pjt222/correlate-observability-signals

Observability-Signale korrelieren

Verbinden metrics, logs, and traces for unified debugging across the three pillars of observability.

Wann verwenden

• Investigating complex incidents that span multiple systems
• Reducing MTTR (mean time to resolution)
• Building unified observability dashboards
• Implementing distributed tracing
• Moving from siloed tools to unified observability

Eingaben

• Erforderlich: Prometheus (metrics)
• Erforderlich: Log aggregation system (Loki, Elasticsearch, CloudWatch)
• Erforderlich: Distributed tracing backend (Tempo, Jaeger, Zipkin)
• Optional: Grafana for unified visualization
• Optional: OpenTelemetry instrumentation

Vorgehensweise

See Extended Examples for complete configuration files and templates.

Schritt 1: Implementieren Trace Context Propagation

Hinzufuegen trace IDs to all logs and metrics using OpenTelemetry:

// Go example: Propagate trace context to logs
package main

import (
    "context"
    "log"

    "go.opentelemetry.io/otel"
    "go.opentelemetry.io/otel/trace"
)

func handleRequest(ctx context.Context, userID string) {
    // Extract trace context
    span := trace.SpanFromContext(ctx)
    traceID := span.SpanContext().TraceID().String()

    // Include trace ID in structured logs
    log.Printf("trace_id=%s user_id=%s action=process_request", traceID, userID)

    // Business logic here
    processData(ctx, userID)
}

func processData(ctx context.Context, userID string) {
    tracer := otel.Tracer("my-service")
    ctx, span := tracer.Start(ctx, "processData")
    defer span.End()

    traceID := span.SpanContext().TraceID().String()
    log.Printf("trace_id=%s user_id=%s action=process_data", traceID, userID)

    // More work
}

Python example:

# Python: Flask with OpenTelemetry
from flask import Flask, request
from opentelemetry import trace
from opentelemetry.instrumentation.flask import FlaskInstrumentor
import logging

app = Flask(__name__)
FlaskInstrumentor().instrument_app(app)

logging.basicConfig(
    format='%(asctime)s trace_id=%(otelTraceID)s span_id=%(otelSpanID)s %(message)s',
    level=logging.INFO
)

@app.route('/api/users/<user_id>')
def get_user(user_id):
    span = trace.get_current_span()
    trace_id = format(span.get_span_context().trace_id, '032x')

    logging.info(f"Fetching user {user_id}", extra={
        'otelTraceID': trace_id,
        'otelSpanID': format(span.get_span_context().span_id, '016x')
    })

    # Business logic
    return {"user_id": user_id}

Erwartet: All logs include trace_id field, enabling log-to-trace correlation.

Bei Fehler: If trace IDs missing, check OpenTelemetry SDK initialization and context propagation.

Schritt 2: Konfigurieren Exemplars in Prometheus

Exemplars link metrics to traces:

# prometheus.yml
global:
  scrape_interval: 15s
  # Enable exemplar storage
  exemplars:
    max_exemplars: 100000  # Per TSDB block

scrape_configs:
  - job_name: 'api-service'
    static_configs:
      - targets: ['api-service:8080']
    # Scrape exemplars
    metric_relabel_configs:
      - source_labels: [__name__]
        regex: 'http_request_duration_seconds.*'
        action: keep

Instrument application to emit exemplars:

// Go: Emit exemplars with Prometheus histogram
package main

import (
    "github.com/prometheus/client_golang/prometheus"
    "github.com/prometheus/client_golang/prometheus/promauto"
    "go.opentelemetry.io/otel/trace"
)

var httpDuration = promauto.NewHistogramVec(
    prometheus.HistogramOpts{
        Name:    "http_request_duration_seconds",
        Help:    "HTTP request duration",
        Buckets: prometheus.DefBuckets,
    },
    []string{"method", "endpoint", "status"},
)

func recordRequest(ctx context.Context, method, endpoint, status string, duration float64) {
    // Get trace ID for exemplar
    span := trace.SpanFromContext(ctx)
    traceID := span.SpanContext().TraceID().String()

    // Record metric with exemplar
    observer := httpDuration.WithLabelValues(method, endpoint, status)
    observer.(prometheus.ExemplarObserver).ObserveWithExemplar(
        duration,
        prometheus.Labels{"trace_id": traceID},
    )
}

Query exemplars in Prometheus:

# Histogram with exemplars
histogram_quantile(0.95, rate(http_request_duration_seconds_bucket[5m]))

In Grafana, exemplars appear as dots on histogram graphs that link to traces.

