gitadmin/foxhunt
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
ffc83da6439ab76da19f2769a8ad9d34ee5f66a7
foxhunt/infra/k8s/monitoring/dashboards/foxhunt-infrastructure.json
jgrusewski ddf697641a feat(monitoring): add per-service and platform dashboards, reorganize folder structure 
- 7 application service dashboards (Services folder): API Gateway, Trading
  Service, Backtesting, Data Acquisition, ML Training, Trading Agent, Web Gateway
- 9 platform dashboards (Platform folder): Prometheus, Grafana, Loki & Promtail,
  Tempo, Redis, PostgreSQL, Argo Workflows, QuestDB, Cluster & Costs
- Cluster & Costs dashboard includes Scaleway node cost estimates (€/hr, €/month)
  for platform, GPU (H100), and CI compile nodes
- Reorganized existing dashboards into 7 folders: CI, Infrastructure, Operations,
  Platform, Services, Trading, Training (dropped "Foxhunt" prefix from titles)
- Rewrote import.sh: removed ConfigMap deployment, switched to API-only imports
  with automatic folder creation and per-dashboard folder mapping
- All 24 dashboards deployed via Grafana API (Postgres-backed)

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-14 02:30:52 +01:00

2207 lines
48 KiB
JSON
Raw Blame History

{
"annotations": {
"list": [
{
"builtIn": 1,
"datasource": {
"type": "grafana",
"uid": "-- Grafana --"
},
"enable": true,
"hide": true,
"iconColor": "rgba(0, 211, 255, 1)",
"name": "Annotations & Alerts",
"type": "dashboard"
}
]
},
"description": "K8s cluster health, node resources, GPU, storage, and Prometheus health",
"editable": true,
"fiscalYearStartMonth": 0,
"links": [],
"liveNow": false,
"panels": [
{
"collapsed": false,
"gridPos": {
"h": 1,
"w": 24,
"x": 0,
"y": 0
},
"id": 1,
"panels": [],
"title": "Cluster Overview",
"type": "row"
},
{
"datasource": {
"type": "prometheus",
"uid": "${DS_PROMETHEUS}"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "thresholds"
},
"mappings": [],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "red",
"value": null
},
{
"color": "green",
"value": 1
}
]
}
},
"overrides": []
},
"gridPos": {
"h": 4,
"w": 4,
"x": 0,
"y": 1
},
"id": 2,
"options": {
"colorMode": "value",
"graphMode": "none",
"justifyMode": "auto",
"orientation": "auto",
"reduceOptions": {
"calcs": [
"lastNotNull"
],
"fields": "",
"values": false
},
"textMode": "auto"
},
"targets": [
{
"expr": "count(kube_node_info)",
"legendFormat": "",
"refId": "A"
}
],
"title": "Total Nodes",
"type": "stat"
},
{
"datasource": {
"type": "prometheus",
"uid": "${DS_PROMETHEUS}"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "thresholds"
},
"mappings": [],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "red",
"value": null
},
{
"color": "green",
"value": 1
}
]
}
},
"overrides": []
},
"gridPos": {
"h": 4,
"w": 4,
"x": 4,
"y": 1
},
"id": 3,
"options": {
"colorMode": "value",
"graphMode": "none",
"justifyMode": "auto",
"orientation": "auto",
"reduceOptions": {
"calcs": [
"lastNotNull"
],
"fields": "",
"values": false
},
"textMode": "auto"
},
"targets": [
{
"expr": "sum(kube_node_status_condition{condition=\"Ready\", status=\"true\"})",
"legendFormat": "",
"refId": "A"
}
],
"title": "Nodes Ready",
"type": "stat"
},
{
"datasource": {
"type": "prometheus",
"uid": "${DS_PROMETHEUS}"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "thresholds"
},
"mappings": [],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "red",
"value": null
},
{
"color": "green",
"value": 1
}
]
}
},
"overrides": []
},
"gridPos": {
"h": 4,
"w": 4,
"x": 8,
"y": 1
},
"id": 4,
"options": {
"colorMode": "value",
"graphMode": "none",
"justifyMode": "auto",
