mirror of
https://github.com/Alkatrazz24/Funk-lab.git
synced 2026-07-08 13:34:43 +02:00
feat(monitoring): dashboard infra — 10 nouveaux panels
Compute (compute-01/02/03) : - Uptime min (dtdurations, jaune<12h / vert>12h / bleu>7j) - Disk /var % avg (xfs, seuil 60%/85%) storage-01 : - RAID5 état (mapping 1→ACTIF vert / 0→DÉGRADÉ rouge) - RAID5 disques HS (vert=0 / rouge>=1) - Uptime (dtdurations) - Temp CPU (hwmon pci AMD, seuil 60°/80°) gpu-01 : - Uptime (dtdurations) - Temp CPU Tdie (hwmon AMD, seuil 70°/90°) - Swap % (16GB configuré, actuellement 0%) - Timeseries GPU edge/hotspot/VRAM (hwmon RX 6700XT) - Timeseries CPU Tdie/Tctl (hwmon AMD) Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
This commit is contained in:
alkatrazz
parent
a21b986d54
commit
428fa4ce2a
1 changed files with 452 additions and 2 deletions
|
|
@ -142,6 +142,84 @@ data:
|
|||
],
|
||||
"type": "stat"
|
||||
},
|
||||
{
|
||||
"datasource": {"type": "prometheus", "uid": "${datasource}"},
|
||||
"fieldConfig": {
|
||||
"defaults": {
|
||||
"color": {"mode": "thresholds"},
|
||||
"thresholds": {
|
||||
"mode": "absolute",
|
||||
"steps": [
|
||||
{"color": "yellow", "value": null},
|
||||
{"color": "green", "value": 43200},
|
||||
{"color": "blue", "value": 604800}
|
||||
]
|
||||
},
|
||||
"unit": "dtdurations",
|
||||
"mappings": []
|
||||
},
|
||||
"overrides": []
|
||||
},
|
||||
"gridPos": {"h": 4, "w": 6, "x": 12, "y": 1},
|
||||
"id": 40,
|
||||
"options": {
|
||||
"reduceOptions": {"calcs": ["min"], "fields": "", "values": false},
|
||||
"orientation": "auto",
|
||||
"textMode": "auto",
|
||||
"colorMode": "background",
|
||||
"graphMode": "none"
|
||||
},
|
||||
"title": "Uptime min — compute",
|
||||
"targets": [
|
||||
{
|
||||
"datasource": {"type": "prometheus", "uid": "${datasource}"},
|
||||
"expr": "time() - node_boot_time_seconds{job=\"node-exporter\"}",
|
||||
"legendFormat": "{{instance}}",
|
||||
"refId": "A"
|
||||
}
|
||||
],
|
||||
"type": "stat"
|
||||
},
|
||||
{
|
||||
"datasource": {"type": "prometheus", "uid": "${datasource}"},
|
||||
"fieldConfig": {
|
||||
"defaults": {
|
||||
"color": {"mode": "thresholds"},
|
||||
"thresholds": {
|
||||
"mode": "absolute",
|
||||
"steps": [
|
||||
{"color": "green", "value": null},
|
||||
{"color": "yellow", "value": 60},
|
||||
{"color": "red", "value": 85}
|
||||
]
|
||||
},
|
||||
"unit": "percent",
|
||||
"min": 0,
|
||||
"max": 100,
|
||||
"mappings": []
|
||||
},
|
||||
"overrides": []
|
||||
},
|
||||
"gridPos": {"h": 4, "w": 6, "x": 18, "y": 1},
|
||||
"id": 41,
|
||||
"options": {
|
||||
"reduceOptions": {"calcs": ["lastNotNull"], "fields": "", "values": false},
|
||||
"orientation": "auto",
|
||||
"textMode": "auto",
|
||||
"colorMode": "background",
|
||||
"graphMode": "none"
|
||||
},
|
||||
"title": "Disk /var avg % — compute",
|
||||
"targets": [
|
||||
{
|
||||
"datasource": {"type": "prometheus", "uid": "${datasource}"},
|
||||
"expr": "avg(100 - (node_filesystem_avail_bytes{job=\"node-exporter\",mountpoint=\"/var\",fstype=\"xfs\"} / node_filesystem_size_bytes{job=\"node-exporter\",mountpoint=\"/var\",fstype=\"xfs\"} * 100))",