Erwartet: Grafana shows exemplars in metric graphs, clicking opens corresponding trace.

Bei Fehler: Verifizieren Prometheus version ≥2.26 (exemplar support), check Grafana Datenquelle config enables exemplars.

Schritt 3: Erstellen Unified Dashboard with RED Method

RED Method: Rate, Errors, Duration (for services)

{
  "dashboard": {
    "title": "API Service - RED Dashboard",
    "panels": [
      {
        "title": "Request Rate (req/s)",
        "type": "graph",
        "targets": [
          {
            "expr": "sum(rate(http_requests_total{job=\"api-service\"}[5m])) by (endpoint)",
            "legendFormat": "{{ endpoint }}"
          }
        ],
        "exemplars": true
      },
      {
        "title": "Error Rate (%)",
        "type": "graph",
        "targets": [
          {
            "expr": "sum(rate(http_requests_total{job=\"api-service\", status=~\"5..\"}[5m])) / sum(rate(http_requests_total{job=\"api-service\"}[5m])) * 100",
            "legendFormat": "Error %"
          }
        ],
        "exemplars": true
      },
      {
        "title": "Request Duration (p50, p95, p99)",
        "type": "graph",
        "targets": [
          {
            "expr": "histogram_quantile(0.50, rate(http_request_duration_seconds_bucket{job=\"api-service\"}[5m]))",
            "legendFormat": "p50"
          },
          {
            "expr": "histogram_quantile(0.95, rate(http_request_duration_seconds_bucket{job=\"api-service\"}[5m]))",
            "legendFormat": "p95"
          },
          {
            "expr": "histogram_quantile(0.99, rate(http_request_duration_seconds_bucket{job=\"api-service\"}[5m]))",
            "legendFormat": "p99"
          }
        ],
        "exemplars": true
      },
      {
        "title": "Correlated Logs",
        "type": "logs",
        "datasource": "Loki",
        "targets": [
          {
            "expr": "{job=\"api-service\"} |= \"error\""
          }
        ],
        "options": {
          "showTime": true,
          "enableLogDetails": true
        }
      }
    ]
  }
}

Erwartet: Single dashboard showing rate, errors, duration + correlated logs.

Bei Fehler: If panels show "No Data", verify metric names match your instrumentation.

Schritt 4: Implementieren USE Method for Resources

USE Method: Utilization, Saturation, Errors (for resources like CPU, memory, disk)

{
  "dashboard": {
    "title": "Node Resources - USE Dashboard",
    "panels": [
      {
        "title": "CPU Utilization (%)",
        "type": "graph",
        "targets": [
          {
            "expr": "100 - (avg(rate(node_cpu_seconds_total{mode=\"idle\"}[5m])) * 100)",
            "legendFormat": "CPU Usage %"
          }
        ]
      },
      {
        "title": "CPU Saturation (Load Average)",
        "type": "graph",
        "targets": [
          {
            "expr": "node_load1",
            "legendFormat": "1min load"
          },
          {
            "expr": "node_load5",
            "legendFormat": "5min load"
          },
          {
            "expr": "count(node_cpu_seconds_total{mode=\"idle\"})",
            "legendFormat": "CPU cores (threshold)"
          }
        ]
      },
      {
        "title": "Memory Utilization (%)",
        "type": "graph",
        "targets": [
          {
            "expr": "(node_memory_MemTotal_bytes - node_memory_MemAvailable_bytes) / node_memory_MemTotal_bytes * 100",
            "legendFormat": "Memory Usage %"
          }
        ]
      },
      {
        "title": "Memory Saturation (Page Faults)",
        "type": "graph",
        "targets": [
          {
            "expr": "rate(node_vmstat_pgmajfault[5m])",
            "legendFormat": "Major page faults/s"
          }
        ]
      },
      {
        "title": "Disk Utilization (%)",
        "type": "graph",
        "targets": [
          {
            "expr": "(node_filesystem_size_bytes - node_filesystem_free_bytes) / node_filesystem_size_bytes * 100",
            "legendFormat": "{{ device }}"
          }
        ]
      },
      {
        "title": "Disk Saturation (IO Wait %)",
        "type": "graph",
        "targets": [
          {
            "expr": "rate(node_cpu_seconds_total{mode=\"iowait\"}[5m]) * 100",
            "legendFormat": "IO Wait %"
          }
        ]
      }
    ]
  }
}

Erwartet: Dashboard showing resource health across all USE dimensions.

Bei Fehler: Sicherstellen node_exporter is running and scraping system metrics.

Schritt 5: Link Logs to Traces in Loki

Konfigurieren Loki to extract trace IDs:

# loki-config.yml
schema_config:
  configs:
    - from: 2024-01-01
      store: boltdb-shipper
      object_store: s3
      schema: v11
      index:
        prefix: index_
        period: 24h

# Derived fields for trace linking
query_config:
  derived_fields:
    - name: TraceID
      source: trace_id
      url: 'https://tempo.company.com/trace/${__value.raw}'
      urlDisplayLabel: 'View Trace'

In Grafana, configure Loki Datenquelle:

{
  "name": "Loki",
  "type": "loki",
  "url": "http://loki:3100",
  "jsonData": {
    "derivedFields": [
      {
        "datasourceUid": "tempo-uid",
        "matcherRegex": "trace_id=(\\w+)",
        "name": "TraceID",
        "url": "$${__value.raw}"
      }
    ]
  }
}

Erwartet: Clicking trace ID in Loki logs opens corresponding trace in Tempo.

Bei Fehler: Verifizieren regex matches your log format, check Tempo Datenquelle UID.

Schritt 6: Erstellen Unified Incident View

Erstellen a dashboard that brings all signals together:

{
  "dashboard": {
    "title": "Incident Investigation",
    "templating": {
      "list": [
        {
# ... (see EXAMPLES.md for complete configuration)

Workflow waehrend incident:

1. Alarmieren fires for high error rate
2. On-call engineer opens Grafana dashboard
3. Identifies spike in error rate at specific time
4. Clicks exemplar dot on duration histogram → opens trace
5. Trace shows slow database query
6. Clicks "View Logs" on span → opens logs for that trace
7. Logs reveal specific SQL query causing timeout
8. Root cause identified in <2 minutes

Erwartet: Single pane of glass for debugging, jumping zwischen metrics/logs/traces.

Bei Fehler: If links don't work, check Datenquelle configurations and trace ID propagation.

Validierung

• [ ] Trace IDs present in all application logs
• [ ] Prometheus scraping exemplars
• [ ] Grafana dashboards show exemplar dots on histograms
• [ ] Clicking exemplar opens corresponding trace in Tempo/Jaeger
• [ ] Loki logs have "View Trace" links that work
• [ ] RED dashboard created for key services
• [ ] USE dashboard created for infrastructure
• [ ] Unified incident dashboard tested waehrend GameDay

Haeufige Stolperfallen

• Inconsistent trace ID format: OpenTelemetry uses 32-char hex, Jaeger uses 16-char. Waehlen one.
• Missing context propagation: If trace IDs don't flow across services, distributed tracing breaks. Use OpenTelemetry auto-instrumentation.
• Exemplar overload: Too many exemplars (>100k) can slow Prometheus. Sample high-volume metrics.
• Clock skew: Traces span multiple services. Sicherstellen NTP is configured; clock drift causes trace ordering issues.
• Data retention mismatch: If traces expire vor metrics, correlation breaks. Ausrichten retention policies.

Verwandte Skills

• setup-prometheus-monitoring - metrics foundation for correlation
• configure-log-aggregation - logs foundation for correlation
• instrument-distributed-tracing - traces foundation for correlation
• build-grafana-dashboards - unified visualization layer