"orientation": "auto",
"reduceOptions": {
"calcs": [
"lastNotNull"
],
"fields": "",
"values": false
},
"textMode": "auto"
},
"targets": [
{
"expr": "sum(kube_pod_status_phase{namespace=\"foxhunt\", phase=\"Running\"})",
"legendFormat": "",
"refId": "A"
}
],
"title": "Pods Running",
"type": "stat"
},
{
"datasource": {
"type": "prometheus",
"uid": "${DS_PROMETHEUS}"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "thresholds"
},
"mappings": [],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": null
}
]
},
"unit": "short"
},
"overrides": []
},
"gridPos": {
"h": 4,
"w": 4,
"x": 12,
"y": 1
},
"id": 5,
"options": {
"colorMode": "value",
"graphMode": "none",
"justifyMode": "auto",
"orientation": "auto",
"reduceOptions": {
"calcs": [
"lastNotNull"
],
"fields": "",
"values": false
},
"text": {
"titleSize": 12
},
"textMode": "auto"
},
"targets": [
{
"expr": "sum(kube_node_status_capacity{resource=\"cpu\"})",
"legendFormat": "",
"refId": "A"
}
],
"title": "Total CPU Capacity",
"type": "stat"
},
{
"datasource": {
"type": "prometheus",
"uid": "${DS_PROMETHEUS}"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "thresholds"
},
"mappings": [],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": null
}
]
},
"unit": "bytes"
},
"overrides": []
},
"gridPos": {
"h": 4,
"w": 4,
"x": 16,
"y": 1
},
"id": 6,
"options": {
"colorMode": "value",
"graphMode": "none",
"justifyMode": "auto",
"orientation": "auto",
"reduceOptions": {
"calcs": [
"lastNotNull"
],
"fields": "",
"values": false
},
"textMode": "auto"
},
"targets": [
{
"expr": "sum(kube_node_status_capacity{resource=\"memory\"})",
"legendFormat": "",
"refId": "A"
}
],
"title": "Total Memory Capacity",
"type": "stat"
},
{
"datasource": {
"type": "prometheus",
"uid": "${DS_PROMETHEUS}"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "thresholds"
},
"mappings": [],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "yellow",
"value": null
},
{
"color": "green",
"value": 1
}
]
}
},
"overrides": []
},
"gridPos": {
"h": 4,
"w": 4,
"x": 20,
"y": 1
},
"id": 7,
"options": {
"colorMode": "value",
"graphMode": "none",
"justifyMode": "auto",
"orientation": "auto",
"reduceOptions": {
"calcs": [
"lastNotNull"
],
"fields": "",
"values": false
},
"textMode": "auto"
},
"targets": [
{
"expr": "sum(kube_node_status_capacity{resource=\"nvidia_com_gpu\"}) or vector(0)",
"legendFormat": "",
"refId": "A"
}
],
"title": "GPUs Available",
"type": "stat"
},
{
"collapsed": false,
"gridPos": {
"h": 1,
"w": 24,
"x": 0,
"y": 5
},
"id": 10,
"panels": [],
"title": "Node Resources",
"type": "row"
},
{
"datasource": {
"type": "prometheus",
"uid": "${DS_PROMETHEUS}"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisSoftMax": 1,
"axisSoftMin": 0,
"drawStyle": "line",
"fillOpacity": 30,
"lineInterpolation": "smooth",
"lineWidth": 1,
"pointSize": 5,
"showPoints": "never",
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "normal"
}
},
"max": 1,
"min": 0,
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": null
}
]
},
"unit": "percentunit"
},
"overrides": []
},
"gridPos": {
"h": 8,
"w": 12,
"x": 0,
"y": 6
},
"id": 11,
"options": {
"legend": {
"calcs": [
"mean",
"max"
],
"displayMode": "table",
"placement": "bottom"
},
"tooltip": {
"mode": "multi",
"sort": "desc"
}
},
"targets": [
{
"expr": "(1 - avg by (instance) (rate(node_cpu_seconds_total{mode=\"idle\"}[$__rate_interval]))) * on(instance) group_left(nodename) node_uname_info",
"legendFormat": "{{ nodename }}",
"refId": "A"
}
],
"title": "CPU Usage by Node (stacked)",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "${DS_PROMETHEUS}"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"drawStyle": "line",
