HPCE/grafana-dashb-save
0
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
master
grafana-dashb-save/dashboards/Merlin7/Merlin7_Single_Node_Monitoring.json
T
gitea-actions[bot] e7ae99ebc3 grafana: update resources 
Changed files: alerts/Tier3/SlurmNodeUnstable.json,dashboards/Merlin7/Merlin7_Single_Node_Monitoring.json receivers/Teir3_Alerts.json
2026-07-15 11:01:01 +00:00

5840 lines
158 KiB
JSON
Raw Permalink Blame History

{
"annotations": {
"list": [
{
"builtIn": 1,
"datasource": {
"type": "grafana",
"uid": "grafana"
},
"enable": true,
"hide": true,
"iconColor": "rgba(0, 211, 255, 1)",
"name": "Annotations & Alerts",
"target": {
"limit": 100,
"matchAny": false,
"tags": [],
"type": "dashboard"
},
"type": "dashboard"
}
]
},
"editable": true,
"fiscalYearStartMonth": 0,
"graphTooltip": 1,
"id": null,
"liveNow": false,
"panels": [
{
"collapsed": false,
"gridPos": {
"h": 1,
"w": 24,
"x": 0,
"y": 0
},
"id": 68,
"title": "Overall View",
"type": "row"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"decimals": 1,
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"max": 100,
"min": 0,
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": null
},
{
"color": "orange",
"value": 70
},
{
"color": "red",
"value": 85
}
]
},
"unit": "percent"
}
},
"gridPos": {
"h": 4,
"w": 3,
"x": 0,
"y": 1
},
"id": 2,
"options": {
"barShape": "flat",
"barWidthFactor": 0.5,
"effects": {
"barGlow": false,
"centerGlow": false,
"gradient": true
},
"endpointMarker": "point",
"minVizHeight": 75,
"minVizWidth": 75,
"orientation": "horizontal",
"reduceOptions": {
"calcs": [
"lastNotNull"
],
"fields": "",
"values": false
},
"segmentCount": 1,
"segmentSpacing": 0.3,
"shape": "gauge",
"showThresholdLabels": false,
"showThresholdMarkers": true,
"sizing": "auto",
"sparkline": true,
"textMode": "auto"
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "avg(((nvidia_smi_utilization_gpu_ratio{instance=~\"${server}\"} * 100) or nvidia_smi_utilization_gpu{instance=~\"${server}\"} or nvidia_utilization_gpu{instance=~\"${server}\"} or DCGM_FI_DEV_GPU_UTIL{instance=~\"${server}\"}))",
"instant": false,
"legendFormat": "",
"range": true,
"refId": "A"
}
],
"title": "GPU avg usage",
"type": "gauge"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"decimals": 1,
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"max": 100,
"min": 0,
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": null
},
{
"color": "orange",
"value": 70
},
{
"color": "red",
"value": 90
}
]
},
"unit": "percent"
}
},
"gridPos": {
"h": 4,
"w": 3,
"x": 3,
"y": 1
},
"id": 3,
"options": {
"barShape": "flat",
"barWidthFactor": 0.5,
"effects": {
"barGlow": false,
"centerGlow": false,
"gradient": true
},
"endpointMarker": "point",
"minVizHeight": 75,
"minVizWidth": 75,
"orientation": "horizontal",
"reduceOptions": {
"calcs": [
"lastNotNull"
],
"fields": "",
"values": false
},
"segmentCount": 1,
"segmentSpacing": 0.3,
"shape": "gauge",
"showThresholdLabels": false,
"showThresholdMarkers": true,
"sizing": "auto",
"sparkline": true,
"textMode": "auto"
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "avg((100 * (((nvidia_smi_memory_used_bytes{instance=~\"${server}\"} or nvidia_memory_used_bytes{instance=~\"${server}\"} or (DCGM_FI_DEV_FB_USED{instance=~\"${server}\"} * 1024 * 1024))) / ((nvidia_smi_memory_total_bytes{instance=~\"${server}\"} or nvidia_memory_total_bytes{instance=~\"${server}\"} or (DCGM_FI_DEV_FB_TOTAL{instance=~\"${server}\"} * 1024 * 1024)))) or nvidia_smi_utilization_memory{instance=~\"${server}\"} or nvidia_utilization_memory{instance=~\"${server}\"} or DCGM_FI_DEV_MEM_COPY_UTIL{instance=~\"${server}\"}))",
"instant": false,
"legendFormat": "",
"range": true,
"refId": "A"
}
],
"title": "GPU avg memory",
"type": "gauge"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"decimals": 0,
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": null
},
{
"color": "orange",
"value": 70
},
{
"color": "red",
"value": 80
}
]
},
"unit": "celsius"
}
},
"gridPos": {
"h": 4,
"w": 3,
"x": 6,
"y": 1
},
"id": 4,
"options": {
"colorMode": "value",
"graphMode": "area",
"justifyMode": "auto",
"orientation": "horizontal",
"percentChangeColorMode": "standard",
"reduceOptions": {
"calcs": [
"lastNotNull"
],
"fields": "",
"values": false
},
"showPercentChange": false,
"textMode": "auto",
"wideLayout": true
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "max((nvidia_smi_temperature_gpu{instance=~\"${server}\"} or nvidia_temperatures{instance=~\"${server}\"} or nvidia_temperature_gpu{instance=~\"${server}\"} or DCGM_FI_DEV_GPU_TEMP{instance=~\"${server}\"}))",
"instant": false,
"legendFormat": "",
"range": true,
"refId": "A"
}
],
"title": "GPU max temp",
"type": "stat"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"decimals": 0,
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": null
}
]
},
"unit": "watt"
}
},
"gridPos": {
"h": 4,
"w": 3,
"x": 9,
"y": 1
},
"id": 5,
"options": {
"colorMode": "value",
"graphMode": "area",
"justifyMode": "auto",
"orientation": "horizontal",
"percentChangeColorMode": "standard",
"reduceOptions": {
"calcs": [
"lastNotNull"
],
"fields": "",
"values": false
},
"showPercentChange": false,
"textMode": "auto",
"wideLayout": true
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "sum((nvidia_smi_power_draw_watts{instance=~\"${server}\"} or nvidia_power_usage{instance=~\"${server}\"} or DCGM_FI_DEV_POWER_USAGE{instance=~\"${server}\"}))",
"instant": false,
"legendFormat": "",
"range": true,
"refId": "A"
}
],
"title": "GPU power",
"type": "stat"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"decimals": 1,
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"max": 100,
"min": 0,
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": 0
},
{
"color": "orange",
"value": 70
},
{
"color": "red",
"value": 90
}
]
},
"unit": "percent"
}
},
"gridPos": {
"h": 4,
"w": 3,
"x": 12,
"y": 1
},
"id": 6,
"options": {
"barShape": "flat",
"barWidthFactor": 0.5,
"effects": {
"barGlow": false,
"centerGlow": false,
"gradient": true
},
"endpointMarker": "point",
"minVizHeight": 75,
"minVizWidth": 75,
"orientation": "horizontal",
"reduceOptions": {
"calcs": [
"lastNotNull"
],
"fields": "",
"values": false
},
"segmentCount": 1,
"segmentSpacing": 0.3,
"shape": "gauge",
"showThresholdLabels": false,
"showThresholdMarkers": true,
"sizing": "auto",
"sparkline": true,
"textMode": "auto"
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "100 * (1 - (node_memory_MemAvailable_bytes{instance=~\"${server}\"} / node_memory_MemTotal_bytes{instance=~\"${server}\"}))",
"instant": false,
"legendFormat": "",
"range": true,
"refId": "A"
}
],
"title": "RAM usage",
"type": "gauge"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"decimals": 0,
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": 0
}
]
},
"unit": "s"
}
},
"gridPos": {
"h": 4,
"w": 3,
"x": 15,
"y": 1
},
"id": 9,
"options": {
"colorMode": "value",
"graphMode": "none",
"justifyMode": "auto",
"orientation": "horizontal",
"percentChangeColorMode": "standard",
"reduceOptions": {
"calcs": [
"lastNotNull"
],
"fields": "",
"values": false
},
"showPercentChange": false,
"textMode": "auto",
"wideLayout": true
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "time() - node_boot_time_seconds{instance=~\"${server}\"}",
"instant": false,
"legendFormat": "",
"range": true,
"refId": "A"
}
],
"title": "Uptime",
"type": "stat"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"decimals": 0,
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": 0
}
]
},
"unit": "none"
}
},
"gridPos": {
"h": 4,
"w": 3,
"x": 0,
"y": 5
},
"id": 10,
"options": {
"colorMode": "value",
"graphMode": "area",
"justifyMode": "auto",
"orientation": "horizontal",
"percentChangeColorMode": "standard",
"reduceOptions": {
"calcs": [
"lastNotNull"
],
"fields": "",
"values": false
},
"showPercentChange": false,
"textMode": "auto",
"wideLayout": true
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "node_processes_pids{job=\"integrations/unix\", instance=~\"${server}\"}",
"instant": false,
"legendFormat": "",
"range": true,
"refId": "A"
}
],
"title": "Processes",
"type": "stat"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"decimals": 0,
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": 0
}
]
},
"unit": "none"
}
},
"gridPos": {
"h": 4,
"w": 3,
"x": 3,
"y": 5
},
