id": null,
|
|
"links": [
|
|
{
|
|
"icon": "external link",
|
|
"title": "App Uptime Detail",
|
|
"url": "/d/app-uptime-dashboard",
|
|
"type": "link"
|
|
}
|
|
],
|
|
"panels": [
|
|
{
|
|
"title": "Application Status",
|
|
"type": "stat",
|
|
"gridPos": { "h": 4, "w": 24, "x": 0, "y": 0 },
|
|
"datasource": { "type": "prometheus" },
|
|
"fieldConfig": {
|
|
"defaults": {
|
|
"mappings": [
|
|
{ "options": { "0": { "color": "red", "text": "DOWN" } }, "type": "value" },
|
|
{ "options": { "1": { "color": "green", "text": "UP" } }, "type": "value" }
|
|
],
|
|
"thresholds": {
|
|
"mode": "absolute",
|
|
"steps": [
|
|
{ "color": "red", "value": null },
|
|
{ "color": "green", "value": 1 }
|
|
]
|
|
}
|
|
}
|
|
},
|
|
"options": {
|
|
"colorMode": "background",
|
|
"graphMode": "none",
|
|
"justifyMode": "center",
|
|
"reduceOptions": { "calcs": ["lastNotNull"], "fields": "", "values": false },
|
|
"textMode": "auto",
|
|
"orientation": "horizontal"
|
|
},
|
|
"targets": [
|
|
{
|
|
"expr": "probe_success{job=\"blackbox-http\"}",
|
|
"legendFormat": "{{ app }}",
|
|
"refId": "A"
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"title": "Response Time",
|
|
"type": "stat",
|
|
"gridPos": { "h": 4, "w": 24, "x": 0, "y": 4 },
|
|
"datasource": { "type": "prometheus" },
|
|
"fieldConfig": {
|
|
"defaults": {
|
|
"unit": "s",
|
|
"decimals": 2,
|
|
"thresholds": {
|
|
"mode": "absolute",
|
|
"steps": [
|
|
{ "color": "green", "value": null },
|
|
{ "color": "yellow", "value": 1 },
|
|
{ "color": "orange", "value": 3 },
|
|
{ "color": "red", "value": 5 }
|
|
]
|
|
}
|
|
}
|
|
},
|
|
"options": {
|
|
"colorMode": "background",
|
|
"graphMode": "area",
|
|
"justifyMode": "center",
|
|
"reduceOptions": { "calcs": ["lastNotNull"], "fields": "", "values": false },
|
|
"textMode": "auto",
|
|
"orientation": "horizontal"
|
|
},
|
|
"targets": [
|
|
{
|
|
"expr": "probe_http_duration_seconds{job=\"blackbox-http\"}",
|
|
"legendFormat": "{{ app }}",
|
|
"refId": "A"
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"title": "Uptime (7d)",
|
|
"type": "gauge",
|
|
"gridPos": { "h": 5, "w": 12, "x": 0, "y": 8 },
|
|
"datasource": { "type": "prometheus" },
|
|
"fieldConfig": {
|
|
"defaults": {
|
|
"decimals": 3,
|
|
"max": 100,
|
|
"min": 0,
|
|
"unit": "percent",
|
|
"thresholds": {
|
|
"mode": "absolute",
|
|
"steps": [
|
|
{ "color": "red", "value": null },
|
|
{ "color": "orange", "value": 95 },
|
|
{ "color": "yellow", "value": 99 },
|
|
{ "color": "green", "value": 99.9 }
|
|
]
|
|
}
|
|
}
|
|
},
|
|
"options": {
|
|
"reduceOptions": { "calcs": ["lastNotNull"], "fields": "", "values": false },
|
|
"showThresholdLabels": false,
|
|
"showThresholdMarkers": true
|
|
},
|
|
"targets": [
|
|
{
|
|
"expr": "avg_over_time(probe_success{job=\"blackbox-http\"}[7d]) * 100",
|
|
"legendFormat": "{{ app }}",
|
|
"refId": "A"
