mgr/dashboard: monitoring: grafonnet refactoring for radosgw dashboards

This PR intends to refactor radosgw dashboards using grafonnet

Fixes:https://tracker.ceph.com/issues/52777
Signed-off-by: Aashish Sharma <aasharma@redhat.com>
This commit is contained in:
Aashish Sharma 2021-10-08 15:00:09 +05:30
parent c576184cd8
commit eb01954cd9
4 changed files with 1279 additions and 1105 deletions

View File

@ -181,4 +181,132 @@ local addStyle(alias, colorMode, colors, dateFormat, decimals, mappingType, patt
{},'$ceph_hosts Disk utilization', 'Show disk utilization % (util) of any OSD devices on the host by the physical device name and associated OSD id.', 'connected', 'percent', '%Util', 'label_replace(((irate(node_disk_io_time_ms{instance=~\"($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\", \"$1\", \"instance\", \"([^:.]*).*\")', '{{device}}({{ceph_daemon}})', 12, 21, 11, 9
)
])
}
{
"radosgw-sync-overview.json":
local RgwSyncOverviewPanel(title, formatY1, labelY1, rgwMetric, x, y, w, h) =
graphPanelSchema({}, title, '', 'null as zero', true, formatY1, 'short', labelY1, null, 0, 1, '$datasource')
.addTargets(
[addTargetSchema('sum by (source_zone) (rate(%s[30s]))' % rgwMetric, 1, 'time_series', '{{source_zone}}')]) + {gridPos: {x: x, y: y, w: w, h: h}};
dashboardSchema(
'RGW Sync Overview', '', 'rgw-sync-overview', 'now-1h', '15s', 16, ["overview"], '', {refresh_intervals:['5s','10s','15s','30s','1m','5m','15m','30m','1h','2h','1d'],time_options:['5m','15m','1h','6h','12h','24h','2d','7d','30d']}
)
.addAnnotation(
addAnnotationSchema(
1, '-- Grafana --', true, true, 'rgba(0, 211, 255, 1)', 'Annotations & Alerts', 'dashboard')
)
.addRequired(
type='grafana', id='grafana', name='Grafana', version='5.0.0'
)
.addRequired(
type='panel', id='graph', name='Graph', version='5.0.0'
)
.addTemplate(
addTemplateSchema('rgw_servers', '$datasource', 'prometehus', 1, true, 1, '', '')
)
.addTemplate(
g.template.datasource('datasource', 'prometheus', 'default', label='Data Source')
)
.addPanels([
RgwSyncOverviewPanel(
'Replication (throughput) from Source Zone', 'Bps', null, 'ceph_data_sync_from_zone_fetch_bytes_sum', 0, 0, 8, 7),
RgwSyncOverviewPanel(
'Replication (objects) from Source Zone', 'short', 'Objects/s', 'ceph_data_sync_from_zone_fetch_bytes_count', 8, 0, 8, 7),
RgwSyncOverviewPanel(
'Polling Request Latency from Source Zone', 'ms', null, 'ceph_data_sync_from_zone_poll_latency_sum', 16, 0, 8, 7),
RgwSyncOverviewPanel(
'Unsuccessful Object Replications from Source Zone', 'short', 'Count/s', 'ceph_data_sync_from_zone_fetch_errors', 0, 7, 8, 7)
])
}
{
"radosgw-overview.json":
local RgwOverviewPanel(title, description, formatY1, formatY2, expr1, legendFormat1, x, y, w, h) =
graphPanelSchema({}, title, description, 'null', false, formatY1, formatY2, null, null, 0, 1, '$datasource')
.addTargets(
[addTargetSchema(expr1, 1, 'time_series', legendFormat1)]) + {gridPos: {x: x, y: y, w: w, h: h}};
dashboardSchema(
'RGW Overview', '', 'WAkugZpiz', 'now-1h', '15s', 16, ['overview'], '', {refresh_intervals:['5s','10s','15s','30s','1m','5m','15m','30m','1h','2h','1d'],time_options:['5m','15m','1h','6h','12h','24h','2d','7d','30d']}
)
.addAnnotation(
addAnnotationSchema(