"id": 11,
"options": {
"colorMode": "value",
"graphMode": "area",
"justifyMode": "auto",
"orientation": "horizontal",
"percentChangeColorMode": "standard",
"reduceOptions": {
"calcs": [
"lastNotNull"
],
"fields": "",
"values": false
},
"showPercentChange": false,
"textMode": "auto",
"wideLayout": true
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "node_processes_threads{instance=~\"${server}\"}",
"instant": false,
"legendFormat": "",
"range": true,
"refId": "A"
}
],
"title": "Threads",
"type": "stat"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"decimals": 0,
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": 0
}
]
},
"unit": "none"
}
},
"gridPos": {
"h": 4,
"w": 3,
"x": 6,
"y": 5
},
"id": 12,
"options": {
"colorMode": "value",
"graphMode": "area",
"justifyMode": "auto",
"orientation": "horizontal",
"percentChangeColorMode": "standard",
"reduceOptions": {
"calcs": [
"lastNotNull"
],
"fields": "",
"values": false
},
"showPercentChange": false,
"textMode": "auto",
"wideLayout": true
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "node_processes_state{instance=~\"${server}\",state=\"Z\"}",
"instant": false,
"legendFormat": "__auto",
"range": true,
"refId": "A"
}
],
"title": "Zombie processes",
"type": "stat"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"decimals": 0,
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": 0
}
]
},
"unit": "none"
}
},
"gridPos": {
"h": 4,
"w": 3,
"x": 9,
"y": 5
},
"id": 67,
"options": {
"colorMode": "value",
"graphMode": "area",
"justifyMode": "auto",
"orientation": "horizontal",
"percentChangeColorMode": "standard",
"reduceOptions": {
"calcs": [
"lastNotNull"
],
"fields": "",
"values": false
},
"showPercentChange": false,
"textMode": "auto",
"wideLayout": true
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "node_processes_state{instance=~\"${server}\",state=\"D\"}",
"instant": false,
"legendFormat": "__auto",
"range": true,
"refId": "A"
}
],
"title": "Processes in D state",
"type": "stat"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"decimals": 1,
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": 0
}
]
},
"unit": "none"
}
},
"gridPos": {
"h": 4,
"w": 3,
"x": 12,
"y": 5
},
"id": 13,
"options": {
"colorMode": "value",
"graphMode": "area",
"justifyMode": "auto",
"orientation": "horizontal",
"percentChangeColorMode": "standard",
"reduceOptions": {
"calcs": [
"lastNotNull"
],
"fields": "",
"values": false
},
"showPercentChange": false,
"textMode": "auto",
"wideLayout": true
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "node_load5{instance=~\"${server}\"}",
"instant": false,
"legendFormat": "",
"range": true,
"refId": "A"
}
],
"title": "Load 5m",
"type": "stat"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"decimals": 1,
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"max": 100,
"min": 0,
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": 0
}
]
},
"unit": "percent"
}
},
"gridPos": {
"h": 4,
"w": 3,
"x": 15,
"y": 5
},
"id": 14,
"options": {
"colorMode": "value",
"graphMode": "area",
"justifyMode": "auto",
"orientation": "horizontal",
"percentChangeColorMode": "standard",
"reduceOptions": {
"calcs": [
"lastNotNull"
],
"fields": "",
"values": false
},
"showPercentChange": false,
"textMode": "auto",
"wideLayout": true
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "100 * avg by (instance) (rate(node_cpu_seconds_total{instance=~\"${server}\",mode=\"iowait\"}[$__rate_interval]))",
"instant": false,
"interval": "1m",
"legendFormat": "__auto",
"range": true,
"refId": "A"
}
],
"title": "IOWait",
"type": "stat"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"decimals": 0,
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": 0
}
]
},
"unit": "Bps"
}
},
"gridPos": {
"h": 8,
"w": 6,
"x": 0,
"y": 9
},
"id": 15,
"options": {
"colorMode": "value",
"graphMode": "area",
"justifyMode": "auto",
"orientation": "horizontal",
"percentChangeColorMode": "standard",
"reduceOptions": {
"calcs": [
"lastNotNull"
],
"fields": "",
"values": false
},
"showPercentChange": false,
"textMode": "auto",
"wideLayout": true
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "sum by (instance) (rate(lustre_receive_bytes_total{instance=~\"${server}\"}[$__rate_interval]))",
"instant": false,
"interval": "1m",
"legendFormat": "Reads",
"range": true,
"refId": "A"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "sum by (instance) (rate(lustre_send_bytes_total{instance=~\"${server}\"}[$__rate_interval]))",
"instant": false,
"legendFormat": "Writes",
"range": true,
"refId": "B"
}
],
"title": "Lustre Data Rate",
"type": "stat"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"decimals": 0,
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": 0
}
]
},
"unit": "ops"
}
},
"gridPos": {
"h": 8,
"w": 6,
"x": 6,
"y": 9
},
"id": 66,
"options": {
"colorMode": "value",
"graphMode": "area",
"justifyMode": "auto",
"orientation": "horizontal",
"percentChangeColorMode": "standard",
"reduceOptions": {
"calcs": [
"lastNotNull"
],
"fields": "",
"values": false
},
"showPercentChange": false,
"textMode": "auto",
"wideLayout": true
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "sum by (instance) (rate(lustre_receive_count_total{instance=~\"${server}\"}[$__rate_interval]))",
"instant": false,
"interval": "1m",
"legendFormat": "Reads",
"range": true,
"refId": "A"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "sum by (instance) (rate(lustre_send_count_total{instance=~\"${server}\"}[$__rate_interval]))",
"instant": false,
"legendFormat": "Writes",
"range": true,
"refId": "B"
}
],
"title": "Lustre Operations",
"type": "stat"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"decimals": 0,
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": 0
}
]
},
"unit": "Bps"
}
},
"gridPos": {
"h": 8,
"w": 6,
"x": 12,
"y": 9
},
"id": 63,
"options": {
"colorMode": "value",
"graphMode": "area",
"justifyMode": "auto",
"orientation": "horizontal",
"percentChangeColorMode": "standard",
"reduceOptions": {
"calcs": [
"lastNotNull"
],
"fields": "",
"values": false
},
"showPercentChange": false,
"textMode": "auto",
"wideLayout": true
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "sum by (instance) (8 * rate(node_network_receive_bytes_total{instance=~\"$server\",device=~\"(vlan|eth|enp|eno|ib|bond|vmbr|ens|hsn|nmn).*\",device!~\"lo|docker.*|veth.*\"}[$__rate_interval]))",
"instant": false,
"interval": "1m",
"legendFormat": "In",
"range": true,
"refId": "A"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "sum by (instance) (8 * rate(node_network_transmit_bytes_total{instance=~\"$server\",device=~\"(vlan|eth|enp|eno|ib|bond|vmbr|ens|hsn|nmn).*\",device!~\"lo|docker.*|veth.*\"}[$__rate_interval]))",
"instant": false,
"interval": "1m",
"legendFormat": "Out",
"range": true,
"refId": "B"
}
],
"title": "Network Rate",
"type": "stat"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"decimals": 0,
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": 0
}
]
},
"unit": "ops"
}
},
"gridPos": {
"h": 6,
"w": 11,
"x": 0,
"y": 17
},
"id": 65,
"options": {
"colorMode": "value",
"graphMode": "area",
"justifyMode": "auto",
"orientation": "horizontal",
"percentChangeColorMode": "standard",
"reduceOptions": {
"calcs": [
"lastNotNull"
],
"fields": "",
"values": false
},
"showPercentChange": false,
"textMode": "auto",
"wideLayout": true
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "sum (avg by (instance) (rate(lustre_receive_count_total{instance=~\"${server}\"}[$__rate_interval])))",
"instant": false,
"interval": "1m",
"legendFormat": "Reads",
"range": true,
"refId": "A"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "sum (avg by (instance) (rate(lustre_send_count_total{instance=~\"${server}\"}[$__rate_interval])))",
"instant": false,
"legendFormat": "Writes",
"range": true,
"refId": "B"
}
],
"title": "Lustre Operations",
"type": "stat"
},
{
"collapsed": true,
"gridPos": {
"h": 1,
"w": 24,
"x": 0,
"y": 23
},
"id": 69,
"panels": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": false,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"barAlignment": 0,
"barWidthFactor": 0.6,
"drawStyle": "line",
"fillOpacity": 14,
"gradientMode": "none",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "smooth",
"lineStyle": {
"fill": "solid"
},
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"log": 2,