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"title": "SSL Certificate Expiry",
|
|
"type": "stat",
|
|
"gridPos": { "h": 5, "w": 12, "x": 12, "y": 8 },
|
|
"datasource": { "type": "prometheus" },
|
|
"fieldConfig": {
|
|
"defaults": {
|
|
"unit": "dtdurations",
|
|
"thresholds": {
|
|
"mode": "absolute",
|
|
"steps": [
|
|
{ "color": "red", "value": null },
|
|
{ "color": "orange", "value": 604800 },
|
|
{ "color": "yellow", "value": 2592000 },
|
|
{ "color": "green", "value": 7776000 }
|
|
]
|
|
}
|
|
}
|
|
},
|
|
"options": {
|
|
"colorMode": "background",
|
|
"graphMode": "none",
|
|
"justifyMode": "center",
|
|
"reduceOptions": { "calcs": ["lastNotNull"], "fields": "", "values": false },
|
|
"textMode": "auto",
|
|
"orientation": "horizontal"
|
|
},
|
|
"targets": [
|
|
{
|
|
"expr": "probe_ssl_earliest_cert_expiry{job=\"blackbox-http\"} - time()",
|
|
"legendFormat": "{{ app }}",
|
|
"refId": "A"
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"title": "",
|
|
"type": "text",
|
|
"gridPos": { "h": 1, "w": 24, "x": 0, "y": 13 },
|
|
"options": {
|
|
"mode": "markdown",
|
|
"content": "### Pod Health"
|
|
}
|
|
},
|
|
{
|
|
"title": "Pods by Status",
|
|
"type": "stat",
|
|
"gridPos": { "h": 4, "w": 8, "x": 0, "y": 14 },
|
|
"datasource": { "type": "prometheus" },
|
|
"fieldConfig": {
|
|
"defaults": {
|
|
"thresholds": {
|
|
"mode": "absolute",
|
|
"steps": [
|
|
{ "color": "green", "value": null }
|
|
]
|
|
}
|
|
},
|
|
"overrides": [
|
|
{
|
|
"matcher": { "id": "byName", "options": "Failed" },
|
|
"properties": [
|
|
{ "id": "thresholds", "value": { "mode": "absolute", "steps": [{ "color": "red", "value": null }] } }
|
|
]
|
|
},
|
|
{
|
|
"matcher": { "id": "byName", "options": "Pending" },
|
|
"properties": [
|
|
{ "id": "thresholds", "value": { "mode": "absolute", "steps": [{ "color": "yellow", "value": null }] } }
|
|
]
|
|
}
|
|
]
|
|
},
|
|
"options": {
|
|
"colorMode": "background",
|
|
"graphMode": "none",
|
|
"justifyMode": "center",
|
|
"reduceOptions": { "calcs": ["lastNotNull"], "fields": "", "values": false },
|
|
"textMode": "auto",
|
|
"orientation": "horizontal"
|
|
},
|
|
"targets": [
|
|
{
|
|
"expr": "sum(kube_pod_status_phase{phase=\"Running\"})",
|
|
"legendFormat": "Running",
|
|
"refId": "A"
|
|
},
|
|
{
|
|
"expr": "sum(kube_pod_status_phase{phase=\"Pending\"})",
|
|
"legendFormat": "Pending",
|
|
"refId": "B"
|
|
},
|
|
{
|
|
"expr": "sum(kube_pod_status_phase{phase=\"Failed\"})",
|
|
"legendFormat": "Failed",
|
|
"refId": "C"
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"title": "Pod Restarts (last 1h)",
|
|
"type": "table",
|
|
"gridPos": { "h": 4, "w": 16, "x": 8, "y": 14 },
|
|
"datasource": { "type": "prometheus" },
|
|
"fieldConfig": {
|
|
"defaults": {},
|
|
"overrides": [
|
|
{
|
|
"matcher": { "id": "byName", "options": "Value" },
|
|
"properties": [
|
|
{ "id": "displayName", "value": "Restarts" },
|
|
{
|
|