|
||||
"legendFormat": "/var avg",
|
||||
"refId": "A"
|
||||
}
|
||||
],
|
||||
"type": "stat"
|
||||
},
|
||||
{
|
||||
"datasource": {"type": "prometheus", "uid": "${datasource}"},
|
||||
"fieldConfig": {
|
||||
|
|
@ -365,12 +443,167 @@ data:
|
|||
"colorMode": "background",
|
||||
"graphMode": "none"
|
||||
},
|
||||
"title": "Disk /srv usage % — storage-01",
|
||||
"title": "Disk /srv/data % — storage-01",
|
||||
"targets": [
|
||||
{
|
||||
"datasource": {"type": "prometheus", "uid": "${datasource}"},
|
||||
"expr": "100 - (node_filesystem_avail_bytes{job=\"storage-01\",mountpoint=\"/srv/data\",fstype!=\"tmpfs\"} / node_filesystem_size_bytes{job=\"storage-01\",mountpoint=\"/srv/data\",fstype!=\"tmpfs\"} * 100)",
|
||||
"legendFormat": "/srv",
|
||||
"legendFormat": "/srv/data",
|
||||
"refId": "A"
|
||||
}
|
||||
],
|
||||
"type": "stat"
|
||||
},
|
||||
{
|
||||
"datasource": {"type": "prometheus", "uid": "${datasource}"},
|
||||
"fieldConfig": {
|
||||
"defaults": {
|
||||
"color": {"mode": "thresholds"},
|
||||
"thresholds": {"mode": "absolute", "steps": [{"color": "red", "value": null}, {"color": "green", "value": 1}]},
|
||||
"unit": "none",
|
||||
"mappings": [
|
||||
{
|
||||
"type": "value",
|
||||
"options": {
|
||||
"1": {"text": "ACTIF", "color": "green", "index": 0},
|
||||
"0": {"text": "DÉGRADÉ", "color": "red", "index": 1}
|
||||
}
|
||||
}
|
||||
]
|
||||
},
|
||||
"overrides": []
|
||||
},
|
||||
"gridPos": {"h": 4, "w": 3, "x": 12, "y": 14},
|
||||
"id": 42,
|
||||
"options": {
|
||||
"reduceOptions": {"calcs": ["lastNotNull"], "fields": "", "values": false},
|
||||
"orientation": "auto",
|
||||
"textMode": "auto",
|
||||
"colorMode": "background",
|
||||
"graphMode": "none"
|
||||
},
|
||||
"title": "RAID5 état",
|
||||
"targets": [
|
||||
{
|
||||
"datasource": {"type": "prometheus", "uid": "${datasource}"},
|
||||
"expr": "node_md_state{job=\"storage-01\",device=\"md0\",state=\"active\"}",
|
||||
"legendFormat": "md0",
|
||||
"refId": "A"
|
||||
}
|
||||
],
|
||||
"type": "stat"
|
||||
},
|
||||
{
|
||||
"datasource": {"type": "prometheus", "uid": "${datasource}"},
|
||||
"fieldConfig": {
|
||||
"defaults": {
|
||||
"color": {"mode": "thresholds"},
|
||||
"thresholds": {
|
||||
"mode": "absolute",
|
||||
"steps": [
|
||||
{"color": "green", "value": null},
|
||||
{"color": "red", "value": 1}
|
||||
]
|
||||
},
|
||||
"unit": "none",
|
||||
"mappings": [],
|
||||
"noValue": "0"
|
||||
},
|
||||
"overrides": []
|
||||
},
|
||||
"gridPos": {"h": 4, "w": 3, "x": 15, "y": 14},
|
||||
"id": 43,
|
||||
"options": {
|
||||
"reduceOptions": {"calcs": ["lastNotNull"], "fields": "", "values": false},
|
||||
"orientation": "auto",
|
||||
"textMode": "auto",
|
||||
"colorMode": "background",
|
||||
"graphMode": "none"
|
||||
},
|
||||
"title": "RAID5 disques HS",
|
||||
"targets": [
|
||||
{
|
||||
"datasource": {"type": "prometheus", "uid": "${datasource}"},
|
||||
"expr": "node_md_disks{job=\"storage-01\",device=\"md0\",state=\"failed\"}",
|
||||
"legendFormat": "failed",
|