"type": "symlog"
},
"showPoints": "auto",
"showValues": false,
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"max": 100,
"min": 0,
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": 0
}
]
},
"unit": "percent"
}
},
"gridPos": {
"h": 8,
"w": 12,
"x": 0,
"y": 24
},
"id": 26,
"options": {
"annotations": {
"clustering": -1,
"multiLane": false
},
"legend": {
"calcs": [
"lastNotNull",
"min",
"mean",
"max"
],
"displayMode": "table",
"placement": "bottom",
"showLegend": true,
"sortBy": "Last *",
"sortDesc": true
},
"tooltip": {
"hideZeros": false,
"mode": "multi",
"sort": "desc"
}
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "100 * avg by (mode) (irate(node_cpu_seconds_total{job=\"integrations/unix\",instance=~\"$server\",mode!=\"idle\"}[$__rate_interval]))",
"instant": false,
"interval": "1m",
"legendFormat": "__auto",
"range": true,
"refId": "A"
}
],
"title": "$server - CPU usage",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": false,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"barAlignment": 0,
"barWidthFactor": 0.6,
"drawStyle": "line",
"fillOpacity": 0,
"gradientMode": "none",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "linear",
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "auto",
"showValues": false,
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": 0
}
]
},
"unit": "short"
}
},
"gridPos": {
"h": 8,
"w": 12,
"x": 12,
"y": 24
},
"id": 27,
"options": {
"annotations": {
"clustering": -1,
"multiLane": false
},
"legend": {
"calcs": [
"lastNotNull",
"min",
"mean",
"max"
],
"displayMode": "table",
"placement": "bottom",
"showLegend": true
},
"tooltip": {
"hideZeros": false,
"mode": "multi",
"sort": "none"
}
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "node_load1{job=\"integrations/unix\",instance=~\"$server\"}",
"instant": false,
"interval": "1m",
"legendFormat": "load1",
"range": true,
"refId": "A"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "node_load5{job=\"integrations/unix\",instance=~\"$server\"}",
"instant": false,
"interval": "1m",
"legendFormat": "load5",
"range": true,
"refId": "B"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "node_load15{job=\"integrations/unix\",instance=~\"$server\"}",
"instant": false,
"interval": "1m",
"legendFormat": "load15",
"range": true,
"refId": "C"
}
],
"title": "$server - Load averages",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": false,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"barAlignment": 0,
"barWidthFactor": 0.6,
"drawStyle": "line",
"fillOpacity": 0,
"gradientMode": "none",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "linear",
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "auto",
"showValues": false,
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": 0
}
]
},
"unit": "short"
}
},
"gridPos": {
"h": 8,
"w": 12,
"x": 0,
"y": 32
},
"id": 28,
"options": {
"annotations": {
"clustering": -1,
"multiLane": false
},
"legend": {
"calcs": [
"lastNotNull",
"min",
"mean",
"max"
],
"displayMode": "table",
"placement": "bottom",
"showLegend": true
},
"tooltip": {
"hideZeros": false,
"mode": "multi",
"sort": "none"
}
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "node_procs_running{instance=~\"${server}\"}",
"instant": false,
"interval": "1m",
"legendFormat": "running",
"range": true,
"refId": "A"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "node_procs_blocked{instance=~\"${server}\"}",
"instant": false,
"interval": "1m",
"legendFormat": "blocked",
"range": true,
"refId": "B"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "node_processes_state{instance=~\"${server}\",state!=\"R\",state!=\"I\",state!=\"S\"}",
"instant": false,
"interval": "1m",
"legendFormat": "{{state}} state",
"range": true,
"refId": "C"
}
],
"title": "$server - Processes",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": false,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"barAlignment": 0,
"barWidthFactor": 0.6,
"drawStyle": "line",
"fillOpacity": 15,
"gradientMode": "none",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "linear",
"lineStyle": {
"fill": "solid"
},
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "auto",
"showValues": false,
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": 0
}
]
},
"unit": "bytes"
}
},
"gridPos": {
"h": 8,
"w": 12,
"x": 12,
"y": 32
},
"id": 29,
"options": {
"annotations": {
"clustering": -1,
"multiLane": false
},
"legend": {
"calcs": [
"lastNotNull",
"min",
"mean",
"max"
],
"displayMode": "table",
"placement": "bottom",
"showLegend": true
},
"tooltip": {
"hideZeros": false,
"mode": "multi",
"sort": "none"
}
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "node_memory_MemTotal_bytes{instance=~\"${server}\"}",
"instant": false,
"legendFormat": "total",
"range": true,
"refId": "A"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "node_memory_MemAvailable_bytes{instance=~\"${server}\"}",
"hide": true,
"instant": false,
"legendFormat": "available",
"range": true,
"refId": "B"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "node_memory_Buffers_bytes{instance=~\"${server}\"}",
"instant": false,
"legendFormat": "buffers",
"range": true,
"refId": "C"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "node_memory_Cached_bytes{instance=~\"${server}\"}",
"instant": false,
"legendFormat": "cached",
"range": true,
"refId": "D"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "node_memory_MemTotal_bytes{instance=~\"${server}\"} - node_memory_MemAvailable_bytes{instance=~\"${server}\"}",
"instant": false,
"legendFormat": "used",
"range": true,
"refId": "E"
}
],
"title": "$server - Memory usage",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": false,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"barAlignment": 0,
"barWidthFactor": 0.6,
"drawStyle": "line",
"fillOpacity": 0,
"gradientMode": "none",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "linear",
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "auto",
"showValues": false,
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": 0
}
]
},
"unit": "ops"
}
},
"gridPos": {
"h": 8,
"w": 12,
"x": 0,
"y": 40
},
"id": 30,
"options": {
"annotations": {
"clustering": -1,
"multiLane": false
},
"legend": {
"calcs": [
"lastNotNull",
"min",
"mean",
"max"
],
"displayMode": "table",
"placement": "bottom",
"showLegend": true
},
"tooltip": {
"hideZeros": false,
"mode": "multi",
"sort": "none"
}
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"exemplar": false,
"expr": "rate(node_context_switches_total{instance=~\"${server}\"}[$__rate_interval])",
"instant": false,
"interval": "1m",
"legendFormat": "Context Switches",
"range": true,
"refId": "A"
}
],
"title": "$server - Context switches",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": false,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"barAlignment": 0,
"barWidthFactor": 0.6,
"drawStyle": "line",
"fillOpacity": 0,
"gradientMode": "none",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "linear",
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "auto",
"showValues": false,
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": 0
}
]
},
"unit": "ops"
}
},
"gridPos": {
"h": 8,
"w": 12,
"x": 12,
"y": 40
},
"id": 31,
"options": {
"annotations": {
"clustering": -1,
"multiLane": false
},
"legend": {
"calcs": [
"lastNotNull",
"min",
"mean",
"max"
],
"displayMode": "table",
"placement": "bottom",
"showLegend": true
},
"tooltip": {
"hideZeros": false,
"mode": "multi",
"sort": "none"
}
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "rate(node_forks_total{instance=~\"${server}\"}[$__rate_interval])",
"instant": false,
"interval": "1m",
"legendFormat": "forks",
"range": true,
"refId": "A"
}
],
"title": "$server - Forks",
"type": "timeseries"
}
],
"title": "System",
"type": "row"
},
{
"collapsed": true,
"gridPos": {
"h": 1,
"w": 24,
"x": 0,
"y": 24
},
"id": 70,
"panels": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": true,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"barAlignment": 0,
"barWidthFactor": 0.6,
"drawStyle": "line",
"fillOpacity": 15,
"gradientMode": "opacity",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "linear",
"lineStyle": {
"fill": "solid"
},
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "auto",
"showValues": false,
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": 0