"fillOpacity": 20,
"lineInterpolation": "smooth",
"lineWidth": 1,
"pointSize": 5,
"showPoints": "never",
"spanNulls": false,
"stacking": {
"mode": "none"
}
},
"max": 1,
"min": 0,
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": null
},
{
"color": "orange",
"value": 0.8
},
{
"color": "red",
"value": 0.95
}
]
},
"unit": "percentunit"
},
"overrides": []
},
"gridPos": {
"h": 8,
"w": 12,
"x": 12,
"y": 6
},
"id": 12,
"options": {
"legend": {
"calcs": [
"mean",
"max"
],
"displayMode": "table",
"placement": "bottom"
},
"tooltip": {
"mode": "multi",
"sort": "desc"
}
},
"targets": [
{
"expr": "(1 - (node_memory_MemAvailable_bytes / node_memory_MemTotal_bytes)) * on(instance) group_left(nodename) node_uname_info",
"legendFormat": "{{ nodename }}",
"refId": "A"
}
],
"title": "Memory Usage by Node",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "${DS_PROMETHEUS}"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"drawStyle": "line",
"fillOpacity": 15,
"lineInterpolation": "smooth",
"lineWidth": 1,
"pointSize": 5,
"showPoints": "never",
"spanNulls": false,
"stacking": {
"mode": "none"
}
},
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": null
}
]
},
"unit": "Bps"
},
"overrides": []
},
"gridPos": {
"h": 8,
"w": 8,
"x": 0,
"y": 14
},
"id": 13,
"options": {
"legend": {
"displayMode": "list",
"placement": "bottom"
},
"tooltip": {
"mode": "multi",
"sort": "desc"
}
},
"targets": [
{
"expr": "sum by (instance) (rate(node_disk_read_bytes_total[$__rate_interval])) * on(instance) group_left(nodename) node_uname_info",
"legendFormat": "{{ nodename }} read",
"refId": "A"
},
{
"expr": "sum by (instance) (rate(node_disk_written_bytes_total[$__rate_interval])) * on(instance) group_left(nodename) node_uname_info",
"legendFormat": "{{ nodename }} write",
"refId": "B"
}
],
"title": "Disk I/O (Read / Write)",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "${DS_PROMETHEUS}"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"drawStyle": "line",
"fillOpacity": 15,
"lineInterpolation": "smooth",
"lineWidth": 1,
"pointSize": 5,
"showPoints": "never",
"spanNulls": false,
"stacking": {
"mode": "none"
}
},
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": null
}
]
},
"unit": "Bps"
},
"overrides": []
},
"gridPos": {
"h": 8,
"w": 8,
"x": 8,
"y": 14
},
"id": 14,
"options": {
"legend": {
"displayMode": "list",
"placement": "bottom"
},
"tooltip": {
"mode": "multi",
"sort": "desc"
}
},
"targets": [
{
"expr": "sum by (instance) (rate(node_network_receive_bytes_total{device!~\"lo|veth.*|cali.*|flannel.*\"}[$__rate_interval])) * on(instance) group_left(nodename) node_uname_info",
"legendFormat": "{{ nodename }} rx",
"refId": "A"
},
{
"expr": "sum by (instance) (rate(node_network_transmit_bytes_total{device!~\"lo|veth.*|cali.*|flannel.*\"}[$__rate_interval])) * on(instance) group_left(nodename) node_uname_info",
"legendFormat": "{{ nodename }} tx",
"refId": "B"
}
],
"title": "Network I/O (Receive / Transmit)",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "${DS_PROMETHEUS}"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"drawStyle": "line",
"fillOpacity": 0,
"lineInterpolation": "smooth",
"lineWidth": 1,
"pointSize": 5,
"showPoints": "never",
"spanNulls": false,
"stacking": {
"mode": "none"
}
},
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": null
}
]
},
"unit": "short"
},
"overrides": []
},
"gridPos": {
"h": 8,
"w": 8,
"x": 16,
"y": 14
},
"id": 15,
"options": {
"legend": {
"displayMode": "list",
"placement": "bottom"
},
"tooltip": {
"mode": "multi",
"sort": "desc"
}
},
"targets": [
{
"expr": "node_load1 * on(instance) group_left(nodename) node_uname_info",
"legendFormat": "{{ nodename }} load1",
"refId": "A"
},
{
"expr": "node_load5 * on(instance) group_left(nodename) node_uname_info",
"legendFormat": "{{ nodename }} load5",
"refId": "B"
},
{
"expr": "node_load15 * on(instance) group_left(nodename) node_uname_info",
"legendFormat": "{{ nodename }} load15",
"refId": "C"
}
],
"title": "Load Average (1m / 5m / 15m)",
"type": "timeseries"
},
{
"collapsed": false,
"gridPos": {
"h": 1,
"w": 24,
"x": 0,
"y": 22
},
"id": 20,
"panels": [],
"title": "Kubernetes Workloads",
"type": "row"
},
{
"datasource": {
"type": "prometheus",
"uid": "${DS_PROMETHEUS}"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"drawStyle": "line",
"fillOpacity": 0,
"lineInterpolation": "stepAfter",
"lineWidth": 2,
"pointSize": 5,
"showPoints": "never",
"spanNulls": false,
"stacking": {
"mode": "none"
}
},
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": null
}
]
},
"unit": "short"
},
"overrides": [
{
"matcher": {
"id": "byRegexp",
"options": ".*desired"
},
"properties": [
{
"id": "custom.lineStyle",
"value": {
"dash": [
10,
10
],
"fill": "dash"
}
}
]
}
]
},
"gridPos": {
"h": 8,
"w": 12,
"x": 0,
"y": 23
},
"id": 21,
"options": {
"legend": {
"calcs": [
"lastNotNull"
],
"displayMode": "table",
"placement": "bottom"
},
"tooltip": {
"mode": "multi",
"sort": "desc"
}
},
"targets": [
{
"expr": "kube_deployment_spec_replicas{namespace=\"foxhunt\"}",
"legendFormat": "{{ deployment }} desired",
"refId": "A"
},
{
"expr": "kube_deployment_status_replicas_available{namespace=\"foxhunt\"}",
"legendFormat": "{{ deployment }} available",
"refId": "B"
}
],
"title": "Deployment Replicas (Desired vs Available)",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "${DS_PROMETHEUS}"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": null
}
]
}
},
"overrides": [
{
"matcher": {
"id": "byName",
"options": "Running"
},
"properties": [
{
"id": "color",
"value": {
"fixedColor": "green",
"mode": "fixed"
}
}
]
},
{
"matcher": {
"id": "byName",
"options": "Pending"
},
"properties": [
{
"id": "color",
"value": {
"fixedColor": "yellow",
"mode": "fixed"
}
}
]
},
{
"matcher": {
"id": "byName",
"options": "Failed"
},
"properties": [
{
"id": "color",
"value": {
"fixedColor": "red",
"mode": "fixed"
}
}
]
},
{
"matcher": {
"id": "byName",
"options": "Succeeded"
},
"properties": [
{
"id": "color",
"value": {
"fixedColor": "blue",
"mode": "fixed"
}
}
]
}
]
},
"gridPos": {
"h": 8,
"w": 6,
"x": 12,
"y": 23
},
"id": 22,
"options": {
"legend": {
"displayMode": "list",
"placement": "right"
},
"pieType": "donut",
"reduceOptions": {
"calcs": [
"lastNotNull"
],
"fields": "",
"values": false
},
"tooltip": {
"mode": "single"
}
},
"targets": [
{
"expr": "sum by (phase) (kube_pod_status_phase{namespace=\"foxhunt\"})",
"legendFormat": "{{ phase }}",
"refId": "A"
}
],
"title": "Pod Phase Distribution",
"type": "piechart"
},
{
"datasource": {
"type": "prometheus",
"uid": "${DS_PROMETHEUS}"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"drawStyle": "bars",
"fillOpacity": 80,
"lineInterpolation": "linear",
"lineWidth": 1,
"pointSize": 5,
"showPoints": "never",
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "normal"
}
},
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": null
},
{
"color": "red",
"value": 1
}
]
},
"unit": "short"
},
"overrides": []
},
"gridPos": {
"h": 8,
"w": 6,
"x": 18,
"y": 23
},
"id": 23,
"options": {
"legend": {
"displayMode": "list",
"placement": "bottom"