"id": "thresholds",
|
|
"value": {
|
|
"mode": "absolute",
|
|
"steps": [
|
|
{ "color": "green", "value": null },
|
|
{ "color": "yellow", "value": 1 },
|
|
{ "color": "red", "value": 5 }
|
|
]
|
|
}
|
|
},
|
|
{ "id": "custom.displayMode", "value": "color-background" }
|
|
]
|
|
}
|
|
]
|
|
},
|
|
"options": {
|
|
"showHeader": true,
|
|
"sortBy": [{ "displayName": "Restarts", "desc": true }]
|
|
},
|
|
"transformations": [
|
|
{ "id": "reduce", "options": { "reducers": ["lastNotNull"] } },
|
|
{ "id": "sortBy", "options": { "fields": {}, "sort": [{ "field": "Value", "desc": true }] } }
|
|
],
|
|
"targets": [
|
|
{
|
|
"expr": "sum by (namespace, pod) (increase(kube_pod_container_status_restarts_total[1h])) > 0",
|
|
"legendFormat": "{{ namespace }}/{{ pod }}",
|
|
"refId": "A",
|
|
"instant": true
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"title": "Not Ready Pods",
|
|
"type": "table",
|
|
"gridPos": { "h": 5, "w": 12, "x": 0, "y": 18 },
|
|
"datasource": { "type": "prometheus" },
|
|
"fieldConfig": {
|
|
"defaults": {},
|
|
"overrides": [
|
|
{
|
|
"matcher": { "id": "byName", "options": "Value" },
|
|
"properties": [
|
|
{ "id": "displayName", "value": "Ready" },
|
|
{ "id": "mappings", "value": [
|
|
{ "options": { "0": { "text": "NOT READY", "color": "red" } }, "type": "value" },
|
|
{ "options": { "1": { "text": "Ready", "color": "green" } }, "type": "value" }
|
|
]},
|
|
{ "id": "custom.displayMode", "value": "color-text" }
|
|
]
|
|
}
|
|
]
|
|
},
|
|
"options": { "showHeader": true },
|
|
"targets": [
|
|
{
|
|
"expr": "kube_pod_status_ready{condition=\"true\"} == 0",
|
|
"legendFormat": "{{ namespace }}/{{ pod }}",
|
|
"refId": "A",
|
|
"instant": true
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"title": "Pods per Namespace",
|
|
"type": "piechart",
|
|
"gridPos": { "h": 5, "w": 12, "x": 12, "y": 18 },
|
|
"datasource": { "type": "prometheus" },
|
|
"fieldConfig": {
|
|
"defaults": {
|
|
"custom": { "hideFrom": { "legend": false, "tooltip": false, "viz": false } }
|
|
}
|
|
},
|
|
"options": {
|
|
"pieType": "donut",
|
|
"reduceOptions": { "calcs": ["lastNotNull"], "fields": "", "values": false },
|
|
"legend": { "displayMode": "table", "placement": "right", "values": ["value"] }
|
|
},
|
|
"targets": [
|
|
{
|
|
"expr": "sum by (namespace) (kube_pod_status_phase{phase=\"Running\"})",
|
|
"legendFormat": "{{ namespace }}",
|
|
"refId": "A"
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"title": "",
|
|
"type": "text",
|
|
"gridPos": { "h": 1, "w": 24, "x": 0, "y": 23 },
|
|
"options": {
|
|
"mode": "markdown",
|
|
"content": "### Cluster Resources"
|
|
}
|
|
},
|
|
{
|
|
"title": "CPU Utilization",
|
|
"type": "gauge",
|
|
"gridPos": { "h": 5, "w": 8, "x": 0, "y": 24 },
|
|
"datasource": { "type": "prometheus" },
|
|
"fieldConfig": {
|
|
"defaults": {
|
|
"decimals": 1,
|
|
"max": 100,
|
|
"min": 0,
|