}
]
},
"unit": "binBps"
},
"overrides": [
{
"matcher": {
"id": "byFrameRefID",
"options": "B"
},
"properties": [
{
"id": "custom.transform",
"value": "negative-Y"
}
]
}
]
},
"gridPos": {
"h": 8,
"w": 12,
"x": 0,
"y": 25
},
"id": 33,
"options": {
"annotations": {
"clustering": -1,
"multiLane": false
},
"legend": {
"calcs": [
"lastNotNull",
"min",
"mean",
"max"
],
"displayMode": "table",
"placement": "bottom",
"showLegend": true
},
"tooltip": {
"hideZeros": false,
"mode": "multi",
"sort": "none"
}
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "rate(node_network_receive_bytes_total{instance=~\"${server}\",device=~\"(vlan|eth|enp|eno|ib|bond|vmbr|ens|hsn|nmn).*\",device!~\"lo|docker.*|veth.*\"}[$__rate_interval])",
"instant": false,
"interval": "1m",
"legendFormat": "{{device}} in",
"range": true,
"refId": "A"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "rate(node_network_transmit_bytes_total{instance=~\"${server}\",device=~\"(vlan|eth|enp|eno|ib|bond|vmbr|ens|hsn|nmn).*\",device!~\"lo|docker.*|veth.*\"}[$__rate_interval])",
"instant": false,
"interval": "1m",
"legendFormat": "{{device}} out",
"range": true,
"refId": "B"
}
],
"title": "$server - Network Usage",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": true,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"barAlignment": 0,
"barWidthFactor": 0.6,
"drawStyle": "line",
"fillOpacity": 15,
"gradientMode": "opacity",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "linear",
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "auto",
"showValues": false,
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": 0
}
]
},
"unit": "pps"
},
"overrides": [
{
"matcher": {
"id": "byFrameRefID",
"options": "B"
},
"properties": [
{
"id": "custom.transform",
"value": "negative-Y"
}
]
}
]
},
"gridPos": {
"h": 8,
"w": 12,
"x": 12,
"y": 25
},
"id": 34,
"options": {
"annotations": {
"clustering": -1,
"multiLane": false
},
"legend": {
"calcs": [
"lastNotNull",
"min",
"mean",
"max"
],
"displayMode": "table",
"placement": "bottom",
"showLegend": true
},
"tooltip": {
"hideZeros": false,
"mode": "multi",
"sort": "none"
}
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "rate(node_network_receive_packets_total{instance=~\"${server}\",device=~\"(vlan|eth|enp|eno|ib|bond|vmbr|ens|hsn|nmn).*\",device!~\"lo|docker.*|veth.*\"}[$__rate_interval])",
"instant": false,
"interval": "1m",
"legendFormat": "{{device}} in",
"range": true,
"refId": "A"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "rate(node_network_transmit_packets_total{instance=~\"${server}\",device=~\"(vlan|eth|enp|eno|ib|bond|vmbr|ens|hsn|nmn).*\",device!~\"lo|docker.*|veth.*\"}[$__rate_interval])",
"instant": false,
"interval": "1m",
"legendFormat": "{{device}} out",
"range": true,
"refId": "B"
}
],
"title": "$server - Network Packets",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": true,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"barAlignment": 0,
"barWidthFactor": 0.6,
"drawStyle": "line",
"fillOpacity": 0,
"gradientMode": "none",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "linear",
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "auto",
"showValues": false,
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": 0
}
]
},
"unit": "pps"
},
"overrides": [
{
"matcher": {
"id": "byFrameRefID",
"options": "B"
},
"properties": [
{
"id": "custom.transform",
"value": "negative-Y"
}
]
}
]
},
"gridPos": {
"h": 8,
"w": 12,
"x": 0,
"y": 33
},
"id": 35,
"options": {
"annotations": {
"clustering": -1,
"multiLane": false
},
"legend": {
"calcs": [
"lastNotNull",
"min",
"mean",
"max"
],
"displayMode": "table",
"placement": "bottom",
"showLegend": true
},
"tooltip": {
"hideZeros": false,
"mode": "multi",
"sort": "none"
}
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "rate(node_network_receive_drop_total{instance=~\"${server}\",device=~\"(vlan|eth|enp|eno|ib|bond|vmbr|ens|hsn|nmn).*\",device!~\"lo|docker.*|veth.*\"}[$__rate_interval])",
"instant": false,
"interval": "1m",
"legendFormat": "{{device}} in",
"range": true,
"refId": "A"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "rate(node_network_transmit_drop_total{instance=~\"${server}\",device=~\"(vlan|eth|enp|eno|ib|bond|vmbr|ens|hsn|nmn).*\",device!~\"lo|docker.*|veth.*\"}[$__rate_interval])",
"instant": false,
"interval": "1m",
"legendFormat": "{{device}} out",
"range": true,
"refId": "B"
}
],
"title": "$server - Network Drops",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": true,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"barAlignment": 0,
"barWidthFactor": 0.6,
"drawStyle": "line",
"fillOpacity": 0,
"gradientMode": "none",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "linear",
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "auto",
"showValues": false,
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": 0
}
]
},
"unit": "pps"
},
"overrides": [
{
"matcher": {
"id": "byFrameRefID",
"options": "B"
},
"properties": [
{
"id": "custom.transform",
"value": "negative-Y"
}
]
}
]
},
"gridPos": {
"h": 8,
"w": 12,
"x": 12,
"y": 33
},
"id": 36,
"options": {
"annotations": {
"clustering": -1,
"multiLane": false
},
"legend": {
"calcs": [
"lastNotNull",
"min",
"mean",
"max"
],
"displayMode": "table",
"placement": "bottom",
"showLegend": true
},
"tooltip": {
"hideZeros": false,
"mode": "multi",
"sort": "none"
}
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "rate(node_network_receive_errs_total{instance=~\"${server}\",device=~\"(vlan|eth|enp|eno|ib|bond|vmbr|ens|hsn|nmn).*\",device!~\"lo|docker.*|veth.*\"}[$__rate_interval])",
"instant": false,
"interval": "1m",
"legendFormat": "{{device}} in",
"range": true,
"refId": "A"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "rate(node_network_transmit_errs_total{instance=~\"${server}\",device=~\"(vlan|eth|enp|eno|ib|bond|vmbr|ens|hsn|nmn).*\",device!~\"lo|docker.*|veth.*\"}[$__rate_interval])",
"instant": false,
"interval": "1m",
"legendFormat": "{{device}} out",
"range": true,
"refId": "B"
}
],
"title": "$server - Network Errors",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": false,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"barAlignment": 0,
"barWidthFactor": 0.6,
"drawStyle": "line",
"fillOpacity": 0,
"gradientMode": "none",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "linear",
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "auto",
"showValues": false,
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": 0
}
]
},
"unit": "short"
}
},
"gridPos": {
"h": 8,
"w": 24,
"x": 0,
"y": 41
},
"id": 37,
"options": {
"annotations": {
"clustering": -1,
"multiLane": false
},
"legend": {
"calcs": [
"lastNotNull",
"min",
"mean",
"max"
],
"displayMode": "table",
"placement": "bottom",
"showLegend": true
},
"tooltip": {
"hideZeros": false,
"mode": "multi",
"sort": "none"
}
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "node_netstat_Tcp_CurrEstab{instance=~\"${server}\"}",
"instant": false,
"legendFormat": "established",
"range": true,
"refId": "A"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "node_sockstat_TCP_inuse{instance=~\"${server}\"}",
"instant": false,
"legendFormat": "tcp in use",
"range": true,
"refId": "B"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "node_sockstat_TCP_tw{instance=~\"${server}\"}",
"instant": false,
"legendFormat": "time_wait",
"range": true,
"refId": "C"
}
],
"title": "$server - TCP Connections",
"type": "timeseries"
}
],
"title": "Network",
"type": "row"
},
{
"collapsed": true,
"gridPos": {
"h": 1,
"w": 24,
"x": 0,
"y": 25
},
"id": 71,
"panels": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": true,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"barAlignment": 0,
"barWidthFactor": 0.6,
"drawStyle": "line",
"fillOpacity": 0,
"gradientMode": "none",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "linear",
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "auto",
"showValues": false,
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": []
},
"unit": "Bps"
},
"overrides": [
{
"matcher": {
"id": "byFrameRefID",
"options": "B"
},
"properties": [
{
"id": "custom.transform",
"value": "negative-Y"
}
]
}
]
},
"gridPos": {
"h": 8,
"w": 12,
"x": 0,
"y": 26
},
"id": 41,
"options": {
"annotations": {
"clustering": -1,
"multiLane": false
},
"legend": {
"calcs": [
"lastNotNull",