},
"tooltip": {
"mode": "multi",
"sort": "desc"
}
},
"targets": [
{
"expr": "increase(kube_pod_container_status_restarts_total{namespace=\"foxhunt\"}[$__rate_interval])",
"legendFormat": "{{ pod }} / {{ container }}",
"refId": "A"
}
],
"title": "Container Restarts",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "${DS_PROMETHEUS}"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "thresholds"
},
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": null
}
]
}
},
"overrides": [
{
"matcher": {
"id": "byName",
"options": "Time"
},
"properties": [
{
"id": "custom.hidden",
"value": true
}
]
},
{
"matcher": {
"id": "byName",
"options": "__name__"
},
"properties": [
{
"id": "custom.hidden",
"value": true
}
]
},
{
"matcher": {
"id": "byName",
"options": "job"
},
"properties": [
{
"id": "custom.hidden",
"value": true
}
]
},
{
"matcher": {
"id": "byName",
"options": "instance"
},
"properties": [
{
"id": "custom.hidden",
"value": true
}
]
}
]
},
"gridPos": {
"h": 6,
"w": 24,
"x": 0,
"y": 31
},
"id": 24,
"options": {
"footer": {
"show": false
},
"showHeader": true,
"sortBy": [
{
"desc": true,
"displayName": "Value"
}
]
},
"targets": [
{
"expr": "kube_pod_status_phase{namespace=\"foxhunt\", phase=~\"Pending|Failed\"} > 0",
"format": "table",
"instant": true,
"legendFormat": "",
"refId": "A"
}
],
"title": "Pending / Failed Pods",
"transformations": [
{
"id": "organize",
"options": {
"excludeByName": {
"Time": true,
"__name__": true,
"instance": true,
"job": true
},
"indexByName": {},
"renameByName": {
"Value": "Count",
"namespace": "Namespace",
"phase": "Phase",
"pod": "Pod"
}
}
}
],
"type": "table"
},
{
"collapsed": false,
"gridPos": {
"h": 1,
"w": 24,
"x": 0,
"y": 37
},
"id": 30,
"panels": [],
"title": "Storage",
"type": "row"
},
{
"datasource": {
"type": "prometheus",
"uid": "${DS_PROMETHEUS}"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "thresholds"
},
"max": 1,
"min": 0,
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": null
},
{
"color": "yellow",
"value": 0.7
},
{
"color": "orange",
"value": 0.85
},
{
"color": "red",
"value": 0.95
}
]
},
"unit": "percentunit"
},
"overrides": []
},
"gridPos": {
"h": 8,
"w": 12,
"x": 0,
"y": 38
},
"id": 31,
"options": {
"displayMode": "gradient",
"minVizHeight": 10,
"minVizWidth": 0,
"orientation": "horizontal",
"reduceOptions": {
"calcs": [
"lastNotNull"
],
"fields": "",
"values": false
},
"showUnfilled": true
},
"targets": [
{
"expr": "(1 - (node_filesystem_avail_bytes{fstype=~\"ext4|xfs|btrfs\"} / node_filesystem_size_bytes{fstype=~\"ext4|xfs|btrfs\"})) * on(instance) group_left(nodename) node_uname_info",
"instant": true,
"legendFormat": "{{ nodename }} {{ mountpoint }}",
"refId": "A"
}
],
"title": "Disk Usage by Mountpoint",
"type": "bargauge"
},
{
"datasource": {
"type": "prometheus",
"uid": "${DS_PROMETHEUS}"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "thresholds"
},
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": null
}
]
}
},
"overrides": [
{
"matcher": {
"id": "byName",
"options": "Storage"
},
"properties": [
{
"id": "unit",
"value": "bytes"
}
]
}
]
},
"gridPos": {
"h": 8,
"w": 12,
"x": 12,
"y": 38
},
"id": 32,
"options": {
"footer": {
"show": false
},
"showHeader": true
},
"targets": [
{
"expr": "kube_persistentvolumeclaim_status_phase{namespace=\"foxhunt\"}",
"format": "table",
"instant": true,
"legendFormat": "",
"refId": "A"
},
{
"expr": "kube_persistentvolumeclaim_resource_requests_storage_bytes{namespace=\"foxhunt\"}",
"format": "table",
"instant": true,
"legendFormat": "",
"refId": "B"
}
],
"title": "PVC Status & Requested Storage",
"transformations": [
{
"id": "seriesToColumns",
"options": {
"byField": "persistentvolumeclaim"
}
},
{
"id": "organize",
"options": {
"excludeByName": {
"Time 1": true,
"Time 2": true,
"__name__ 1": true,
"__name__ 2": true,
"instance 1": true,
"instance 2": true,
"job 1": true,
"job 2": true,
"namespace 1": true,
"namespace 2": true,
"phase": false
},
"renameByName": {
"Value 1": "Phase Value",
"Value 2": "Storage",
"persistentvolumeclaim": "PVC",
"phase": "Phase"
}
}
}
],
"type": "table"
},
{
"collapsed": false,
"gridPos": {
"h": 1,
"w": 24,
"x": 0,
"y": 46
},
"id": 40,
"panels": [],
"title": "GPU Resources",
"type": "row"
},
{
"datasource": {
"type": "prometheus",
"uid": "${DS_PROMETHEUS}"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"drawStyle": "line",
"fillOpacity": 25,
"lineInterpolation": "smooth",
"lineWidth": 2,
"pointSize": 5,
"showPoints": "never",
"spanNulls": false,
"stacking": {
"mode": "none"
},
"thresholdsStyle": {
"mode": "line"
}
},
"max": 100,
"min": 0,
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": null
},
{
"color": "yellow",
"value": 80
},
{
"color": "red",
"value": 95
}
]
},
"unit": "percent"
},
"overrides": []
},
"gridPos": {
"h": 8,
"w": 8,
"x": 0,
"y": 47
},
"id": 41,
"options": {
"legend": {
"displayMode": "list",
"placement": "bottom"
},
"tooltip": {
"mode": "multi",
"sort": "desc"
}
},
"targets": [
{
"expr": "DCGM_FI_DEV_GPU_UTIL",
"legendFormat": "GPU {{ gpu }} ({{ instance }})",
"refId": "A"
}
],
"title": "GPU Utilization %",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "${DS_PROMETHEUS}"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"drawStyle": "line",
"fillOpacity": 20,
"lineInterpolation": "smooth",
"lineWidth": 1,
"pointSize": 5,
"showPoints": "never",
"spanNulls": false,
"stacking": {
"mode": "none"
}
},
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": null
}
]
},
"unit": "decmbytes"
},
"overrides": [
{
"matcher": {
"id": "byRegexp",
"options": ".*free"
},
"properties": [
{
"id": "custom.fillOpacity",
"value": 5
},
{
"id": "custom.lineStyle",
"value": {
"dash": [
10,
10
],
"fill": "dash"
}
}
]
}
]
},
"gridPos": {
"h": 8,
"w": 8,
"x": 8,
"y": 47
},
"id": 42,
"options": {
"legend": {
"displayMode": "list",
"placement": "bottom"
},
"tooltip": {
"mode": "multi",
"sort": "desc"
}
},
"targets": [
{
"expr": "DCGM_FI_DEV_FB_USED",
"legendFormat": "GPU {{ gpu }} used",
"refId": "A"
},
{
"expr": "DCGM_FI_DEV_FB_FREE",
"legendFormat": "GPU {{ gpu }} free",
"refId": "B"
}
],
"title": "VRAM Usage",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "${DS_PROMETHEUS}"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"drawStyle": "line",
"fillOpacity": 0,
"lineInterpolation": "smooth",
"lineWidth": 2,
"pointSize": 5,
"showPoints": "never",
"spanNulls": false,
"stacking": {
"mode": "none"
},
"thresholdsStyle": {
"mode": "line"
}
},
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": null
},
{
"color": "orange",
"value": 75
},
{
"color": "red",
"value": 90
}
]
},
"unit": "celsius"
},
"overrides": []
},
"gridPos": {
"h": 8,
"w": 8,
"x": 16,
"y": 47
},
"id": 43,
"options": {
"legend": {
"displayMode": "list",
"placement": "bottom"
},
"tooltip": {
"mode": "multi",
"sort": "desc"
}
},
"targets": [
{
"expr": "DCGM_FI_DEV_GPU_TEMP",
"legendFormat": "GPU {{ gpu }} core",
"refId": "A"
},
{
"expr": "DCGM_FI_DEV_MEMORY_TEMP",
"legendFormat": "GPU {{ gpu }} mem",
"refId": "B"
}
],
"title": "GPU Temperature",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "${DS_PROMETHEUS}"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"drawStyle": "line",
"fillOpacity": 15,
"lineInterpolation": "smooth",
"lineWidth": 2,
"pointSize": 5,
"showPoints": "never",
"spanNulls": false,
"stacking": {
"mode": "none"
}
},
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": null
}
]
},
"unit": "watt"
},
"overrides": []
},
"gridPos": {
"h": 8,
"w": 12,
"x": 0,
"y": 55
},
"id": 44,
"options": {
"legend": {
"displayMode": "list",
"placement": "bottom"
},
"tooltip": {
"mode": "multi",
"sort": "desc"
}
},
"targets": [
{
"expr": "DCGM_FI_DEV_POWER_USAGE",
"legendFormat": "GPU {{ gpu }}",
"refId": "A"
}
],
"title": "GPU Power Draw",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "${DS_PROMETHEUS}"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"drawStyle": "line",
"fillOpacity": 15,
"lineInterpolation": "smooth",
"lineWidth": 1,
"pointSize": 5,
"showPoints": "never",
"spanNulls": false,
"stacking": {
"mode": "none"
}
},
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": null
}
]
},
"unit": "Bps"
},
"overrides": []
},
"gridPos": {
"h": 8,
"w": 12,
"x": 12,
"y": 55
},
"id": 45,
"options": {
"legend": {
"displayMode": "list",
"placement": "bottom"
},
"tooltip": {
"mode": "multi",
"sort": "desc"
}
},
"targets": [
{
"expr": "DCGM_FI_PROF_PCIE_RX_BYTES",
"legendFormat": "GPU {{ gpu }} rx",
"refId": "A"
},
{
"expr": "DCGM_FI_PROF_PCIE_TX_BYTES",
"legendFormat": "GPU {{ gpu }} tx",
"refId": "B"
}
],
"title": "GPU PCIe Throughput",
"type": "timeseries"
},
{
"collapsed": false,
"gridPos": {
"h": 1,
"w": 24,
"x": 0,
"y": 63
},
"id": 50,
"panels": [],
"title": "Prometheus Health",
"type": "row"
},
{
"datasource": {
"type": "prometheus",
"uid": "${DS_PROMETHEUS}"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "thresholds"
},
"mappings": [],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": null
},
{
"color": "yellow",
"value": 500000
},
{
"color": "red",
"value": 1000000
}
]
},
"unit": "short"
},
"overrides": []
},
"gridPos": {
"h": 4,
"w": 6,
"x": 0,
"y": 64
},
"id": 51,
"options": {
"colorMode": "value",
"graphMode": "area",
"justifyMode": "auto",
"orientation": "auto",
"reduceOptions": {
"calcs": [
"lastNotNull"
],
"fields": "",
"values": false
},
"textMode": "auto"
},
"targets": [
{
"expr": "prometheus_tsdb_head_series",
"legendFormat": "",
"refId": "A"
}
],
"title": "Active Time Series",
"type": "stat"
},
{
"datasource": {
"type": "prometheus",
"uid": "${DS_PROMETHEUS}"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "thresholds"
},
"mappings": [],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": null
},
{
"color": "yellow",
"value": 5368709120
},
{
"color": "red",
"value": 10737418240
}
]
},
"unit": "bytes"
},
"overrides": []
},
"gridPos": {
"h": 4,
"w": 6,
"x": 6,
"y": 64
},
"id": 52,
"options": {
"colorMode": "value",
"graphMode": "area",
"justifyMode": "auto",
"orientation": "auto",
"reduceOptions": {
"calcs": [
"lastNotNull"
],
"fields": "",
"values": false
},
"textMode": "auto"
},
"targets": [
{
"expr": "prometheus_tsdb_storage_blocks_bytes",
"legendFormat": "",
"refId": "A"
}
],
"title": "TSDB Storage Size",
"type": "stat"
},
{
"datasource": {
"type": "prometheus",
"uid": "${DS_PROMETHEUS}"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "thresholds"
},
"mappings": [],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": null
}
]
}
},
"overrides": [
{
"matcher": {
"id": "byName",
"options": "down"
},
"properties": [
{
"id": "color",
"value": {
"fixedColor": "red",
"mode": "fixed"
}
}
]
}
]
},
"gridPos": {
"h": 4,
"w": 6,
"x": 12,
"y": 64
},
"id": 53,
"options": {
"colorMode": "value",
"graphMode": "none",