1, '-- Grafana --', true, true, 'rgba(0, 211, 255, 1)', 'Annotations & Alerts', 'dashboard')
)
.addRequired(
type='grafana', id='grafana', name='Grafana', version='5.0.0'
)
.addRequired(
type='panel', id='graph', name='Graph', version='5.0.0'
)
.addTemplate(
addTemplateSchema('rgw_servers', '$datasource', 'label_values(ceph_rgw_req, ceph_daemon)', 1, true, 1, '', '')
)
.addTemplate(
g.template.datasource('datasource', 'prometheus', 'default', label='Data Source')
)
.addPanels([
addRowSchema(false, true, 'RGW Overview - All Gateways') + {gridPos: {x: 0, y: 0, w: 24, h: 1}},
RgwOverviewPanel(
'Average GET/PUT Latencies', '', 's', 'short', 'rate(ceph_rgw_get_initial_lat_sum[30s]) / rate(ceph_rgw_get_initial_lat_count[30s])', 'GET AVG', 0, 1, 8, 7).addTargets(
[addTargetSchema('rate(ceph_rgw_put_initial_lat_sum[30s]) / rate(ceph_rgw_put_initial_lat_count[30s])', 1, 'time_series', 'PUT AVG')]),
RgwOverviewPanel(
'Total Requests/sec by RGW Instance', '', 'none', 'short', 'sum by(rgw_host) (label_replace(rate(ceph_rgw_req[30s]), \"rgw_host\", \"$1\", \"ceph_daemon\", \"rgw.(.*)\"))', '{{rgw_host}}', 8, 1, 7, 7),
RgwOverviewPanel(
'GET Latencies by RGW Instance', 'Latencies are shown stacked, without a yaxis to provide a visual indication of GET latency imbalance across RGW hosts', 's', 'short', 'label_replace(rate(ceph_rgw_get_initial_lat_sum[30s]),\"rgw_host\",\"$1\",\"ceph_daemon\",\"rgw.(.*)\") / \nlabel_replace(rate(ceph_rgw_get_initial_lat_count[30s]),\"rgw_host\",\"$1\",\"ceph_daemon\",\"rgw.(.*)\")', '{{rgw_host}}', 15, 1, 6, 7),
RgwOverviewPanel(
'Bandwidth Consumed by Type', 'Total bytes transferred in/out of all radosgw instances within the cluster', 'bytes', 'short', 'sum(rate(ceph_rgw_get_b[30s]))', 'GETs', 0, 8, 8, 6).addTargets(
[addTargetSchema('sum(rate(ceph_rgw_put_b[30s]))', 1, 'time_series', 'PUTs')]),
RgwOverviewPanel(
'Bandwidth by RGW Instance', 'Total bytes transferred in/out through get/put operations, by radosgw instance', 'bytes', 'short', 'sum by(rgw_host) (\n (label_replace(rate(ceph_rgw_get_b[30s]), \"rgw_host\",\"$1\",\"ceph_daemon\",\"rgw.(.*)\")) + \n (label_replace(rate(ceph_rgw_put_b[30s]), \"rgw_host\",\"$1\",\"ceph_daemon\",\"rgw.(.*)\"))\n)', '{{rgw_host}}', 8, 8, 7, 6),
RgwOverviewPanel(
'PUT Latencies by RGW Instance', 'Latencies are shown stacked, without a yaxis to provide a visual indication of PUT latency imbalance across RGW hosts', 's', 'short', 'label_replace(rate(ceph_rgw_put_initial_lat_sum[30s]),\"rgw_host\",\"$1\",\"ceph_daemon\",\"rgw.(.*)\") / \nlabel_replace(rate(ceph_rgw_put_initial_lat_count[30s]),\"rgw_host\",\"$1\",\"ceph_daemon\",\"rgw.(.*)\")', '{{rgw_host}}', 15, 8, 6, 6)
])
}
{
"radosgw-detail.json":
local RgwDetailsPanel(aliasColors, title, description, formatY1, formatY2, expr1, expr2, legendFormat1, legendFormat2, x, y, w, h) =
graphPanelSchema(aliasColors, title, description, 'null', false, formatY1, formatY2, null, null, 0, 1, '$datasource')
.addTargets(
[addTargetSchema(expr1, 1, 'time_series', legendFormat1),addTargetSchema(expr2, 1, 'time_series', legendFormat2)]) + {gridPos: {x: x, y: y, w: w, h: h}};
dashboardSchema(