||||
"refId": "A"
|
||||
}
|
||||
],
|
||||
"type": "stat"
|
||||
},
|
||||
{
|
||||
"datasource": {"type": "prometheus", "uid": "${datasource}"},
|
||||
"fieldConfig": {
|
||||
"defaults": {
|
||||
"color": {"mode": "thresholds"},
|
||||
"thresholds": {
|
||||
"mode": "absolute",
|
||||
"steps": [
|
||||
{"color": "yellow", "value": null},
|
||||
{"color": "green", "value": 43200},
|
||||
{"color": "blue", "value": 604800}
|
||||
]
|
||||
},
|
||||
"unit": "dtdurations",
|
||||
"mappings": []
|
||||
},
|
||||
"overrides": []
|
||||
},
|
||||
"gridPos": {"h": 4, "w": 3, "x": 18, "y": 14},
|
||||
"id": 44,
|
||||
"options": {
|
||||
"reduceOptions": {"calcs": ["lastNotNull"], "fields": "", "values": false},
|
||||
"orientation": "auto",
|
||||
"textMode": "auto",
|
||||
"colorMode": "background",
|
||||
"graphMode": "none"
|
||||
},
|
||||
"title": "Uptime — storage-01",
|
||||
"targets": [
|
||||
{
|
||||
"datasource": {"type": "prometheus", "uid": "${datasource}"},
|
||||
"expr": "time() - node_boot_time_seconds{job=\"storage-01\"}",
|
||||
"legendFormat": "uptime",
|
||||
"refId": "A"
|
||||
}
|
||||
],
|
||||
"type": "stat"
|
||||
},
|
||||
{
|
||||
"datasource": {"type": "prometheus", "uid": "${datasource}"},
|
||||
"fieldConfig": {
|
||||
"defaults": {
|
||||
"color": {"mode": "thresholds"},
|
||||
"thresholds": {
|
||||
"mode": "absolute",
|
||||
"steps": [
|
||||
{"color": "green", "value": null},
|
||||
{"color": "yellow", "value": 60},
|
||||
{"color": "red", "value": 80}
|
||||
]
|
||||
},
|
||||
"unit": "celsius",
|
||||
"min": 0,
|
||||
"max": 100,
|
||||
"mappings": []
|
||||
},
|
||||
"overrides": []
|
||||
},
|
||||
"gridPos": {"h": 4, "w": 3, "x": 21, "y": 14},
|
||||
"id": 45,
|
||||
"options": {
|
||||
"reduceOptions": {"calcs": ["lastNotNull"], "fields": "", "values": false},
|
||||
"orientation": "auto",
|
||||
"textMode": "auto",
|
||||
"colorMode": "background",
|
||||
"graphMode": "none"
|
||||
},
|
||||
"title": "Temp CPU — storage-01",
|
||||
"targets": [
|
||||
{
|
||||
"datasource": {"type": "prometheus", "uid": "${datasource}"},
|
||||
"expr": "node_hwmon_temp_celsius{job=\"storage-01\",chip=\"pci0000:00_0000:00:18_3\",sensor=\"temp1\"}",
|
||||
"legendFormat": "CPU",
|
||||
"refId": "A"
|
||||
}
|
||||
],
|
||||
|
|
@ -609,6 +842,124 @@ data:
|
|||
],
|
||||
"type": "stat"
|
||||
},
|
||||
{
|
||||
"datasource": {"type": "prometheus", "uid": "${datasource}"},
|
||||
"fieldConfig": {
|
||||
"defaults": {
|
||||
"color": {"mode": "thresholds"},
|
||||
"thresholds": {
|
||||
"mode": "absolute",
|
||||
"steps": [
|
||||
{"color": "yellow", "value": null},
|
||||
{"color": "green", "value": 43200},
|
||||
{"color": "blue", "value": 604800}
|
||||
]
|
||||
},
|
||||
"unit": "dtdurations",
|
||||
"mappings": []
|
||||
},
|
||||
"overrides": []
|
||||
},
|
||||
"gridPos": {"h": 4, "w": 5, "x": 8, "y": 27},
|
||||
"id": 46,
|
||||
"options": {
|
||||
"reduceOptions": {"calcs": ["lastNotNull"], "fields": "", "values": false},
|
||||
"orientation": "auto",
|
||||
"textMode": "auto",
|
||||
"colorMode": "background",
|