"min",
"mean",
"max"
],
"displayMode": "table",
"placement": "bottom",
"showLegend": true
},
"tooltip": {
"hideZeros": false,
"mode": "multi",
"sort": "none"
}
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "sum (avg by (instance) (rate(lustre_receive_bytes_total{instance=~\"${server}\"}[$__rate_interval])))\n",
"instant": false,
"interval": "1m",
"legendFormat": "read",
"range": true,
"refId": "A"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "sum (avg by (instance) (rate(lustre_send_bytes_total{instance=~\"${server}\"}[$__rate_interval])))",
"instant": false,
"interval": "1m",
"legendFormat": "write",
"range": true,
"refId": "B"
}
],
"title": "$server - Lustre Data Rate",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": true,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"barAlignment": 0,
"barWidthFactor": 0.6,
"drawStyle": "line",
"fillOpacity": 0,
"gradientMode": "none",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "linear",
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "auto",
"showValues": false,
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": []
},
"unit": "iops"
},
"overrides": [
{
"matcher": {
"id": "byFrameRefID",
"options": "B"
},
"properties": [
{
"id": "custom.transform",
"value": "negative-Y"
}
]
}
]
},
"gridPos": {
"h": 8,
"w": 12,
"x": 12,
"y": 26
},
"id": 42,
"options": {
"annotations": {
"clustering": -1,
"multiLane": false
},
"legend": {
"calcs": [
"lastNotNull",
"min",
"mean",
"max"
],
"displayMode": "table",
"placement": "bottom",
"showLegend": true
},
"tooltip": {
"hideZeros": false,
"mode": "multi",
"sort": "none"
}
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "sum (avg by (instance) (rate(lustre_send_count_total{instance=~\"${server}\"}[$__rate_interval])))\n",
"instant": false,
"interval": "1m",
"legendFormat": "read",
"range": true,
"refId": "A"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "sum (avg by (instance) (rate(lustre_receive_count_total{instance=~\"${server}\"}[$__rate_interval])))",
"instant": false,
"interval": "1m",
"legendFormat": "write",
"range": true,
"refId": "B"
}
],
"title": "$server - Lustre IOPS",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": true,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"barAlignment": 0,
"barWidthFactor": 0.6,
"drawStyle": "line",
"fillOpacity": 0,
"gradientMode": "none",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "linear",
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "auto",
"showValues": false,
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": []
},
"unit": "binBps"
},
"overrides": [
{
"matcher": {
"id": "byFrameRefID",
"options": "B"
},
"properties": [
{
"id": "custom.transform",
"value": "negative-Y"
}
]
}
]
},
"gridPos": {
"h": 8,
"w": 12,
"x": 0,
"y": 34
},
"id": 43,
"options": {
"annotations": {
"clustering": -1,
"multiLane": false
},
"legend": {
"calcs": [
"lastNotNull",
"min",
"mean",
"max"
],
"displayMode": "table",
"placement": "bottom",
"showLegend": true
},
"tooltip": {
"hideZeros": false,
"mode": "multi",
"sort": "none"
}
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "sum (avg by (instance) (rate(lustre_drop_bytes_total{instance=~\"${server}\"}[$__rate_interval])))\n",
"instant": false,
"interval": "1m",
"legendFormat": "read",
"range": true,
"refId": "A"
}
],
"title": "$server - Lustre Drop MB/s",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": true,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"barAlignment": 0,
"barWidthFactor": 0.6,
"drawStyle": "line",
"fillOpacity": 0,
"gradientMode": "none",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "linear",
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "auto",
"showValues": false,
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": []
},
"unit": "iops"
},
"overrides": [
{
"matcher": {
"id": "byFrameRefID",
"options": "B"
},
"properties": [
{
"id": "custom.transform",
"value": "negative-Y"
}
]
}
]
},
"gridPos": {
"h": 8,
"w": 12,
"x": 12,
"y": 34
},
"id": 44,
"options": {
"annotations": {
"clustering": -1,
"multiLane": false
},
"legend": {
"calcs": [
"lastNotNull",
"min",
"mean",
"max"
],
"displayMode": "table",
"placement": "bottom",
"showLegend": true
},
"tooltip": {
"hideZeros": false,
"mode": "multi",
"sort": "none"
}
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "sum (avg by (instance) (rate(lustre_drop_count_total{instance=~\"${server}\"}[$__rate_interval])))\n",
"instant": false,
"interval": "1m",
"legendFormat": "read",
"range": true,
"refId": "A"
}
],
"title": "$server - Lustre Drop Count",
"type": "timeseries"
}
],
"title": "Lustre",
"type": "row"
},
{
"collapsed": false,
"gridPos": {
"h": 1,
"w": 24,
"x": 0,
"y": 26
},
"id": 72,
"title": "Overall GPU Statistics",
"type": "row"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": false,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"barAlignment": 0,
"barWidthFactor": 0.6,
"drawStyle": "line",
"fillOpacity": 0,
"gradientMode": "none",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "linear",
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "auto",
"showValues": false,
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"max": 100,
"min": 0,
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": null
}
]
},
"unit": "percent"
}
},
"gridPos": {
"h": 8,
"w": 12,
"x": 0,
"y": 27
},
"id": 19,
"options": {
"annotations": {
"clustering": -1,
"multiLane": false
},
"legend": {
"calcs": [
"lastNotNull",
"min",
"mean",
"max"
],
"displayMode": "table",
"placement": "bottom",
"showLegend": true
},
"tooltip": {
"hideZeros": false,
"mode": "multi",
"sort": "none"
}
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "((nvidia_smi_utilization_gpu_ratio{instance=~\"${server}\"} * 100) or nvidia_smi_utilization_gpu{instance=~\"${server}\"} or nvidia_utilization_gpu{instance=~\"${server}\"} or DCGM_FI_DEV_GPU_UTIL{instance=~\"${server}\"})",
"instant": false,
"legendFormat": "{{instance}} gpu{{gpu}}{{index}}{{minor_number}}{{minor}}",
"range": true,
"refId": "A"
}
],
"title": "$server - GPU Usage",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": false,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"barAlignment": 0,
"barWidthFactor": 0.6,
"drawStyle": "line",
"fillOpacity": 0,
"gradientMode": "none",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "linear",
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "auto",
"showValues": false,
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"max": 100,
"min": 0,
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": null
}
]
},
"unit": "percent"
}
},
"gridPos": {
"h": 8,
"w": 12,
"x": 12,
"y": 27
},
"id": 20,
"options": {
"annotations": {
"clustering": -1,
"multiLane": false
},
"legend": {
"calcs": [
"lastNotNull",
"min",
"mean",
"max"
],
"displayMode": "table",
"placement": "bottom",
"showLegend": true
},
"tooltip": {
"hideZeros": false,
"mode": "multi",
"sort": "none"
}
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "(100 * (((nvidia_smi_memory_used_bytes{instance=~\"${server}\"} or nvidia_memory_used_bytes{instance=~\"${server}\"} or (DCGM_FI_DEV_FB_USED{instance=~\"${server}\"} * 1024 * 1024))) / ((nvidia_smi_memory_total_bytes{instance=~\"${server}\"} or nvidia_memory_total_bytes{instance=~\"${server}\"} or (DCGM_FI_DEV_FB_TOTAL{instance=~\"${server}\"} * 1024 * 1024)))) or nvidia_smi_utilization_memory{instance=~\"${server}\"} or nvidia_utilization_memory{instance=~\"${server}\"} or DCGM_FI_DEV_MEM_COPY_UTIL{instance=~\"${server}\"})",
"instant": false,
"legendFormat": "{{instance}} gpu{{gpu}}{{index}}{{minor_number}}{{minor}}",
"range": true,
"refId": "A"
}
],
"title": "$server - GPU Memory Usage",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": false,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"barAlignment": 0,
"barWidthFactor": 0.6,
"drawStyle": "line",
"fillOpacity": 0,
"gradientMode": "none",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "linear",
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "auto",
"showValues": false,
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": null
}
]
},
"unit": "celsius"
}
},
"gridPos": {
"h": 8,
"w": 8,
"x": 0,
"y": 35
},
"id": 21,
"options": {
"annotations": {
"clustering": -1,
"multiLane": false
},
"legend": {
"calcs": [
"lastNotNull",
"min",
"mean",
"max"
],