'RGW Instance Detail', '', 'x5ARzZtmk', 'now-1h', '15s', 16, ['overview'], '', {refresh_intervals:['5s','10s','15s','30s','1m','5m','15m','30m','1h','2h','1d'],time_options:['5m','15m','1h','6h','12h','24h','2d','7d','30d']}
)
.addAnnotation(
addAnnotationSchema(
1, '-- Grafana --', true, true, 'rgba(0, 211, 255, 1)', 'Annotations & Alerts', 'dashboard')
)
.addRequired(
type='grafana', id='grafana', name='Grafana', version='5.0.0'
)
.addRequired(
type='panel', id='grafana-piechart-panel', name='Pie Chart', version='1.3.3'
)
.addRequired(
type='panel', id='graph', name='Graph', version='5.0.0'
)
.addTemplate(
g.template.datasource('datasource', 'prometheus', 'default', label='Data Source')
)
.addTemplate(
addTemplateSchema('rgw_servers', '$datasource', 'label_values(ceph_rgw_req, ceph_daemon)', 1, true, 1, '', '')
)
.addPanels([
addRowSchema(false, true, 'RGW Host Detail : $rgw_servers') + {gridPos: {x: 0, y: 0, w: 24, h: 1}},
RgwDetailsPanel(
{}, '$rgw_servers GET/PUT Latencies', '', 's', 'short', 'sum by (ceph_daemon) (rate(ceph_rgw_get_initial_lat_sum{ceph_daemon=~\"($rgw_servers)\"}[30s]) / rate(ceph_rgw_get_initial_lat_count{ceph_daemon=~\"($rgw_servers)\"}[30s]))', 'sum by (ceph_daemon)(rate(ceph_rgw_put_initial_lat_sum{ceph_daemon=~\"($rgw_servers)\"}[30s]) / rate(ceph_rgw_put_initial_lat_count{ceph_daemon=~\"($rgw_servers)\"}[30s]))', 'GET {{ceph_daemon}}', 'PUT {{ceph_daemon}}', 0, 1, 6, 8),
RgwDetailsPanel(
{}, 'Bandwidth by HTTP Operation', '', 'bytes', 'short', 'rate(ceph_rgw_get_b{ceph_daemon=~\"$rgw_servers\"}[30s])', 'rate(ceph_rgw_put_b{ceph_daemon=~\"$rgw_servers\"}[30s])', 'GETs {{ceph_daemon}}', 'PUTs {{ceph_daemon}}', 6, 1, 7, 8),
RgwDetailsPanel(
{"GETs": "#7eb26d","Other": "#447ebc","PUTs": "#eab839","Requests": "#3f2b5b","Requests Failed": "#bf1b00"},'HTTP Request Breakdown', '', 'short', 'short', 'rate(ceph_rgw_failed_req{ceph_daemon=~\"$rgw_servers\"}[30s])', 'rate(ceph_rgw_get{ceph_daemon=~\"$rgw_servers\"}[30s])', 'Requests Failed {{ceph_daemon}}', 'GETs {{ceph_daemon}}', 13, 1, 7, 8)
.addTargets(
[addTargetSchema('rate(ceph_rgw_put{ceph_daemon=~\"$rgw_servers\"}[30s])', 1, 'time_series', 'PUTs {{ceph_daemon}}'),addTargetSchema('rate(ceph_rgw_req{ceph_daemon=~\"$rgw_servers\"}[30s]) -\n (rate(ceph_rgw_get{ceph_daemon=~\"$rgw_servers\"}[30s]) +\n rate(ceph_rgw_put{ceph_daemon=~\"$rgw_servers\"}[30s]))', 1, 'time_series', 'Other {{ceph_daemon}}')]),
addPieChartSchema(
{"GETs": "#7eb26d","Other (HEAD,POST,DELETE)": "#447ebc","PUTs": "#eab839","Requests": "#3f2b5b","Failures": "#bf1b00"},'$datasource', '', 'Under graph', 'pie', 'Workload Breakdown', 'current')
.addTarget(addTargetSchema('rate(ceph_rgw_failed_req{ceph_daemon=~\"$rgw_servers\"}[30s])', 1, 'time_series', 'Failures {{ceph_daemon}}'))
.addTarget(addTargetSchema('rate(ceph_rgw_get{ceph_daemon=~\"$rgw_servers\"}[30s])', 1, 'time_series', 'GETs {{ceph_daemon}}'))
.addTarget(addTargetSchema('rate(ceph_rgw_put{ceph_daemon=~\"$rgw_servers\"}[30s])', 1, 'time_series', 'PUTs {{ceph_daemon}}'))
.addTarget(addTargetSchema('rate(ceph_rgw_req{ceph_daemon=~\"$rgw_servers\"}[30s]) -\n (rate(ceph_rgw_get{ceph_daemon=~\"$rgw_servers\"}[30s]) +\n rate(ceph_rgw_put{ceph_daemon=~\"$rgw_servers\"}[30s]))', 1, 'time_series', 'Other (DELETE,LIST) {{ceph_daemon}}')) + {gridPos: {x: 20, y: 1, w: 4, h: 8}}
