grafana/meta/prometheus.json
{
"annotations": {
"list": [
{
"builtIn": 1,
"datasource": "-- Grafana --",
"enable": true,
"hide": true,
"iconColor": "rgba(0, 211, 255, 1)",
"name": "Annotations & Alerts",
"type": "dashboard"
}
]
},
"editable": true,
"gnetId": null,
"graphTooltip": 0,
"id": 8,
"links": [],
"panels": [
{
"datasource": null,
"gridPos": {
"h": 1,
"w": 24,
"x": 0,
"y": 0
},
"id": 10,
"title": "Times",
"type": "row"
},
{
"cacheTimeout": null,
"datasource": "monitor-timescale",
"description": "time since last recorded metric",
"gridPos": {
"h": 5,
"w": 4,
"x": 0,
"y": 1
},
"id": 6,
"links": [],
"options": {
"fieldOptions": {
"calcs": [
"mean"
],
"defaults": {
"decimals": 0,
"mappings": [
{
"id": 0,
"op": "=",
"text": "N/A",
"type": 1,
"value": "null"
}
],
"max": 150,
"min": 0,
"nullValueMode": "connected",
"thresholds": [
{
"color": "green",
"value": null
},
{
"color": "#EAB839",
"value": 90
},
{
"color": "red",
"value": 120
}
],
"unit": "dtdurations"
},
"override": {},
"values": false
},
"orientation": "horizontal",
"showThresholdLabels": false,
"showThresholdMarkers": true
},
"pluginVersion": "6.4.4",
"targets": [
{
"format": "time_series",
"group": [],
"metricColumn": "none",
"rawQuery": true,
"rawSql": " SELECT\n metrics.labels->>'pod' AS \"metric\",\n MAX(\"time\") AS \"time\",\n EXTRACT(EPOCH FROM NOW()) - MAX(value) AS \"value\"\n FROM metrics\n WHERE\n time > NOW() - INTERVAL '15 minute'\n AND name = 'prometheus_remote_storage_highest_timestamp_in_seconds'\n AND value != 'NaN'\n GROUP BY metric\n ORDER BY time DESC\n LIMIT 4",
"refId": "A",
"select": [
[
{
"params": [
"value"
],
"type": "column"
}
]
],
"table": "metrics",
"timeColumn": "\"time\"",
"timeColumnType": "timestamp",
"where": [
{
"name": "$__timeFilter",
"params": [],
"type": "macro"
}
]
}
],
"timeFrom": null,
"timeShift": null,
"title": "Absolute Lag",
"type": "gauge"
},
{
"cacheTimeout": null,
"datasource": "monitor-prometheus",
"gridPos": {
"h": 5,
"w": 4,
"x": 4,
"y": 1
},
"id": 11,
"links": [],
"options": {
"fieldOptions": {
"calcs": [
"lastNotNull"
],
"defaults": {
"mappings": [
{
"id": 0,
"op": "=",
"text": "N/A",
"type": 1,
"value": "null"
}
],
"max": 150,
"min": 0,
"nullValueMode": "connected",
"thresholds": [
{
"color": "green",
"value": null
},
{
"color": "#EAB839",
"value": 90
},
{
"color": "red",
"value": 120
}
],
"unit": "dtdurations"
},
"override": {},
"values": false
},
"orientation": "horizontal",
"showThresholdLabels": false,
"showThresholdMarkers": true
},
"pluginVersion": "6.4.4",
"targets": [
{
"expr": "prometheus_remote_storage_highest_timestamp_in_seconds - on(pod) prometheus_remote_storage_queue_highest_sent_timestamp_seconds{queue=~\"0.*\"}",
"legendFormat": "{{ pod }}",
"refId": "C"
}
],
"timeFrom": null,
"timeShift": null,
"title": "Storage Lag",
"type": "gauge"
},
{
"collapsed": false,
"datasource": null,
"gridPos": {
"h": 1,
"w": 24,
"x": 0,
"y": 6
},
"id": 8,
"panels": [],
"title": "Samples",
"type": "row"
},
{
"alert": {
"alertRuleTags": {},
"conditions": [
{
"evaluator": {
"params": [
1,
1000000000
],
"type": "outside_range"
},
"operator": {
"type": "and"
},
"query": {
"params": [
"A",
"5m",
"now"
]
},
"reducer": {
"params": [],
"type": "avg"
},
"type": "query"
}
],
"executionErrorState": "alerting",
"for": "5m",
"frequency": "1m",
"handler": 1,
"name": "unusual number of samples sent",
"noDataState": "alerting",
"notifications": []
},
"aliasColors": {},
"bars": false,
"dashLength": 10,
"dashes": false,
"datasource": "monitor-timescale",
"fill": 1,
"fillGradient": 1,
"gridPos": {
"h": 7,
"w": 12,
"x": 0,
"y": 7
},
"id": 4,
"interval": "60s",
"legend": {
"avg": false,
"current": false,
"max": false,
"min": false,
"show": true,
"total": false,
"values": false
},
"lines": true,
"linewidth": 1,
"nullPointMode": "null",
"options": {
"dataLinks": []
},
"percentage": false,
"pointradius": 2,
"points": false,
"renderer": "flot",
"seriesOverrides": [],
"spaceLength": 10,
"stack": false,
"steppedLine": true,
"targets": [
{
"format": "time_series",
"group": [],
"metricColumn": "none",
"rawQuery": true,
"rawSql": "SELECT\n tg AS \"time\",\n metric,\n ((\n CASE \n WHEN lag(value) OVER w IS NULL THEN 0\n WHEN value > lag(value) OVER w THEN (value - lag(value) OVER w)\n ELSE value\n END\n ) / EXTRACT(EPOCH FROM INTERVAL '${__interval}')) AS \"value\"\nFROM (\nSELECT\n labels->>'pod' AS \"metric\",\n time_bucket('${__interval}', \"time\") AS \"tg\",\n MAX(value) AS \"value\"\nFROM metrics\nWHERE\n $__timeFilter(\"time\") and\n name = 'prometheus_remote_storage_succeeded_samples_total'\nGROUP BY metric, tg\nORDER BY metric, tg\n) AS m\nWHERE value > 0\nWINDOW w AS (PARTITION BY metric ORDER BY tg)\nORDER BY time",
"refId": "A",
"select": [
[
{
"params": [
"value"
],
"type": "column"
}
]
],
"table": "metrics",
"timeColumn": "\"time\"",
"timeColumnType": "timestamp",
"where": [
{
"name": "$__timeFilter",
"params": [],
"type": "macro"
}
]
},
{
"format": "time_series",
"group": [],
"metricColumn": "none",
"rawQuery": true,
"rawSql": "SELECT\n tg AS \"time\",\n metric,\n ((\n CASE \n WHEN lag(value) OVER w IS NULL THEN 0\n WHEN value > lag(value) OVER w THEN (value - lag(value) OVER w)\n ELSE value\n END\n ) / EXTRACT(EPOCH FROM INTERVAL '${__interval}')) AS \"value\"\nFROM (\nSELECT\n labels->>'pod' AS \"metric\",\n time_bucket('${__interval}', \"time\") AS \"tg\",\n MAX(value) AS \"value\"\nFROM metrics\nWHERE\n $__timeFilter(\"time\") and\n name = 'adapter_samples_sent_total'\nGROUP BY metric, tg\nORDER BY metric, tg\n) AS m\nWHERE value > 0\nWINDOW w AS (PARTITION BY metric ORDER BY tg)\nORDER BY time",
"refId": "B",
"select": [
[
{
"params": [
"value"
],
"type": "column"
}
]
],
"table": "metrics",
"timeColumn": "\"time\"",
"timeColumnType": "timestamp",
"where": [
{
"name": "$__timeFilter",
"params": [],
"type": "macro"
}
]
}
],
"thresholds": [
{
"colorMode": "critical",
"fill": true,
"line": true,
"op": "lt",
"value": 1,
"yaxis": "left"
},
{
"colorMode": "critical",
"fill": true,
"line": true,
"op": "gt",
"value": 1000000000,
"yaxis": "left"
}
],
"timeFrom": null,
"timeRegions": [],
"timeShift": null,
"title": "Samples Sent",
"tooltip": {
"shared": true,
"sort": 0,
"value_type": "individual"
},
"type": "graph",
"xaxis": {
"buckets": null,
"mode": "time",
"name": null,
"show": true,
"values": []
},
"yaxes": [
{
"format": "short",
"label": null,
"logBase": 10,
"max": "10e4",
"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": "monitor-timescale",
"fill": 1,
"fillGradient": 2,
"gridPos": {
"h": 7,
"w": 12,
"x": 12,
"y": 7
},
"id": 12,
"interval": "60s",
"legend": {
"avg": false,
"current": false,
"max": false,
"min": false,
"show": true,
"total": false,
"values": false
},
"lines": true,
"linewidth": 1,
"nullPointMode": "null",
"options": {
"dataLinks": []
},
"percentage": false,
"pointradius": 2,
"points": false,
"renderer": "flot",
"seriesOverrides": [],
"spaceLength": 10,
"stack": false,
"steppedLine": false,
"targets": [
{
"format": "time_series",
"group": [],
"metricColumn": "none",
"rawQuery": true,
"rawSql": "SELECT\n coalesce(MAX(pod), MAX(instance), 'unknown') AS \"metric\",\n \"time\",\n MAX(\"local\") - MAX(\"remote\") AS \"value\"\nFROM\n(\n SELECT\n metrics.labels->>'pod' AS \"pod\",\n metrics.labels->>'instance' AS \"instance\",\n $__timeGroup(\"time\", ${__interval}),\n CASE WHEN name = 'prometheus_remote_storage_highest_timestamp_in_seconds' THEN value ELSE NULL END AS \"local\",\n CASE WHEN name = 'prometheus_remote_storage_queue_highest_sent_timestamp_seconds' THEN value ELSE NULL END AS \"remote\"\n FROM metrics\n WHERE\n $__timeFilter(\"time\")\n AND name IN ('prometheus_remote_storage_highest_timestamp_in_seconds', 'prometheus_remote_storage_queue_highest_sent_timestamp_seconds')\n AND value != 'NaN'\n) t\nGROUP BY time, instance\nORDER BY time, instance;",
"refId": "A",
"select": [
[
{
"params": [
"value"
],
"type": "column"
}
]
],
"table": "metrics",
"timeColumn": "\"time\"",
"timeColumnType": "timestamp",
"where": [
{
"name": "$__timeFilter",
"params": [],
"type": "macro"
}
]
}
],
"thresholds": [],
"timeFrom": null,
"timeRegions": [],
"timeShift": null,
"title": "Storage Lag",
"tooltip": {
"shared": true,
"sort": 0,
"value_type": "individual"
},
"type": "graph",
"xaxis": {
"buckets": null,
"mode": "time",
"name": null,
"show": true,
"values": []
},
"yaxes": [
{
"decimals": 0,
"format": "s",
"label": "",
"logBase": 1,
"max": null,
"min": null,
"show": true
},
{
"format": "short",
"label": null,
"logBase": 1,
"max": null,
"min": null,
"show": true
}
],
"yaxis": {
"align": false,
"alignLevel": null
}
},
{
"aliasColors": {},
"bars": false,
"dashLength": 10,
"dashes": false,
"datasource": "monitor-prometheus",
"fill": 1,
"fillGradient": 1,
"gridPos": {
"h": 7,
"w": 12,
"x": 0,
"y": 14
},
"id": 14,
"legend": {
"avg": false,
"current": false,
"max": false,
"min": false,
"show": true,
"total": false,
"values": false
},
"lines": true,
"linewidth": 1,
"nullPointMode": "null",
"options": {
"dataLinks": []
},
"percentage": false,
"pointradius": 2,
"points": false,
"renderer": "flot",
"seriesOverrides": [],
"spaceLength": 10,
"stack": false,
"steppedLine": true,
"targets": [
{
"expr": "irate(prometheus_remote_storage_succeeded_samples_total[5m])",
"interval": "60s",
"legendFormat": "{{ pod }} - {{ queue }}",
"refId": "A"
}
],
"thresholds": [],
"timeFrom": null,
"timeRegions": [],
"timeShift": null,
"title": "Samples Sent",
"tooltip": {
"shared": true,
"sort": 0,
"value_type": "individual"
},
"type": "graph",
"xaxis": {
"buckets": null,
"mode": "time",
"name": null,
"show": true,
"values": []
},
"yaxes": [
{
"format": "short",
"label": null,
"logBase": 10,
"max": "10e4",
"min": "0",
"show": true
},
{
"format": "short",
"label": null,
"logBase": 1,
"max": null,
"min": null,
"show": true
}
],
"yaxis": {
"align": false,
"alignLevel": null
}
},
{
"alert": {
"alertRuleTags": {},
"conditions": [
{
"evaluator": {
"params": [
60
],
"type": "gt"
},
"operator": {
"type": "and"
},
"query": {
"params": [
"C",
"5m",
"now"
]
},
"reducer": {
"params": [],
"type": "avg"
},
"type": "query"
}
],
"executionErrorState": "alerting",
"for": "3m",
"frequency": "1m",
"handler": 1,
"name": "storage lag",
"noDataState": "alerting",
"notifications": []
},
"aliasColors": {},
"bars": false,
"dashLength": 10,
"dashes": false,
"datasource": "monitor-prometheus",
"fill": 1,
"fillGradient": 2,
"gridPos": {
"h": 7,
"w": 12,
"x": 12,
"y": 14
},
"id": 2,
"legend": {
"avg": false,
"current": false,
"max": false,
"min": false,
"show": true,
"total": false,
"values": false
},
"lines": true,
"linewidth": 1,
"nullPointMode": "null",
"options": {
"dataLinks": []
},
"percentage": false,
"pointradius": 2,
"points": false,
"renderer": "flot",
"seriesOverrides": [],
"spaceLength": 10,
"stack": false,
"steppedLine": false,
"targets": [
{
"expr": "prometheus_remote_storage_highest_timestamp_in_seconds - on(pod) prometheus_remote_storage_queue_highest_sent_timestamp_seconds{queue=~\"0.*\"}",
"interval": "60s",
"legendFormat": "{{ pod }}",
"refId": "C"
}
],
"thresholds": [
{
"colorMode": "critical",
"fill": true,
"line": true,
"op": "gt",
"value": 60
}
],
"timeFrom": null,
"timeRegions": [],
"timeShift": null,
"title": "Storage Lag",
"tooltip": {
"shared": true,
"sort": 0,
"value_type": "individual"
},
"type": "graph",
"xaxis": {
"buckets": null,
"mode": "time",
"name": null,
"show": true,
"values": []
},
"yaxes": [
{
"decimals": 0,
"format": "s",
"label": "",
"logBase": 1,
"max": null,
"min": null,
"show": true
},
{
"format": "short",
"label": null,
"logBase": 1,
"max": null,
"min": null,
"show": true
}
],
"yaxis": {
"align": false,
"alignLevel": null
}
}
],
"refresh": "10s",
"schemaVersion": 20,
"style": "dark",
"tags": [],
"templating": {
"list": []
},
"time": {
"from": "now-6h",
"to": "now"
},
"timepicker": {
"refresh_intervals": [
"5s",
"10s",
"30s",
"1m",
"5m",
"15m",
"30m",
"1h",
"2h",
"1d"
]
},
"timezone": "",
"title": "Prometheus",
"uid": "0B2rX7AWk",
"version": 31
}