"displayMode": "table",
"placement": "bottom",
"showLegend": true
},
"tooltip": {
"hideZeros": false,
"mode": "multi",
"sort": "none"
}
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "(nvidia_smi_temperature_gpu{instance=~\"${server}\"} or nvidia_temperatures{instance=~\"${server}\"} or nvidia_temperature_gpu{instance=~\"${server}\"} or DCGM_FI_DEV_GPU_TEMP{instance=~\"${server}\"})",
"instant": false,
"legendFormat": "{{instance}} gpu{{gpu}}{{index}}{{minor_number}}{{minor}}",
"range": true,
"refId": "A"
}
],
"title": "$server - GPU Temperatures",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": false,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"barAlignment": 0,
"barWidthFactor": 0.6,
"drawStyle": "line",
"fillOpacity": 0,
"gradientMode": "none",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "linear",
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "auto",
"showValues": false,
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": null
}
]
},
"unit": "watt"
}
},
"gridPos": {
"h": 8,
"w": 8,
"x": 8,
"y": 35
},
"id": 22,
"options": {
"annotations": {
"clustering": -1,
"multiLane": false
},
"legend": {
"calcs": [
"lastNotNull",
"min",
"mean",
"max"
],
"displayMode": "table",
"placement": "bottom",
"showLegend": true
},
"tooltip": {
"hideZeros": false,
"mode": "multi",
"sort": "none"
}
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "(nvidia_smi_power_draw_watts{instance=~\"${server}\"} or nvidia_power_usage{instance=~\"${server}\"} or DCGM_FI_DEV_POWER_USAGE{instance=~\"${server}\"})",
"instant": false,
"legendFormat": "{{instance}} gpu{{gpu}}{{index}}{{minor_number}}{{minor}}",
"range": true,
"refId": "A"
}
],
"title": "$server - GPU Power Usage",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": false,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"barAlignment": 0,
"barWidthFactor": 0.6,
"drawStyle": "line",
"fillOpacity": 0,
"gradientMode": "none",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "linear",
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "auto",
"showValues": false,
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"max": 100,
"min": 0,
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": null
}
]
},
"unit": "percent"
}
},
"gridPos": {
"h": 8,
"w": 8,
"x": 16,
"y": 35
},
"id": 23,
"options": {
"annotations": {
"clustering": -1,
"multiLane": false
},
"legend": {
"calcs": [
"lastNotNull",
"min",
"mean",
"max"
],
"displayMode": "table",
"placement": "bottom",
"showLegend": true
},
"tooltip": {
"hideZeros": false,
"mode": "multi",
"sort": "none"
}
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "(nvidia_smi_fan_speed{instance=~\"${server}\"} or nvidia_fanspeed{instance=~\"${server}\"} or DCGM_FI_DEV_FAN_SPEED{instance=~\"${server}\"})",
"instant": false,
"legendFormat": "{{instance}} gpu{{gpu}}{{index}}{{minor_number}}{{minor}}",
"range": true,
"refId": "A"
}
],
"title": "$server - GPU Fan Speeds",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": false,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"barAlignment": 0,
"barWidthFactor": 0.6,
"drawStyle": "line",
"fillOpacity": 0,
"gradientMode": "none",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "linear",
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "auto",
"showValues": false,
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": null
}
]
},
"unit": "bytes"
}
},
"gridPos": {
"h": 8,
"w": 24,
"x": 0,
"y": 43
},
"id": 24,
"options": {
"annotations": {
"clustering": -1,
"multiLane": false
},
"legend": {
"calcs": [
"lastNotNull",
"min",
"mean",
"max"
],
"displayMode": "table",
"placement": "bottom",
"showLegend": true
},
"tooltip": {
"hideZeros": false,
"mode": "multi",
"sort": "none"
}
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "(nvidia_smi_memory_used_bytes{instance=~\"${server}\"} or nvidia_memory_used_bytes{instance=~\"${server}\"} or (DCGM_FI_DEV_FB_USED{instance=~\"${server}\"} * 1024 * 1024))",
"instant": false,
"legendFormat": "{{instance}} gpu{{gpu}}{{index}}{{minor_number}}{{minor}} used",
"range": true,
"refId": "A"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "(nvidia_smi_memory_total_bytes{instance=~\"${server}\"} or nvidia_memory_total_bytes{instance=~\"${server}\"} or (DCGM_FI_DEV_FB_TOTAL{instance=~\"${server}\"} * 1024 * 1024))",
"instant": false,
"legendFormat": "{{instance}} gpu{{gpu}}{{index}}{{minor_number}}{{minor}} total",
"range": true,
"refId": "B"
}
],
"title": "$server - GPU Memory Bytes",
"type": "timeseries"
},
{
"collapsed": true,
"gridPos": {
"h": 1,
"w": 24,
"x": 0,
"y": 51
},
"id": 73,
"panels": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": true,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"barAlignment": 0,
"barWidthFactor": 0.6,
"drawStyle": "line",
"fillOpacity": 15,
"gradientMode": "none",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "linear",
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "auto",
"showValues": false,
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": []
},
"unit": "Bps"
},
"overrides": [
{
"matcher": {
"id": "byFrameRefID",
"options": "B"
},
"properties": [
{
"id": "custom.transform",
"value": "negative-Y"
}
]
}
]
},
"gridPos": {
"h": 8,
"w": 12,
"x": 0,
"y": 52
},
"id": 46,
"options": {
"annotations": {
"clustering": -1,
"multiLane": false
},
"legend": {
"calcs": [
"lastNotNull",
"min",
"mean",
"max"
],
"displayMode": "table",
"placement": "bottom",
"showLegend": true
},
"tooltip": {
"hideZeros": false,
"mode": "multi",
"sort": "none"
}
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "rate(node_vmstat_pswpin{instance=~\"${server}\"}[$__rate_interval]) * 4096",
"instant": false,
"interval": "1m",
"legendFormat": "in",
"range": true,
"refId": "A"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "rate(node_vmstat_pswpout{instance=~\"${server}\"}[$__rate_interval]) * 4096",
"instant": false,
"interval": "1m",
"legendFormat": "out",
"range": true,
"refId": "B"
}
],
"title": "$server - Swap I/O bytes",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": false,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"axisSoftMin": 0,
"barAlignment": 0,
"barWidthFactor": 0.6,
"drawStyle": "line",
"fillOpacity": 15,
"gradientMode": "none",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "linear",
"lineWidth": 1,
"pointSize": 3,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "auto",
"showValues": false,
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "normal"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": 0
}
]
},
"unit": "bytes"
},
"overrides": [
{
"matcher": {
"id": "byName",
"options": "total",
"scope": "series"
},
"properties": [
{
"id": "custom.stacking",
"value": {
"group": "A",
"mode": "none"
}
},
{
"id": "custom.fillOpacity",
"value": 0
},
{
"id": "custom.lineWidth",
"value": 3
},
{
"id": "color",
"value": {
"fixedColor": "dark-red",
"mode": "fixed"
}
}
]
},
{
"matcher": {
"id": "byName",
"options": "free",
"scope": "series"
},
"properties": [
{
"id": "custom.fillBelowTo",
"value": "used"
}
]
},
{
"matcher": {
"id": "byName",
"options": "free",
"scope": "series"
},
"properties": [
{
"id": "color",
"value": {
"fixedColor": "green",
"mode": "fixed"
}
}
]
}
]
},
"gridPos": {
"h": 8,
"w": 12,
"x": 12,
"y": 52
},
"id": 47,
"options": {
"annotations": {
"clustering": -1,
"multiLane": false
},
"legend": {
"calcs": [
"lastNotNull",
"min",
"mean",
"max"
],
"displayMode": "table",
"placement": "bottom",
"showLegend": true
},
"tooltip": {
"hideZeros": false,
"mode": "multi",
"sort": "none"
}
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "node_memory_SwapTotal_bytes{instance=~\"${server}\"}",
"instant": false,
"interval": "",
"legendFormat": "total",
"range": true,
"refId": "A"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "node_memory_SwapCached_bytes{instance=~\"${server}\"}",
"instant": false,
"interval": "",
"legendFormat": "cached",
"range": true,
"refId": "D"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "node_memory_SwapTotal_bytes{instance=~\"${server}\"} - node_memory_SwapFree_bytes{instance=~\"${server}\"}",
"instant": false,
"legendFormat": "used",
"range": true,
"refId": "C"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "node_memory_SwapFree_bytes{instance=~\"${server}\"}\n",
"instant": false,
"interval": "",
"legendFormat": "free",
"range": true,
"refId": "B"
}
],
"title": "$server - Swap usage",
"type": "timeseries"
}
],
"title": "Swap",
"type": "row"
},
{
"collapsed": true,
"gridPos": {
"h": 1,
"w": 24,
"x": 0,
"y": 52
},
"id": 74,
"panels": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": true,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"barAlignment": 0,
"barWidthFactor": 0.6,
"drawStyle": "line",
"fillOpacity": 15,
"gradientMode": "opacity",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "linear",
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "auto",
"showValues": false,
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": []
},
"unit": "iops"
},
"overrides": [
{
"matcher": {
"id": "byFrameRefID",
"options": "B"
},
"properties": [
{
"id": "custom.transform",
"value": "negative-Y"
}
]
}
]
},
"gridPos": {
"h": 8,
"w": 12,
"x": 0,
"y": 53
},
"id": 49,
"options": {
"annotations": {
"clustering": -1,
"multiLane": false
},
"legend": {
"calcs": [
"lastNotNull",
"min",
"mean",
"max"
],
"displayMode": "table",
"placement": "bottom",
"showLegend": true
},
"tooltip": {
"hideZeros": false,
"mode": "multi",
"sort": "none"
}
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "rate(node_disk_reads_completed_total{instance=~\"${server}\",device!~\"loop.*|ram.*|dm.*\"}[$__rate_interval])",
"instant": false,
"interval": "1m",
"legendFormat": "{{device}} read",
"range": true,
"refId": "A"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "rate(node_disk_writes_completed_total{instance=~\"${server}\",device!~\"loop.*|ram.*|dm.*\"}[$__rate_interval])",
"instant": false,
"interval": "1m",
"legendFormat": "{{device}} write",
"range": true,
"refId": "B"
}
],
"title": "$server - Disk I/O requests",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": false,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"barAlignment": 0,
"barWidthFactor": 0.6,
"drawStyle": "line",
"fillOpacity": 15,
"gradientMode": "opacity",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "linear",
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "auto",
"showValues": false,
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": []
},
"unit": "Bps"
},
"overrides": [
{
"matcher": {
"id": "byFrameRefID",
"options": "B"
},
"properties": [
{
"id": "custom.transform",
"value": "negative-Y"
}
]
}
]
},
"gridPos": {
"h": 8,
"w": 12,
"x": 12,
"y": 53
},
"id": 50,
"options": {
"annotations": {
"clustering": -1,
"multiLane": false
},
"legend": {
"calcs": [
"lastNotNull",
"min",
"mean",
"max"
],
"displayMode": "table",
"placement": "bottom",
"showLegend": true
},
"tooltip": {
"hideZeros": false,
"mode": "multi",
"sort": "none"
}
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "rate(node_disk_read_bytes_total{instance=~\"${server}\",device!~\"loop.*|ram.*|dm.*\"}[$__rate_interval])",
"instant": false,
"interval": "1m",
"legendFormat": "{{device}} read",
"range": true,
"refId": "A"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "rate(node_disk_written_bytes_total{instance=~\"${server}\",device!~\"loop.*|ram.*|dm.*\"}[$__rate_interval])",
"instant": false,
"interval": "1m",
"legendFormat": "{{device}} write",
"range": true,
"refId": "B"
}
],
"title": "$server - Disk I/O bytes",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": true,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"barAlignment": 0,
"barWidthFactor": 0.6,
"drawStyle": "line",
"fillOpacity": 15,
"gradientMode": "opacity",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "linear",
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "auto",
"showValues": false,
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": []
},
"unit": "s"
},
"overrides": [
{
"matcher": {
"id": "byFrameRefID",
"options": "B"
},
"properties": [
{
"id": "custom.transform",
"value": "negative-Y"
}
]
}
]
},
"gridPos": {
"h": 8,
"w": 24,
"x": 0,
"y": 61
},
"id": 51,
"options": {
"annotations": {
"clustering": -1,
"multiLane": false
},
"legend": {
"calcs": [
"lastNotNull",
"min",
"mean",
"max"
],
"displayMode": "table",
"placement": "bottom",
"showLegend": true
},
"tooltip": {
"hideZeros": false,
"mode": "multi",
"sort": "none"
}
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "rate(node_disk_read_time_seconds_total{instance=~\"${server}\",device!~\"loop.*|ram.*|dm.*\"}[$__rate_interval])",
"instant": false,
"interval": "1m",
"legendFormat": "{{device}} read",
"range": true,
"refId": "A"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"editorMode": "code",
"expr": "rate(node_disk_write_time_seconds_total{instance=~\"${server}\",device!~\"loop.*|ram.*|dm.*\"}[$__rate_interval])",
"instant": false,
"interval": "1m",
"legendFormat": "{{device}} write",
"range": true,
"refId": "B"
}
],
"title": "$server - Disk I/O time",
"type": "timeseries"
}
],
"title": "Disk Performance",
"type": "row"
},
{
"collapsed": true,
"gridPos": {
"h": 1,
"w": 24,
"x": 0,
"y": 53
},
"id": 75,
"panels": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin6-mimir"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": false,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"axisSoftMin": 0,
"barAlignment": 0,
"barWidthFactor": 0.6,
"drawStyle": "line",
"fillOpacity": 15,
"gradientMode": "none",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "linear",
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "auto",
"showValues": false,
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": []
},
"unit": "bytes"
},
"overrides": [
{
"matcher": {
"id": "byFrameRefID",
"options": "A"
},
"properties": [
{
"id": "custom.lineWidth",
"value": 3
},
{
"id": "custom.fillBelowTo",
"value": "used"
}
]
}
]
},
"gridPos": {
"h": 8,
"w": 12,
"x": 0,
"y": 54
},
"id": 53,
"options": {
"annotations": {
"clustering": -1,
"multiLane": false
},
"legend": {
"calcs": [
"lastNotNull",
"min",
"mean",
"max"
],
"displayMode": "table",
"placement": "bottom",
"showLegend": true
},
"tooltip": {
"hideZeros": false,
"mode": "multi",
"sort": "none"
}
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin6-mimir"
},
"editorMode": "code",
"expr": "node_filesystem_size_bytes{instance=~\"${server}\",mountpoint=\"/\",fstype!~\"tmpfs|overlay|squashfs\"}",
"instant": false,
"legendFormat": "total",
"range": true,
"refId": "A"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin6-mimir"
},
"editorMode": "code",
"expr": "node_filesystem_size_bytes{instance=~\"${server}\",mountpoint=\"/\",fstype!~\"tmpfs|overlay|squashfs\"} - node_filesystem_avail_bytes{instance=~\"${server}\",mountpoint=\"/\",fstype!~\"tmpfs|overlay|squashfs\"}",
"instant": false,
"interval": "",
"legendFormat": "used",
"range": true,
"refId": "B"
}
],
"title": "$server - Disk usage '/'",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin6-mimir"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": false,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"axisSoftMin": 0,
"barAlignment": 0,
"barWidthFactor": 0.6,
"drawStyle": "line",
"fillOpacity": 15,
"gradientMode": "none",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "linear",
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "auto",
"showValues": false,
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": []
},
"unit": "bytes"
},
"overrides": [
{
"matcher": {
"id": "byFrameRefID",
"options": "A"
},
"properties": [
{
"id": "custom.fillBelowTo",
"value": "used"
},
{
"id": "custom.lineWidth",
"value": 3
}
]
}
]
},
"gridPos": {
"h": 8,
"w": 12,
"x": 12,
"y": 54
},
"id": 54,
"options": {
"annotations": {
"clustering": -1,
"multiLane": false
},
"legend": {
"calcs": [
"lastNotNull",
"min",
"mean",
"max"
],
"displayMode": "table",
"placement": "bottom",
"showLegend": true
},
"tooltip": {
"hideZeros": false,
"mode": "multi",
"sort": "none"
}
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin6-mimir"
},
"editorMode": "code",
"expr": "node_filesystem_size_bytes{instance=~\"${server}\",mountpoint=\"/boot\",fstype!~\"tmpfs|overlay|squashfs\"}",
"instant": false,
"legendFormat": "total",
"range": true,
"refId": "A"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin6-mimir"
},
"editorMode": "code",
"expr": "node_filesystem_size_bytes{instance=~\"${server}\",mountpoint=\"/boot\",fstype!~\"tmpfs|overlay|squashfs\"} - node_filesystem_avail_bytes{instance=~\"${server}\",mountpoint=\"/boot\",fstype!~\"tmpfs|overlay|squashfs\"}",
"instant": false,