])
}

View File

@ -1,491 +1,499 @@
{
"__requires": [
{
"type": "grafana",
"id": "grafana",
"name": "Grafana",
"version": "5.0.0"
},
{
"type": "panel",
"id": "grafana-piechart-panel",
"name": "Pie Chart",
"version": "1.3.3"
},
{
"type": "panel",
"id": "graph",
"name": "Graph",
"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"
"id": "grafana",
"name": "Grafana",
"type": "grafana",
"version": "5.0.0"
},
{
"id": "grafana-piechart-panel",
"name": "Pie Chart",
"type": "panel",
"version": "1.3.3"
},
{
"id": "graph",
"name": "Graph",
"type": "panel",
"version": "5.0.0"
}
]
},
"editable": false,
"gnetId": null,
"graphTooltip": 0,
"id": null,
"iteration": 1534386250869,
"links": [],
"panels": [
{
"collapsed": false,
"gridPos": {
"h": 1,
"w": 24,
"x": 0,
"y": 0
},
"id": 12,
"panels": [],
"repeat": null,
"title": "RGW Host Detail : $rgw_servers",
"type": "row"
},
{
"aliasColors": {},
"bars": false,
"dashLength": 10,
"dashes": false,
"datasource": "$datasource",
"fill": 1,
"gridPos": {
"h": 8,
"w": 6,
"x": 0,
"y": 1
},
"id": 34,
"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": [],
"spaceLength": 10,
"stack": false,
"steppedLine": false,
"targets": [
{
"expr": "sum by (ceph_daemon) (rate(ceph_rgw_get_initial_lat_sum{ceph_daemon=~\"($rgw_servers)\"}[30s]) / rate(ceph_rgw_get_initial_lat_count{ceph_daemon=~\"($rgw_servers)\"}[30s]))",
"format": "time_series",
"intervalFactor": 1,
"legendFormat": "GET {{ceph_daemon}}",
"refId": "A"
},
{
"expr": "sum by (ceph_daemon)(rate(ceph_rgw_put_initial_lat_sum{ceph_daemon=~\"($rgw_servers)\"}[30s]) / rate(ceph_rgw_put_initial_lat_count{ceph_daemon=~\"($rgw_servers)\"}[30s]))",
"format": "time_series",
"intervalFactor": 1,
"legendFormat": "PUT {{ceph_daemon}}",
"refId": "B"
}
],
"thresholds": [],
"timeFrom": null,
"timeRegions": [],
"timeShift": null,
"title": "$rgw_servers GET/PUT Latencies",
"tooltip": {
"shared": true,
"sort": 0,
"value_type": "individual"
},
"type": "graph",
"xaxis": {
"buckets": null,
"mode": "time",
"name": null,
"show": true,
"values": []
},
"yaxes": [
{
"format": "s",
"label": null,
"logBase": 1,
"max": null,
"min": "0",
"show": true
},
{
"format": "short",
"label": null,
"logBase": 1,
"max": null,
"min": null,
"show": false
}
],
"annotations": {
"list": [
{
"builtIn": 1,
"datasource": "-- Grafana --",
"enable": true,
"hide": true,
"iconColor": "rgba(0, 211, 255, 1)",
"name": "Annotations & Alerts",
"showIn": 0,
"tags": [ ],
"type": "dashboard"
}
]
},
{
"aliasColors": {},
"bars": false,
"dashLength": 10,
"dashes": false,
"datasource": "$datasource",
"fill": 1,
"fillGradient": 0,
"gridPos": {
"h": 8,
"w": 7,
"x": 6,
"y": 1
},
"id": 18,
"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": [],
"spaceLength": 10,
"stack": true,
"steppedLine": false,
"targets": [
{
"expr": "rate(ceph_rgw_get_b{ceph_daemon=~\"$rgw_servers\"}[30s])",
"format": "time_series",
"intervalFactor": 1,
"legendFormat": "GETs {{ceph_daemon}}",
"refId": "B"
},
{
"expr": "rate(ceph_rgw_put_b{ceph_daemon=~\"$rgw_servers\"}[30s])",
"format": "time_series",
"intervalFactor": 1,
"legendFormat": "PUTs {{ceph_daemon}}",
"refId": "A"
}
],
"thresholds": [],
"timeFrom": null,
"timeRegions": [],
"timeShift": null,
"title": "Bandwidth by HTTP Operation",
"tooltip": {
"shared": true,
"sort": 0,
"value_type": "individual"
},
"type": "graph",
"xaxis": {
"buckets": null,
"mode": "time",
"name": null,
"show": true,
"values": []
},
"yaxes": [
{
"decimals": 0,
"format": "bytes",
"label": "",
"logBase": 1,
"max": null,
"min": "0",
"show": true
},
{
"format": "short",
"label": null,
"logBase": 1,
"max": null,
"min": null,
"show": false
}
]
},
{
"aliasColors": {
"GETs": "#7eb26d",
"Other": "#447ebc",
"PUTs": "#eab839",
"Requests": "#3f2b5b",
"Requests Failed": "#bf1b00"
},
"bars": false,
"dashLength": 10,
"dashes": false,
"datasource": "$datasource",
"fill": 1,
"fillGradient": 0,
"gridPos": {
"h": 8,
"w": 7,
"x": 13,
"y": 1
},
"id": 14,
"legend": {
"avg": false,
"current": false,
"max": false,
"min": false,
"show": true,
"total": false,
"values": false
},
"lines": true,
"linewidth": 1,
"links": [],
"nullPointMode": "null",
"percentage": false,
"pointradius": 5,
"points": false,
"renderer": "flot",
"seriesOverrides": [],
"spaceLength": 10,
"stack": true,
"steppedLine": false,
"targets": [
{
"expr": "rate(ceph_rgw_failed_req{ceph_daemon=~\"$rgw_servers\"}[30s])",
"format": "time_series",
"intervalFactor": 1,
"legendFormat": "Requests Failed {{ceph_daemon}}",
"refId": "B"
},
{
"expr": "rate(ceph_rgw_get{ceph_daemon=~\"$rgw_servers\"}[30s])",
"format": "time_series",
"intervalFactor": 1,
"legendFormat": "GETs {{ceph_daemon}}",
"refId": "C"
},
{
"expr": "rate(ceph_rgw_put{ceph_daemon=~\"$rgw_servers\"}[30s])",
"format": "time_series",
"intervalFactor": 1,
"legendFormat": "PUTs {{ceph_daemon}}",
"refId": "D"
},
{
"expr": "rate(ceph_rgw_req{ceph_daemon=~\"$rgw_servers\"}[30s]) -\n (rate(ceph_rgw_get{ceph_daemon=~\"$rgw_servers\"}[30s]) +\n rate(ceph_rgw_put{ceph_daemon=~\"$rgw_servers\"}[30s]))",
"format": "time_series",
"intervalFactor": 1,
"legendFormat": "Other {{ceph_daemon}}",
"refId": "A"
}
],
"thresholds": [],
"timeFrom": null,
"timeRegions": [],
"timeShift": null,
"title": "HTTP Request Breakdown",
"tooltip": {
"shared": true,
"sort": 0,
"value_type": "individual"
},
"type": "graph",
"xaxis": {
"buckets": null,
"mode": "time",
"name": null,
"show": true,
"values": []
},
"yaxes": [
{
"format": "short",
"label": null,
"logBase": 1,
"max": null,
"min": null,
"show": true
},
{
"format": "short",
"label": null,
"logBase": 1,
"max": null,
"min": null,
"show": true
}
]
},
{
"aliasColors": {
"Failures": "#bf1b00",
"GETs": "#7eb26d",
"Other (HEAD,POST,DELETE)": "#447ebc",
"PUTs": "#eab839"
},
"breakPoint": "50%",
"cacheTimeout": null,
"combine": {
"label": "Others",
"threshold": 0
},
"datasource": "$datasource",
"fontSize": "80%",
"format": "none",
"gridPos": {
"h": 8,
"w": 4,
"x": 20,
"y": 1
},
"id": 23,
"interval": null,
"legend": {
"show": true,
"values": true
},
"legendType": "Under graph",
"links": [],
"maxDataPoints": 3,
"nullPointMode": "connected",
"pieType": "pie",
"strokeWidth": 1,
"targets": [
{
"expr": "rate(ceph_rgw_failed_req{ceph_daemon=~\"$rgw_servers\"}[30s])",
"format": "time_series",
"intervalFactor": 1,
"legendFormat": "Failures {{ceph_daemon}}",
"refId": "A"
},
{
"expr": "rate(ceph_rgw_get{ceph_daemon=~\"$rgw_servers\"}[30s])",
"format": "time_series",
"intervalFactor": 1,
"legendFormat": "GETs {{ceph_daemon}}",
"refId": "B"
},
{
"expr": "rate(ceph_rgw_put{ceph_daemon=~\"$rgw_servers\"}[30s])",
"format": "time_series",
"intervalFactor": 1,
"legendFormat": "PUTs {{ceph_daemon}}",
"refId": "C"
},
{
"expr": "rate(ceph_rgw_req{ceph_daemon=~\"$rgw_servers\"}[30s]) -\n (rate(ceph_rgw_get{ceph_daemon=~\"$rgw_servers\"}[30s]) +\n rate(ceph_rgw_put{ceph_daemon=~\"$rgw_servers\"}[30s]))",
"format": "time_series",
"intervalFactor": 1,
"legendFormat": "Other (DELETE,LIST) {{ceph_daemon}}",
"refId": "D"
}
],
"title": "Workload Breakdown",
"type": "grafana-piechart-panel",
"valueName": "current"
}
],
"refresh": "15s",
"schemaVersion": 16,
"style": "dark",
"tags": [
"overview"
],
"templating": {
"list": [
},
"description": "",
"editable": false,
"gnetId": null,
"graphTooltip": 0,
"hideControls": false,
"id": null,
"links": [ ],
"panels": [
{
"current": {
"tags": [],
"text": "default",
"value": "default"
},
"hide": 0,
"label": "Data Source",
"name": "datasource",
"options": [],
"query": "prometheus",
"refresh": 1,
"regex": "",
"type": "datasource"
"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": "RGW Host Detail : $rgw_servers",
"titleSize": "h6",
"type": "row"
},
{
"allValue": null,
"current": {},
"datasource": "$datasource",
"hide": 0,
"includeAll": true,
"label": null,
"multi": false,
"name": "rgw_servers",
"options": [],
"query": "label_values(ceph_rgw_req, ceph_daemon)",
"refresh": 1,
"regex": "",
"sort": 1,
"tagValuesQuery": "",
"tags": [],
"tagsQuery": "",
"type": "query",
"useTags": false
"aliasColors": { },
"bars": false,
"dashLength": 10,
"dashes": false,
"datasource": "$datasource",
"description": "",
"fill": 1,
"gridPos": {
"h": 8,
"w": 6,
"x": 0,
"y": 1
},
"id": 3,
"legend": {
"alignAsTable": false,
"avg": false,
"current": false,
"max": false,
"min": false,
"rightSide": false,
"show": true,
"sideWidth": null,
"total": false,
"values": false
},
"lines": true,
"linewidth": 1,
"links": [ ],
"nullPointMode": "null",
"percentage": false,
"pointradius": 5,
"points": false,
"renderer": "flot",
"repeat": null,
"seriesOverrides": [ ],
"spaceLength": 10,
"stack": false,
"steppedLine": false,
"targets": [
{
"expr": "sum by (ceph_daemon) (rate(ceph_rgw_get_initial_lat_sum{ceph_daemon=~\"($rgw_servers)\"}[30s]) / rate(ceph_rgw_get_initial_lat_count{ceph_daemon=~\"($rgw_servers)\"}[30s]))",
"format": "time_series",
"intervalFactor": 1,
"legendFormat": "GET {{ceph_daemon}}",
"refId": "A"
},
{
"expr": "sum by (ceph_daemon)(rate(ceph_rgw_put_initial_lat_sum{ceph_daemon=~\"($rgw_servers)\"}[30s]) / rate(ceph_rgw_put_initial_lat_count{ceph_daemon=~\"($rgw_servers)\"}[30s]))",
"format": "time_series",
"intervalFactor": 1,
"legendFormat": "PUT {{ceph_daemon}}",
"refId": "B"
}
],
"thresholds": [ ],
"timeFrom": null,
"timeShift": null,
"title": "$rgw_servers GET/PUT Latencies",
"tooltip": {
"shared": true,
"sort": 0,
"value_type": "individual"
},
"type": "graph",
"xaxis": {
"buckets": null,
"mode": "time",
"name": null,
"show": true,
"values": [ ]
},
"yaxes": [
{
"format": "s",
"label": null,
"logBase": 1,
"max": null,
"min": 0,
"show": true
},
{
"format": "short",
"label": null,
"logBase": 1,
"max": null,
"min": 0,
"show": true
}
]
},
{
"aliasColors": { },
"bars": false,
"dashLength": 10,
"dashes": false,
"datasource": "$datasource",
"description": "",
"fill": 1,
"gridPos": {
"h": 8,
"w": 7,
"x": 6,
"y": 1
},
"id": 4,
"legend": {
"alignAsTable": false,
"avg": false,
"current": false,
"max": false,
"min": false,
"rightSide": false,
"show": true,
"sideWidth": null,
"total": false,
"values": false
},
"lines": true,
"linewidth": 1,
"links": [ ],
"nullPointMode": "null",
"percentage": false,
"pointradius": 5,
"points": false,
"renderer": "flot",
"repeat": null,
"seriesOverrides": [ ],
"spaceLength": 10,
"stack": false,
"steppedLine": false,
"targets": [
{
"expr": "rate(ceph_rgw_get_b{ceph_daemon=~\"$rgw_servers\"}[30s])",
"format": "time_series",
"intervalFactor": 1,
"legendFormat": "GETs {{ceph_daemon}}",
"refId": "A"
},
{
"expr": "rate(ceph_rgw_put_b{ceph_daemon=~\"$rgw_servers\"}[30s])",
"format": "time_series",
"intervalFactor": 1,
"legendFormat": "PUTs {{ceph_daemon}}",
"refId": "B"
}
],
"thresholds": [ ],
"timeFrom": null,
"timeShift": null,
"title": "Bandwidth by HTTP Operation",
"tooltip": {
"shared": true,
"sort": 0,
"value_type": "individual"
},
"type": "graph",
"xaxis": {
"buckets": null,
"mode": "time",
"name": null,
"show": true,
"values": [ ]
},
"yaxes": [
{
"format": "bytes",
"label": null,
"logBase": 1,
"max": null,
"min": 0,
"show": true
},
{
"format": "short",
"label": null,
"logBase": 1,
"max": null,
"min": 0,
"show": true
}
]
},
{
"aliasColors": {
"GETs": "#7eb26d",
"Other": "#447ebc",
"PUTs": "#eab839",
"Requests": "#3f2b5b",
"Requests Failed": "#bf1b00"
},
"bars": false,
"dashLength": 10,
"dashes": false,
"datasource": "$datasource",
"description": "",
"fill": 1,
"gridPos": {
"h": 8,
"w": 7,
"x": 13,
"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": [ ],
"spaceLength": 10,
"stack": false,
"steppedLine": false,
"targets": [
{
"expr": "rate(ceph_rgw_failed_req{ceph_daemon=~\"$rgw_servers\"}[30s])",
"format": "time_series",
"intervalFactor": 1,
"legendFormat": "Requests Failed {{ceph_daemon}}",
"refId": "A"
},
{
"expr": "rate(ceph_rgw_get{ceph_daemon=~\"$rgw_servers\"}[30s])",
"format": "time_series",
"intervalFactor": 1,
"legendFormat": "GETs {{ceph_daemon}}",
"refId": "B"
},
{
"expr": "rate(ceph_rgw_put{ceph_daemon=~\"$rgw_servers\"}[30s])",
"format": "time_series",
"intervalFactor": 1,
"legendFormat": "PUTs {{ceph_daemon}}",
"refId": "C"
},
{
"expr": "rate(ceph_rgw_req{ceph_daemon=~\"$rgw_servers\"}[30s]) -\n (rate(ceph_rgw_get{ceph_daemon=~\"$rgw_servers\"}[30s]) +\n rate(ceph_rgw_put{ceph_daemon=~\"$rgw_servers\"}[30s]))",
"format": "time_series",
"intervalFactor": 1,
"legendFormat": "Other {{ceph_daemon}}",
"refId": "D"
}
],
"thresholds": [ ],
"timeFrom": null,
"timeShift": null,
"title": "HTTP Request Breakdown",
"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": 1,
"max": null,
"min": 0,
"show": true
},
{
"format": "short",
"label": null,
"logBase": 1,
"max": null,
"min": 0,
"show": true
}
]
},
{
"aliasColors": {
"Failures": "#bf1b00",
"GETs": "#7eb26d",
"Other (HEAD,POST,DELETE)": "#447ebc",
"PUTs": "#eab839",
"Requests": "#3f2b5b"
},
"datasource": "$datasource",
"description": "",
"gridPos": {
"h": 8,
"w": 4,
"x": 20,
"y": 1
},
"id": 6,
"legend": {
"percentage": true,