||||
"graphMode": "none"
|
||||
},
|
||||
"title": "Uptime — gpu-01",
|
||||
"targets": [
|
||||
{
|
||||
"datasource": {"type": "prometheus", "uid": "${datasource}"},
|
||||
"expr": "time() - node_boot_time_seconds{job=\"gpu-01-node\"}",
|
||||
"legendFormat": "uptime",
|
||||
"refId": "A"
|
||||
}
|
||||
],
|
||||
"type": "stat"
|
||||
},
|
||||
{
|
||||
"datasource": {"type": "prometheus", "uid": "${datasource}"},
|
||||
"fieldConfig": {
|
||||
"defaults": {
|
||||
"color": {"mode": "thresholds"},
|
||||
"thresholds": {
|
||||
"mode": "absolute",
|
||||
"steps": [
|
||||
{"color": "green", "value": null},
|
||||
{"color": "yellow", "value": 70},
|
||||
{"color": "red", "value": 90}
|
||||
]
|
||||
},
|
||||
"unit": "celsius",
|
||||
"min": 0,
|
||||
"max": 110,
|
||||
"mappings": []
|
||||
},
|
||||
"overrides": []
|
||||
},
|
||||
"gridPos": {"h": 4, "w": 5, "x": 13, "y": 27},
|
||||
"id": 47,
|
||||
"options": {
|
||||
"reduceOptions": {"calcs": ["lastNotNull"], "fields": "", "values": false},
|
||||
"orientation": "auto",
|
||||
"textMode": "auto",
|
||||
"colorMode": "background",
|
||||
"graphMode": "none"
|
||||
},
|
||||
"title": "Temp CPU Tdie — gpu-01",
|
||||
"targets": [
|
||||
{
|
||||
"datasource": {"type": "prometheus", "uid": "${datasource}"},
|
||||
"expr": "node_hwmon_temp_celsius{job=\"gpu-01-node\",chip=\"pci0000:00_0000:00:18_3\",sensor=\"temp3\"}",
|
||||
"legendFormat": "Tdie",
|
||||
"refId": "A"
|
||||
}
|
||||
],
|
||||
"type": "stat"
|
||||
},
|
||||
{
|
||||
"datasource": {"type": "prometheus", "uid": "${datasource}"},
|
||||
"fieldConfig": {
|
||||
"defaults": {
|
||||
"color": {"mode": "thresholds"},
|
||||
"thresholds": {
|
||||
"mode": "absolute",
|
||||
"steps": [
|
||||
{"color": "green", "value": null},
|
||||
{"color": "yellow", "value": 70},
|
||||
{"color": "red", "value": 90}
|
||||
]
|
||||
},
|
||||
"unit": "percent",
|
||||
"min": 0,
|
||||
"max": 100,
|
||||
"mappings": []
|
||||
},
|
||||
"overrides": []
|
||||
},
|
||||
"gridPos": {"h": 4, "w": 6, "x": 18, "y": 27},
|
||||
"id": 48,
|
||||
"options": {
|
||||
"reduceOptions": {"calcs": ["lastNotNull"], "fields": "", "values": false},
|
||||
"orientation": "auto",
|
||||
"textMode": "auto",
|
||||
"colorMode": "background",
|
||||
"graphMode": "none"
|
||||
},
|
||||
"title": "Swap % — gpu-01",
|
||||
"targets": [
|
||||
{
|
||||
"datasource": {"type": "prometheus", "uid": "${datasource}"},
|
||||
"expr": "(1 - node_memory_SwapFree_bytes{job=\"gpu-01-node\"} / node_memory_SwapTotal_bytes{job=\"gpu-01-node\"}) * 100",
|
||||
"legendFormat": "Swap",
|
||||
"refId": "A"
|
||||
}
|
||||
],
|
||||
"type": "stat"
|
||||
},
|
||||
{
|
||||
"datasource": {"type": "prometheus", "uid": "${datasource}"},
|
||||
"fieldConfig": {
|
||||
|
|
@ -772,6 +1123,105 @@ data:
|
|||
}
|
||||
],
|
||||
"type": "timeseries"
|
||||
},
|
||||
{
|
||||
"datasource": {"type": "prometheus", "uid": "${datasource}"},
|
||||
"fieldConfig": {
|
||||
"defaults": {
|
||||
"color": {"mode": "palette-classic"},
|
||||
"custom": {
|
||||
"lineWidth": 1,
|
||||
"fillOpacity": 8,
|