"legendFormat": "used",
"range": true,
"refId": "B"
}
],
"title": "$server - Disk usage '/boot'",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin6-mimir"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": false,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"axisSoftMin": 0,
"barAlignment": 0,
"barWidthFactor": 0.6,
"drawStyle": "line",
"fillOpacity": 15,
"gradientMode": "none",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "linear",
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "auto",
"showValues": false,
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": []
},
"unit": "bytes"
},
"overrides": [
{
"matcher": {
"id": "byFrameRefID",
"options": "A"
},
"properties": [
{
"id": "custom.lineWidth",
"value": 3
},
{
"id": "custom.fillBelowTo",
"value": "used"
}
]
}
]
},
"gridPos": {
"h": 8,
"w": 12,
"x": 0,
"y": 62
},
"id": 55,
"options": {
"annotations": {
"clustering": -1,
"multiLane": false
},
"legend": {
"calcs": [
"lastNotNull",
"min",
"mean",
"max"
],
"displayMode": "table",
"placement": "bottom",
"showLegend": true
},
"tooltip": {
"hideZeros": false,
"mode": "multi",
"sort": "none"
}
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin6-mimir"
},
"editorMode": "code",
"expr": "node_filesystem_size_bytes{instance=~\"${server}\",mountpoint=\"/tmp\",fstype!~\"tmpfs|overlay|squashfs\"}",
"instant": false,
"legendFormat": "total",
"range": true,
"refId": "A"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin6-mimir"
},
"editorMode": "code",
"expr": "node_filesystem_size_bytes{instance=~\"${server}\",mountpoint=\"/tmp\",fstype!~\"tmpfs|overlay|squashfs\"} - node_filesystem_avail_bytes{instance=~\"${server}\",mountpoint=\"/tmp\",fstype!~\"tmpfs|overlay|squashfs\"}",
"instant": false,
"legendFormat": "used",
"range": true,
"refId": "B"
}
],
"title": "$server - Disk usage '/tmp'",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin6-mimir"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": false,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"axisSoftMin": 0,
"barAlignment": 0,
"barWidthFactor": 0.6,
"drawStyle": "line",
"fillOpacity": 15,
"gradientMode": "none",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "linear",
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "auto",
"showValues": false,
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": []
},
"unit": "bytes"
},
"overrides": [
{
"matcher": {
"id": "byFrameRefID",
"options": "A"
},
"properties": [
{
"id": "custom.lineWidth",
"value": 3
},
{
"id": "custom.fillBelowTo",
"value": "used"
}
]
}
]
},
"gridPos": {
"h": 8,
"w": 12,
"x": 12,
"y": 62
},
"id": 61,
"options": {
"annotations": {
"clustering": -1,
"multiLane": false
},
"legend": {
"calcs": [
"lastNotNull",
"min",
"mean",
"max"
],
"displayMode": "table",
"placement": "bottom",
"showLegend": true
},
"tooltip": {
"hideZeros": false,
"mode": "multi",
"sort": "none"
}
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin6-mimir"
},
"editorMode": "code",
"expr": "node_filesystem_size_bytes{instance=~\"${server}\",mountpoint=\"/scratch\",fstype!~\"tmpfs|overlay|squashfs\"}",
"instant": false,
"legendFormat": "total",
"range": true,
"refId": "A"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin6-mimir"
},
"editorMode": "code",
"expr": "node_filesystem_size_bytes{instance=~\"${server}\",mountpoint=\"/scratch\",fstype!~\"tmpfs|overlay|squashfs\"} - node_filesystem_avail_bytes{instance=~\"${server}\",mountpoint=\"/scratch\",fstype!~\"tmpfs|overlay|squashfs\"}",
"instant": false,
"legendFormat": "used",
"range": true,
"refId": "B"
}
],
"title": "$server - Disk usage '/scratch'",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin6-mimir"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": false,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"axisSoftMin": 0,
"barAlignment": 0,
"barWidthFactor": 0.6,
"drawStyle": "line",
"fillOpacity": 15,
"gradientMode": "none",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "linear",
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "auto",
"showValues": false,
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": []
},
"unit": "bytes"
},
"overrides": [
{
"matcher": {
"id": "byFrameRefID",
"options": "A"
},
"properties": [
{
"id": "custom.lineWidth",
"value": 3
},
{
"id": "custom.fillBelowTo",
"value": "used"
}
]
}
]
},
"gridPos": {
"h": 8,
"w": 12,
"x": 0,
"y": 70
},
"id": 60,
"options": {
"annotations": {
"clustering": -1,
"multiLane": false
},
"legend": {
"calcs": [
"lastNotNull",
"min",
"mean",
"max"
],
"displayMode": "table",
"placement": "bottom",
"showLegend": true
},
"tooltip": {
"hideZeros": false,
"mode": "multi",
"sort": "none"
}
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin6-mimir"
},
"editorMode": "code",
"expr": "node_filesystem_size_bytes{instance=~\"${server}\",mountpoint=\"/var\",fstype!~\"tmpfs|overlay|squashfs\"}",
"instant": false,
"legendFormat": "total",
"range": true,
"refId": "A"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin6-mimir"
},
"editorMode": "code",
"expr": "node_filesystem_size_bytes{instance=~\"${server}\",mountpoint=\"/var\",fstype!~\"tmpfs|overlay|squashfs\"} - node_filesystem_avail_bytes{instance=~\"${server}\",mountpoint=\"/var\",fstype!~\"tmpfs|overlay|squashfs\"}",
"instant": false,
"legendFormat": "used",
"range": true,
"refId": "B"
}
],
"title": "$server - Disk usage '/var'",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin6-mimir"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": false,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"axisSoftMin": 0,
"barAlignment": 0,
"barWidthFactor": 0.6,
"drawStyle": "line",
"fillOpacity": 15,
"gradientMode": "none",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "linear",
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "auto",
"showValues": false,
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"thresholds": {
"mode": "absolute",
"steps": []
},
"unit": "bytes"
},
"overrides": [
{
"matcher": {
"id": "byFrameRefID",
"options": "A"
},
"properties": [
{
"id": "custom.lineWidth",
"value": 3
},
{
"id": "custom.fillBelowTo",
"value": "used"
}
]
}
]
},
"gridPos": {
"h": 8,
"w": 11,
"x": 12,
"y": 70
},
"id": 62,
"options": {
"annotations": {
"clustering": -1,
"multiLane": false
},
"legend": {
"calcs": [
"lastNotNull",
"min",
"mean",
"max"
],
"displayMode": "table",
"placement": "bottom",
"showLegend": true
},
"tooltip": {
"hideZeros": false,
"mode": "multi",
"sort": "none"
}
},
"pluginVersion": "13.0.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "merlin6-mimir"
},
"editorMode": "code",
"expr": "node_filesystem_size_bytes{instance=~\"${server}\",mountpoint=\"/var/log\",fstype!~\"tmpfs|overlay|squashfs\"}",
"instant": false,
"legendFormat": "total",
"range": true,
"refId": "A"
},
{
"datasource": {
"type": "prometheus",
"uid": "merlin6-mimir"
},
"editorMode": "code",
"expr": "node_filesystem_size_bytes{instance=~\"${server}\",mountpoint=\"/var/log\",fstype!~\"tmpfs|overlay|squashfs\"} - node_filesystem_avail_bytes{instance=~\"${server}\",mountpoint=\"/var/log\",fstype!~\"tmpfs|overlay|squashfs\"}",
"instant": false,
"legendFormat": "used",
"range": true,
"refId": "B"
}
],
"title": "$server - Disk usage '/var/log'",
"type": "timeseries"
}
],
"title": "Disk Usage",
"type": "row"
}
],
"preload": false,
"refresh": "30s",
"schemaVersion": 42,
"tags": [
"hpc",
"gpu",
"mimir",
"alloy",
"converted"
],
"templating": {
"list": [
{
"allowCustomValue": false,
"current": {
"text": "login001",
"value": "login001"
},
"datasource": {
"type": "prometheus",
"uid": "merlin-mimir"
},
"definition": "label_values({nodename=~\"cn.*|login.*|service.*|gpu.*\"},nodename)",
"hide": 0,
"includeAll": false,
"label": "Server",
"multi": false,
"name": "server",
"options": [],
"query": {
"qryType": 1,
"query": "label_values({nodename=~\"cn.*|login.*|service.*|gpu.*\"},nodename)",
"refId": "PrometheusVariableQueryEditor-VariableQuery"
},
"refresh": 1,
"regex": "",
"regexApplyTo": "value",
"skipUrlSync": false,
"sort": 1,
"type": "query"
}
]
},
"time": {
"from": "now-3h",
"to": "now"
},
"timepicker": {
"refresh_intervals": [
"5s",
"10s",
"30s",
"1m",
"5m",
"15m",
"30m",
"1h",
"2h",
"1d"
]
},
"timezone": "browser",
"title": "Merlin7 Single Node Monitoring",
"uid": "mapwzdq",
"version": 50
}
Reference in New Issue View Git Blame Copy Permalink