"justifyMode": "auto",
"orientation": "auto",
"reduceOptions": {
"calcs": [
"lastNotNull"
],
"fields": "",
"values": false
},
"textMode": "value_and_name"
},
"targets": [
{
"expr": "count(up == 1)",
"legendFormat": "up",
"refId": "A"
},
{
"expr": "count(up == 0)",
"legendFormat": "down",
"refId": "B"
}
],
"title": "Scrape Targets Up",
"type": "stat"
},
{
"datasource": {
"type": "prometheus",
"uid": "${DS_PROMETHEUS}"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"drawStyle": "line",
"fillOpacity": 10,
"lineInterpolation": "smooth",
"lineWidth": 1,
"pointSize": 5,
"showPoints": "never",
"spanNulls": false,
"stacking": {
"mode": "none"
}
},
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": null
},
{
"color": "red",
"value": 5
}
]
},
"unit": "s"
},
"overrides": []
},
"gridPos": {
"h": 8,
"w": 12,
"x": 0,
"y": 68
},
"id": 54,
"options": {
"legend": {
"calcs": [
"mean",
"max"
],
"displayMode": "table",
"placement": "bottom"
},
"tooltip": {
"mode": "multi",
"sort": "desc"
}
},
"targets": [
{
"expr": "scrape_duration_seconds",
"legendFormat": "{{ job }} / {{ instance }}",
"refId": "A"
}
],
"title": "Scrape Duration by Job",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "${DS_PROMETHEUS}"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "thresholds"
},
"mappings": [
{
"options": {
"0": {
"color": "red",
"text": "DOWN"
},
"1": {
"color": "green",
"text": "UP"
}
},
"type": "value"
}
],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "red",
"value": null
},
{
"color": "green",
"value": 1
}
]
}
},
"overrides": [
{
"matcher": {
"id": "byName",
"options": "Time"
},
"properties": [
{
"id": "custom.hidden",
"value": true
}
]
},
{
"matcher": {
"id": "byName",
"options": "__name__"
},
"properties": [
{
"id": "custom.hidden",
"value": true
}
]
},
{
"matcher": {
"id": "byName",
"options": "Value"
},
"properties": [
{
"id": "custom.cellOptions",
"value": {
"type": "color-background"
}
}
]
}
]
},
"gridPos": {
"h": 8,
"w": 12,
"x": 12,
"y": 68
},
"id": 55,
"options": {
"footer": {
"show": false
},
"showHeader": true,
"sortBy": [
{
"desc": false,
"displayName": "Value"
}
]
},
"targets": [
{
"expr": "up",
"format": "table",
"instant": true,
"legendFormat": "",
"refId": "A"
}
],
"title": "Scrape Targets Status",
"transformations": [
{
"id": "organize",
"options": {
"excludeByName": {
"Time": true,
"__name__": true
},
"renameByName": {
"Value": "Status",
"instance": "Instance",
"job": "Job"
}
}
}
],
"type": "table"
}
],
"refresh": "30s",
"schemaVersion": 39,
"style": "dark",
"tags": [
"foxhunt",
"infrastructure",
"k8s"
],
"templating": {
"list": [
{
"auto": true,
"auto_count": 30,
"auto_min": "1m",
"current": {
"selected": true,
"text": "auto",
"value": "$__auto_interval_interval"
},
"hide": 0,
"name": "interval",
"query": "1m,5m,15m,30m,1h",
"refresh": 2,
"type": "interval"
},
{
"allValue": ".*",
"current": {
"selected": true,
"text": "All",
"value": "$__all"
},
"datasource": {
"type": "prometheus",
"uid": "${DS_PROMETHEUS}"
},
"hide": 0,
"includeAll": true,
"multi": true,
"name": "node",
"query": "label_values(node_uname_info, nodename)",
"refresh": 2,
"sort": 1,
"type": "query"
}
]
},
"time": {
"from": "now-3h",
"to": "now"
},
"timezone": "browser",
"title": "Infrastructure & Cluster",
"uid": "foxhunt-infrastructure",
"weekStart": "",
"__inputs": [
{
"name": "DS_PROMETHEUS",
"type": "datasource",
"pluginId": "prometheus"
},
{
"name": "DS_LOKI",
"type": "datasource",
"pluginId": "loki"
},
{
"name": "DS_TEMPO",
"type": "datasource",
"pluginId": "${DS_TEMPO}"
}
]
}
Reference in New Issue View Git Blame Copy Permalink