|
"unit": "percent",
|
|
"thresholds": {
|
|
"mode": "absolute",
|
|
"steps": [
|
|
{ "color": "green", "value": null },
|
|
{ "color": "yellow", "value": 70 },
|
|
{ "color": "orange", "value": 85 },
|
|
{ "color": "red", "value": 95 }
|
|
]
|
|
}
|
|
}
|
|
},
|
|
"options": {
|
|
"reduceOptions": { "calcs": ["lastNotNull"], "fields": "", "values": false },
|
|
"showThresholdLabels": false,
|
|
"showThresholdMarkers": true
|
|
},
|
|
"targets": [
|
|
{
|
|
"expr": "100 - (avg(rate(node_cpu_seconds_total{mode=\"idle\"}[5m])) * 100)",
|
|
"legendFormat": "Cluster CPU",
|
|
"refId": "A"
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"title": "Memory Utilization",
|
|
"type": "gauge",
|
|
"gridPos": { "h": 5, "w": 8, "x": 8, "y": 24 },
|
|
"datasource": { "type": "prometheus" },
|
|
"fieldConfig": {
|
|
"defaults": {
|
|
"decimals": 1,
|
|
"max": 100,
|
|
"min": 0,
|
|
"unit": "percent",
|
|
"thresholds": {
|
|
"mode": "absolute",
|
|
"steps": [
|
|
{ "color": "green", "value": null },
|
|
{ "color": "yellow", "value": 70 },
|
|
{ "color": "orange", "value": 85 },
|
|
{ "color": "red", "value": 95 }
|
|
]
|
|
}
|
|
}
|
|
},
|
|
"options": {
|
|
"reduceOptions": { "calcs": ["lastNotNull"], "fields": "", "values": false },
|
|
"showThresholdLabels": false,
|
|
"showThresholdMarkers": true
|
|
},
|
|
"targets": [
|
|
{
|
|
"expr": "(1 - sum(node_memory_MemAvailable_bytes) / sum(node_memory_MemTotal_bytes)) * 100",
|
|
"legendFormat": "Cluster Memory",
|
|
"refId": "A"
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"title": "Disk Utilization",
|
|
"type": "gauge",
|
|
"gridPos": { "h": 5, "w": 8, "x": 16, "y": 24 },
|
|
"datasource": { "type": "prometheus" },
|
|
"fieldConfig": {
|
|
"defaults": {
|
|
"decimals": 1,
|
|
"max": 100,
|
|
"min": 0,
|
|
"unit": "percent",
|
|
"thresholds": {
|
|
"mode": "absolute",
|
|
"steps": [
|
|
{ "color": "green", "value": null },
|
|
{ "color": "yellow", "value": 70 },
|
|
{ "color": "orange", "value": 85 },
|
|
{ "color": "red", "value": 95 }
|
|
]
|
|
}
|
|
}
|
|
},
|
|
"options": {
|
|
"reduceOptions": { "calcs": ["lastNotNull"], "fields": "", "values": false },
|
|
"showThresholdLabels": false,
|
|
"showThresholdMarkers": true
|
|
},
|
|
"targets": [
|
|
{
|
|
"expr": "(1 - sum(node_filesystem_avail_bytes{mountpoint=\"/\",fstype!=\"tmpfs\"}) / sum(node_filesystem_size_bytes{mountpoint=\"/\",fstype!=\"tmpfs\"})) * 100",
|
|
"legendFormat": "Cluster Disk",
|
|
"refId": "A"
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"title": "Node Count",
|
|
"type": "stat",
|
|
"gridPos": { "h": 3, "w": 8, "x": 0, "y": 29 },
|
|
"datasource": { "type": "prometheus" },
|
|
"fieldConfig": {
|
|
"defaults": {
|
|
"thresholds": {
|
|
"mode": "absolute",
|
|
"steps": [{ "color": "blue", "value": null }]
|
|
}
|
|
}
|
|
},
|
|
"options": {
|
|
"colorMode": "background",
|
|
"graphMode": "none",
|
|
"justifyMode": "center",
|
|
"reduceOptions": { "calcs": ["lastNotNull"], "fields": "", "values": false },
|
|
"textMode": "auto"
|
|
},
|
|
"targets": [
|
|
{
|
|
"expr": "count(kube_node_info)",
|
|
"legendFormat": "Nodes",
|
|
"refId": "A"
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"title": "Total Pods",
|
|
"type": "stat",
|
|
"gridPos": { "h": 3, "w": 8, "x": 8, "y": 29 },
|
|
"datasource": { "type": "prometheus" },
|
|
"fieldConfig": {
|
|
"defaults": {
|
|
"thresholds": {
|
|
"mode": "absolute",
|
|
"steps": [{ "color": "blue", "value": null }]
|
|
}
|
|
}
|
|
},
|
|
"options": {
|
|
"colorMode": "background",
|
|
"graphMode": "none",
|
|
"justifyMode": "center",
|
|
"reduceOptions": { "calcs": ["lastNotNull"], "fields": "", "values": false },
|
|
"textMode": "auto"
|
|
},
|
|
"targets": [
|
|
{
|
|
"expr": "sum(kube_pod_status_phase{phase=\"Running\"})",
|
|
"legendFormat": "Running Pods",
|
|
"refId": "A"
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"title": "PVC Usage",
|
|
"type": "stat",
|
|
"gridPos": { "h": 3, "w": 8, "x": 16, "y": 29 },
|
|
"datasource": { "type": "prometheus" },
|
|
"fieldConfig": {
|
|
"defaults": {
|
|
"thresholds": {
|
|
"mode": "absolute",
|
|
"steps": [{ "color": "blue", "value": null }]
|
|
}
|
|
}
|
|
},
|
|
"options": {
|
|
"colorMode": "background",
|
|
"graphMode": "none",
|
|
"justifyMode": "center",
|
|
"reduceOptions": { "calcs": ["lastNotNull"], "fields": "", "values": false },
|
|
"textMode": "auto"
|
|
},
|
|
"targets": [
|
|
{
|
|
"expr": "count(kube_persistentvolumeclaim_info)",
|
|
"legendFormat": "PVCs",
|
|
"refId": "A"
|
|
}
|
|
]
|
|
},
|
|
{
|
|
"title": "",
|
|
"type": "text",
|
|
"gridPos": { "h": 1, "w": 24, "x": 0, "y": 32 },
|
|
"options": {
|
|
"mode": "markdown",
|
|
"content": "### Recent Alerts"
|
|
}
|
|
},
|
|
{
|
|
"title": "Firing Alerts",
|
|
"type": "table",
|
|
"gridPos": { "h": 6, "w": 24, "x": 0, "y": 33 },
|
|
"datasource": { "type": "prometheus" },
|
|
"fieldConfig": {
|
|
"defaults": {},
|
|
"overrides": [
|
|
{
|
|
"matcher": { "id": "byName", "options": "Value" },
|
|
"properties": [
|
|
{ "id": "displayName", "value": "Active" },
|
|
{ "id": "mappings", "value": [
|
|
{ "options": { "1": { "text": "FIRING", "color": "red" } }, "type": "value" }
|
|
]},
|
|
{ "id": "custom.displayMode", "value": "color-text" }
|
|
]
|
|
}
|
|
]
|
|
},
|
|
"options": { "showHeader": true },
|
|
"targets": [
|
|
{
|
|
"expr": "ALERTS{alertstate=\"firing\"}",
|
|
"legendFormat": "{{ alertname }} — {{ app }}{{ namespace }}{{ pod }}",
|
|
"refId": "A",
|
|
"instant": true
|
|
}
|
|
]
|
|
}
|
|
],
|
|
"schemaVersion": 39,
|
|
"tags": ["executive", "overview", "cluster", "apps"],
|
|
"templating": { "list": [] },
|
|
"time": { "from": "now-1h", "to": "now" },
|
|
"timepicker": {},
|
|
"timezone": "browser",
|
|
"title": "Executive Summary",
|
|
"uid": "executive-summary"
|
|
}
|