||||
"drawStyle": "line",
|
||||
"spanNulls": false
|
||||
},
|
||||
"unit": "celsius",
|
||||
"min": 0,
|
||||
"max": 110
|
||||
},
|
||||
"overrides": [
|
||||
{
|
||||
"matcher": {"id": "byName", "options": "edge"},
|
||||
"properties": [{"id": "color", "value": {"fixedColor": "blue", "mode": "fixed"}}]
|
||||
},
|
||||
{
|
||||
"matcher": {"id": "byName", "options": "hotspot"},
|
||||
"properties": [{"id": "color", "value": {"fixedColor": "orange", "mode": "fixed"}}]
|
||||
},
|
||||
{
|
||||
"matcher": {"id": "byName", "options": "VRAM"},
|
||||
"properties": [{"id": "color", "value": {"fixedColor": "purple", "mode": "fixed"}}]
|
||||
}
|
||||
]
|
||||
},
|
||||
"gridPos": {"h": 8, "w": 12, "x": 0, "y": 47},
|
||||
"id": 49,
|
||||
"options": {
|
||||
"tooltip": {"mode": "multi"},
|
||||
"legend": {"displayMode": "list", "placement": "bottom"}
|
||||
},
|
||||
"title": "GPU Températures (edge / hotspot / VRAM) — gpu-01",
|
||||
"targets": [
|
||||
{
|
||||
"datasource": {"type": "prometheus", "uid": "${datasource}"},
|
||||
"expr": "node_hwmon_temp_celsius{job=\"gpu-01-node\",chip=\"0000:09:00_0_0000:0a:00_0\",sensor=\"temp1\"}",
|
||||
"legendFormat": "edge",
|
||||
"refId": "A"
|
||||
},
|
||||
{
|
||||
"datasource": {"type": "prometheus", "uid": "${datasource}"},
|
||||
"expr": "node_hwmon_temp_celsius{job=\"gpu-01-node\",chip=\"0000:09:00_0_0000:0a:00_0\",sensor=\"temp2\"}",
|
||||
"legendFormat": "hotspot",
|
||||
"refId": "B"
|
||||
},
|
||||
{
|
||||
"datasource": {"type": "prometheus", "uid": "${datasource}"},
|
||||
"expr": "node_hwmon_temp_celsius{job=\"gpu-01-node\",chip=\"0000:09:00_0_0000:0a:00_0\",sensor=\"temp3\"}",
|
||||
"legendFormat": "VRAM",
|
||||
"refId": "C"
|
||||
}
|
||||
],
|
||||
"type": "timeseries"
|
||||
},
|
||||
{
|
||||
"datasource": {"type": "prometheus", "uid": "${datasource}"},
|
||||
"fieldConfig": {
|
||||
"defaults": {
|
||||
"color": {"mode": "palette-classic"},
|
||||
"custom": {
|
||||
"lineWidth": 1,
|
||||
"fillOpacity": 8,
|
||||
"drawStyle": "line",
|
||||
"spanNulls": false
|
||||
},
|
||||
"unit": "celsius",
|
||||
"min": 0,
|
||||
"max": 110
|
||||
},
|
||||
"overrides": []
|
||||
},
|
||||
"gridPos": {"h": 8, "w": 12, "x": 12, "y": 47},
|
||||
"id": 50,
|
||||
"options": {
|
||||
"tooltip": {"mode": "multi"},
|
||||
"legend": {"displayMode": "list", "placement": "bottom"}
|
||||
},
|
||||
"title": "CPU Températures — gpu-01 (AMD Tdie)",
|
||||
"targets": [
|
||||
{
|
||||
"datasource": {"type": "prometheus", "uid": "${datasource}"},
|
||||
"expr": "node_hwmon_temp_celsius{job=\"gpu-01-node\",chip=\"pci0000:00_0000:00:18_3\",sensor=\"temp3\"}",
|
||||
"legendFormat": "Tdie",
|
||||
"refId": "A"
|
||||
},
|
||||
{
|
||||
"datasource": {"type": "prometheus", "uid": "${datasource}"},
|
||||
"expr": "node_hwmon_temp_celsius{job=\"gpu-01-node\",chip=\"pci0000:00_0000:00:18_3\",sensor=\"temp1\"}",
|
||||
"legendFormat": "Tctl",
|
||||
"refId": "B"
|
||||
}
|
||||
],
|
||||
"type": "timeseries"
|
||||
}
|
||||
],
|
||||
"refresh": "30s",
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue