livrare lot 2
This commit is contained in:
commit
8ecc78e729
763 changed files with 164593 additions and 0 deletions
105
backend/observability/grafana/dashboards/didi-overview.json
Normal file
105
backend/observability/grafana/dashboards/didi-overview.json
Normal file
|
|
@ -0,0 +1,105 @@
|
|||
{
|
||||
"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": "DiDi Platform — overall health, request rates, latency, errors, infrastructure",
|
||||
"editable": true,
|
||||
"graphTooltip": 1,
|
||||
"id": null,
|
||||
"panels": [
|
||||
{ "type": "row", "title": "Service Health", "gridPos": { "h": 1, "w": 24, "x": 0, "y": 0 }, "id": 100, "collapsed": false },
|
||||
{
|
||||
"type": "stat", "title": "Services UP", "id": 1,
|
||||
"datasource": { "type": "prometheus", "uid": "prometheus" },
|
||||
"gridPos": { "h": 4, "w": 4, "x": 0, "y": 1 },
|
||||
"targets": [ { "expr": "count(up == 1)", "refId": "A" } ],
|
||||
"fieldConfig": { "defaults": { "color": { "mode": "thresholds" }, "thresholds": { "mode": "absolute", "steps": [ { "color": "red", "value": null }, { "color": "green", "value": 5 } ] } } },
|
||||
"options": { "colorMode": "background", "graphMode": "area", "reduceOptions": { "calcs": ["lastNotNull"], "fields": "", "values": false }, "textMode": "auto" }
|
||||
},
|
||||
{
|
||||
"type": "stat", "title": "Services DOWN", "id": 2,
|
||||
"datasource": { "type": "prometheus", "uid": "prometheus" },
|
||||
"gridPos": { "h": 4, "w": 4, "x": 4, "y": 1 },
|
||||
"targets": [ { "expr": "count(up == 0)", "refId": "A" } ],
|
||||
"fieldConfig": { "defaults": { "color": { "mode": "thresholds" }, "thresholds": { "mode": "absolute", "steps": [ { "color": "green", "value": null }, { "color": "red", "value": 1 } ] } } },
|
||||
"options": { "colorMode": "background", "graphMode": "none", "reduceOptions": { "calcs": ["lastNotNull"], "fields": "", "values": false }, "textMode": "auto" }
|
||||
},
|
||||
{
|
||||
"type": "stat", "title": "HTTP req/sec (total)", "id": 3,
|
||||
"datasource": { "type": "prometheus", "uid": "prometheus" },
|
||||
"gridPos": { "h": 4, "w": 8, "x": 8, "y": 1 },
|
||||
"targets": [ { "expr": "sum(rate(http_requests_total[5m]))", "refId": "A" } ],
|
||||
"fieldConfig": { "defaults": { "unit": "reqps", "color": { "mode": "thresholds" } } },
|
||||
"options": { "colorMode": "value", "graphMode": "area", "reduceOptions": { "calcs": ["lastNotNull"], "fields": "", "values": false } }
|
||||
},
|
||||
{
|
||||
"type": "stat", "title": "Active Workers (agent-v3)", "id": 4,
|
||||
"datasource": { "type": "prometheus", "uid": "prometheus" },
|
||||
"gridPos": { "h": 4, "w": 8, "x": 16, "y": 1 },
|
||||
"targets": [ { "expr": "count(up{job=\"agent-v3-workers\"} == 1)", "refId": "A" } ],
|
||||
"fieldConfig": { "defaults": { "color": { "mode": "thresholds" }, "thresholds": { "mode": "absolute", "steps": [ { "color": "red", "value": null }, { "color": "green", "value": 10 } ] } } },
|
||||
"options": { "colorMode": "background", "graphMode": "none", "reduceOptions": { "calcs": ["lastNotNull"], "fields": "", "values": false } }
|
||||
},
|
||||
{ "type": "row", "title": "HTTP traffic", "gridPos": { "h": 1, "w": 24, "x": 0, "y": 5 }, "id": 101, "collapsed": false },
|
||||
{
|
||||
"type": "timeseries", "title": "Request rate per service", "id": 10,
|
||||
"datasource": { "type": "prometheus", "uid": "prometheus" },
|
||||
"gridPos": { "h": 8, "w": 12, "x": 0, "y": 6 },
|
||||
"targets": [ { "expr": "sum by (service) (rate(http_requests_total[1m]))", "refId": "A", "legendFormat": "{{service}}" } ],
|
||||
"fieldConfig": { "defaults": { "unit": "reqps" } },
|
||||
"options": { "legend": { "displayMode": "table", "placement": "bottom", "calcs": ["mean", "max"] } }
|
||||
},
|
||||
{
|
||||
"type": "timeseries", "title": "p95 latency per service (s)", "id": 11,
|
||||
"datasource": { "type": "prometheus", "uid": "prometheus" },
|
||||
"gridPos": { "h": 8, "w": 12, "x": 12, "y": 6 },
|
||||
"targets": [ { "expr": "histogram_quantile(0.95, sum by (service, le) (rate(http_request_duration_seconds_bucket[5m])))", "refId": "A", "legendFormat": "{{service}}" } ],
|
||||
"fieldConfig": { "defaults": { "unit": "s" } },
|
||||
"options": { "legend": { "displayMode": "table", "placement": "bottom", "calcs": ["mean", "max"] } }
|
||||
},
|
||||
{
|
||||
"type": "timeseries", "title": "5xx error rate", "id": 12,
|
||||
"datasource": { "type": "prometheus", "uid": "prometheus" },
|
||||
"gridPos": { "h": 8, "w": 12, "x": 0, "y": 14 },
|
||||
"targets": [ { "expr": "sum by (service) (rate(http_requests_total{status_code=~\"5..\"}[5m]))", "refId": "A", "legendFormat": "{{service}}" } ],
|
||||
"fieldConfig": { "defaults": { "unit": "reqps", "color": { "mode": "palette-classic" } } }
|
||||
},
|
||||
{
|
||||
"type": "timeseries", "title": "Process RSS memory (MB)", "id": 13,
|
||||
"datasource": { "type": "prometheus", "uid": "prometheus" },
|
||||
"gridPos": { "h": 8, "w": 12, "x": 12, "y": 14 },
|
||||
"targets": [ { "expr": "process_resident_memory_bytes / 1024 / 1024", "refId": "A", "legendFormat": "{{service}}" } ],
|
||||
"fieldConfig": { "defaults": { "unit": "MB" } }
|
||||
},
|
||||
{ "type": "row", "title": "Infrastructure", "gridPos": { "h": 1, "w": 24, "x": 0, "y": 22 }, "id": 103, "collapsed": false },
|
||||
{
|
||||
"type": "timeseries", "title": "CPU per container", "id": 30,
|
||||
"datasource": { "type": "prometheus", "uid": "prometheus" },
|
||||
"gridPos": { "h": 8, "w": 12, "x": 0, "y": 23 },
|
||||
"targets": [ { "expr": "sum by (name) (rate(container_cpu_usage_seconds_total{name=~\"didi.*|agent-v3.*\"}[1m]))", "refId": "A", "legendFormat": "{{name}}" } ],
|
||||
"fieldConfig": { "defaults": { "unit": "percentunit" } },
|
||||
"options": { "legend": { "displayMode": "table", "placement": "bottom" } }
|
||||
},
|
||||
{
|
||||
"type": "timeseries", "title": "Memory per container (MB)", "id": 31,
|
||||
"datasource": { "type": "prometheus", "uid": "prometheus" },
|
||||
"gridPos": { "h": 8, "w": 12, "x": 12, "y": 23 },
|
||||
"targets": [ { "expr": "container_memory_usage_bytes{name=~\"didi.*|agent-v3.*\"} / 1024 / 1024", "refId": "A", "legendFormat": "{{name}}" } ],
|
||||
"fieldConfig": { "defaults": { "unit": "MB" } }
|
||||
},
|
||||
{
|
||||
"type": "timeseries", "title": "RabbitMQ queue depth", "id": 40,
|
||||
"datasource": { "type": "prometheus", "uid": "prometheus" },
|
||||
"gridPos": { "h": 8, "w": 24, "x": 0, "y": 31 },
|
||||
"targets": [ { "expr": "rabbitmq_queue_messages", "refId": "A", "legendFormat": "{{queue}}" } ],
|
||||
"options": { "legend": { "displayMode": "table", "placement": "bottom" } }
|
||||
}
|
||||
],
|
||||
"refresh": "30s",
|
||||
"schemaVersion": 39,
|
||||
"tags": ["didi", "overview"],
|
||||
"templating": { "list": [] },
|
||||
"time": { "from": "now-30m", "to": "now" },
|
||||
"timezone": "Europe/Bucharest",
|
||||
"title": "DiDi Platform Overview",
|
||||
"uid": "didi-overview",
|
||||
"version": 1
|
||||
}
|
||||
|
|
@ -0,0 +1,13 @@
|
|||
apiVersion: 1
|
||||
|
||||
providers:
|
||||
- name: 'DiDi Platform Dashboards'
|
||||
orgId: 1
|
||||
folder: 'DiDi'
|
||||
type: file
|
||||
disableDeletion: false
|
||||
updateIntervalSeconds: 30
|
||||
allowUiUpdates: true
|
||||
options:
|
||||
path: /var/lib/grafana/dashboards
|
||||
foldersFromFilesStructure: false
|
||||
|
|
@ -0,0 +1,40 @@
|
|||
apiVersion: 1
|
||||
|
||||
datasources:
|
||||
- name: Prometheus
|
||||
type: prometheus
|
||||
access: proxy
|
||||
url: http://didi-prometheus:9090
|
||||
isDefault: true
|
||||
editable: false
|
||||
jsonData:
|
||||
httpMethod: POST
|
||||
timeInterval: 30s
|
||||
|
||||
- name: Loki
|
||||
type: loki
|
||||
access: proxy
|
||||
url: http://didi-loki:3100
|
||||
editable: false
|
||||
jsonData:
|
||||
derivedFields:
|
||||
- name: TraceID
|
||||
matcherRegex: 'trace_id=(\w+)'
|
||||
url: '$${__value.raw}'
|
||||
datasourceUid: jaeger
|
||||
|
||||
- name: Jaeger
|
||||
type: jaeger
|
||||
uid: jaeger
|
||||
access: proxy
|
||||
url: http://didi-jaeger:16686
|
||||
editable: false
|
||||
jsonData:
|
||||
tracesToLogsV2:
|
||||
datasourceUid: loki
|
||||
tags: ['service.name']
|
||||
spanStartTimeShift: '-1h'
|
||||
spanEndTimeShift: '1h'
|
||||
tracesToMetrics:
|
||||
datasourceUid: prometheus
|
||||
tags: [{ key: 'service.name', value: 'service' }]
|
||||
Loading…
Add table
Add a link
Reference in a new issue