"show": true,
"values": true
},
"legendType": "Under graph",
"pieType": "pie",
"targets": [
{
"expr": "rate(ceph_rgw_failed_req{ceph_daemon=~\"$rgw_servers\"}[30s])",
"format": "time_series",
"intervalFactor": 1,
"legendFormat": "Failures {{ceph_daemon}}",
"refId": "A"
},
{
"expr": "rate(ceph_rgw_get{ceph_daemon=~\"$rgw_servers\"}[30s])",
"format": "time_series",
"intervalFactor": 1,
"legendFormat": "GETs {{ceph_daemon}}",
"refId": "B"
},
{
"expr": "rate(ceph_rgw_put{ceph_daemon=~\"$rgw_servers\"}[30s])",
"format": "time_series",
"intervalFactor": 1,
"legendFormat": "PUTs {{ceph_daemon}}",
"refId": "C"
},
{
"expr": "rate(ceph_rgw_req{ceph_daemon=~\"$rgw_servers\"}[30s]) -\n (rate(ceph_rgw_get{ceph_daemon=~\"$rgw_servers\"}[30s]) +\n rate(ceph_rgw_put{ceph_daemon=~\"$rgw_servers\"}[30s]))",
"format": "time_series",
"intervalFactor": 1,
"legendFormat": "Other (DELETE,LIST) {{ceph_daemon}}",
"refId": "D"
}
],
"title": "Workload Breakdown",
"type": "grafana-piechart-panel",
"valueName": "current"
}
]
},
"time": {
"from": "now-1h",
"to": "now"
},
"timepicker": {
"refresh_intervals": [
"5s",
"10s",
"15s",
"30s",
"1m",
"5m",
"15m",
"30m",
"1h",
"2h",
"1d"
],
"time_options": [
"5m",
"15m",
"1h",
"6h",
"12h",
"24h",
"2d",
"7d",
"30d"
]
},
"timezone": "",
"title": "RGW Instance Detail",
"uid": "x5ARzZtmk",
"version": 2
],
"refresh": "15s",
"rows": [ ],
"schemaVersion": 16,
"style": "dark",
"tags": [
"overview"
],
"templating": {
"list": [
{
"current": {
"text": "default",
"value": "default"
},
"hide": 0,
"label": "Data Source",
"name": "datasource",
"options": [ ],
"query": "prometheus",
"refresh": 1,
"regex": "",
"type": "datasource"
},
{
"allValue": null,
"current": { },
"datasource": "$datasource",
"hide": 0,
"includeAll": true,
"label": "",
"multi": false,
"name": "rgw_servers",
"options": [ ],
"query": "label_values(ceph_rgw_req, ceph_daemon)",
"refresh": 1,
"regex": "",
"sort": 1,
"tagValuesQuery": "",
"tags": [ ],
"tagsQuery": "",
"type": "query",
"useTags": false
}
]
},
"time": {
"from": "now-1h",
"to": "now"
},
"timepicker": {
"refresh_intervals": [
"5s",
"10s",
"15s",
"30s",
"1m",
"5m",
"15m",
"30m",
"1h",
"2h",
"1d"
],
"time_options": [
"5m",
"15m",
"1h",
"6h",
"12h",
"24h",
"2d",
"7d",
"30d"
]
},
"timezone": "",
"title": "RGW Instance Detail",
"uid": "x5ARzZtmk",
"version": 0
}

File diff suppressed because it is too large Load Diff

View File

@ -29,6 +29,7 @@
}
]
},
"description": "",
"editable": false,
"gnetId": null,
"graphTooltip": 0,
@ -42,6 +43,7 @@
"dashLength": 10,
"dashes": false,
"datasource": "$datasource",
"description": "",
"fill": 1,
"gridPos": {
"h": 7,
@ -126,6 +128,7 @@
"dashLength": 10,
"dashes": false,
"datasource": "$datasource",
"description": "",
"fill": 1,
"gridPos": {
"h": 7,
@ -210,6 +213,7 @@
"dashLength": 10,
"dashes": false,
"datasource": "$datasource",
"description": "",
"fill": 1,
"gridPos": {
"h": 7,
@ -294,6 +298,7 @@
"dashLength": 10,
"dashes": false,
"datasource": "$datasource",
"description": "",
"fill": 1,
"gridPos": {
"h": 7,
@ -386,9 +391,9 @@
"allValue": null,
"current": { },
"datasource": "$datasource",
"hide": 2,
"hide": 0,
"includeAll": true,
"label": null,
"label": "",
"multi": false,
"name": "rgw_servers",
"options": [ ],