Otel Observability

OpenTelemetry Observability

Overview

Implement comprehensive observability using OpenTelemetry for distributed tracing, Prometheus-compatible metrics, and structured logging across all AgentStack components.

Observability Architecture

┌─────────────────────────────────────────────────────────────────┐
│                   Observability Stack                           │
├─────────────────────────────────────────────────────────────────┤
│                                                                 │
│  ┌─────────────────────────────────────────────────────────┐   │
│  │                   Applications                           │   │
│  │  API Gateway │ Agents │ Workers │ Control Plane          │   │
│  │  (OTEL SDK instrumentation)                              │   │
│  └─────────────────────────────────────────────────────────┘   │
│                           │                                     │
│                           ▼                                     │
│  ┌─────────────────────────────────────────────────────────┐   │
│  │              OTEL Collector                              │   │
│  │  Receives │ Processes │ Exports                          │   │
│  └─────────────────────────────────────────────────────────┘   │
│           │              │              │                       │
│           ▼              ▼              ▼                       │
│  ┌─────────────┐ ┌─────────────┐ ┌─────────────┐               │
│  │   Tempo     │ │ Prometheus  │ │    Loki     │               │
│  │  (Traces)   │ │  (Metrics)  │ │   (Logs)    │               │
│  └─────────────┘ └─────────────┘ └─────────────┘               │
│           └──────────────┴──────────────┘                       │
│                          │                                      │
│                          ▼                                      │
│                   ┌─────────────┐                               │
│                   │   Grafana   │                               │
│                   │ (Dashboards)│                               │
│                   └─────────────┘                               │
│                                                                 │
└─────────────────────────────────────────────────────────────────┘

Go OTEL Setup

Initialize Provider

// internal/pkg/telemetry/provider.go
package telemetry

import (
    "context"
    "time"

    "go.opentelemetry.io/otel"
    "go.opentelemetry.io/otel/exporters/otlp/otlptrace/otlptracegrpc"
    "go.opentelemetry.io/otel/exporters/prometheus"
    "go.opentelemetry.io/otel/propagation"
    "go.opentelemetry.io/otel/sdk/metric"
    "go.opentelemetry.io/otel/sdk/resource"
    "go.opentelemetry.io/otel/sdk/trace"
    semconv "go.opentelemetry.io/otel/semconv/v1.21.0"
)

type Config struct {
    ServiceName    string
    ServiceVersion string
    Environment    string
    OTLPEndpoint   string
}

type Provider struct {
    TracerProvider *trace.TracerProvider
    MeterProvider  *metric.MeterProvider
}

func NewProvider(ctx context.Context, cfg Config) (*Provider, error) {
    // Create resource
    res, err := resource.New(ctx,
        resource.WithAttributes(
            semconv.ServiceNameKey.String(cfg.ServiceName),
            semconv.ServiceVersionKey.String(cfg.ServiceVersion),
            semconv.DeploymentEnvironmentKey.String(cfg.Environment),
        ),
    )
    if err != nil {
        return nil, err
    }

    // Create trace exporter
    traceExporter, err := otlptracegrpc.New(ctx,
        otlptracegrpc.WithEndpoint(cfg.OTLPEndpoint),
        otlptracegrpc.WithInsecure(),
    )
    if err != nil {
        return nil, err
    }

    // Create tracer provider
    tp := trace.NewTracerProvider(
        trace.WithResource(res),
        trace.WithBatcher(traceExporter,
            trace.WithBatchTimeout(5*time.Second),
            trace.WithMaxExportBatchSize(512),
        ),
        trace.WithSampler(trace.ParentBased(trace.TraceIDRatioBased(0.1))),
    )

    // Create Prometheus exporter for metrics
    promExporter, err := prometheus.New()
    if err != nil {
        return nil, err
    }

    // Create meter provider
    mp := metric.NewMeterProvider(
        metric.WithResource(res),
        metric.WithReader(promExporter),
    )

    // Set global providers
    otel.SetTracerProvider(tp)
    otel.SetMeterProvider(mp)
    otel.SetTextMapPropagator(propagation.NewCompositeTextMapPropagator(
        propagation.TraceContext{},
        propagation.Baggage{},
    ))

    return &Provider{
        TracerProvider: tp,
        MeterProvider:  mp,
    }, nil
}

func (p *Provider) Shutdown(ctx context.Context) error {
    if err := p.TracerProvider.Shutdown(ctx); err != nil {
        return err
    }
    return p.MeterProvider.Shutdown(ctx)
}

Initialize in main.go

// cmd/api/main.go
func main() {
    ctx := context.Background()
    
    // Initialize telemetry
    telemetryProvider, err := telemetry.NewProvider(ctx, telemetry.Config{
        ServiceName:    "agentstack-api",
        ServiceVersion: version,
        Environment:    os.Getenv("ENVIRONMENT"),
        OTLPEndpoint:   os.Getenv("OTEL_EXPORTER_OTLP_ENDPOINT"),
    })
    if err != nil {
        log.Fatal(err)
    }
    defer telemetryProvider.Shutdown(ctx)
    
    // ... rest of app initialization
}

Distributed Tracing

Custom Spans

// internal/domain/agent/service.go
package agent

import (
    "context"
    
    "go.opentelemetry.io/otel"
    "go.opentelemetry.io/otel/attribute"
    "go.opentelemetry.io/otel/codes"
    "go.opentelemetry.io/otel/trace"
)

var tracer = otel.Tracer("agentstack/agent")

func (s *Service) Create(ctx context.Context, projectID string, input CreateInput) (*Agent, error) {
    ctx, span := tracer.Start(ctx, "agent.Create",
        trace.WithAttributes(
            attribute.String("project.id", projectID),
            attribute.String("agent.name", input.Name),
            attribute.String("agent.framework", input.Framework),
        ),
    )
    defer span.End()

    // Validate
    if err := s.validate(ctx, input); err != nil {
        span.RecordError(err)
        span.SetStatus(codes.Error, "validation failed")
        return nil, err
    }

    // Create agent
    agent, err := s.repo.Create(ctx, projectID, &Agent{
        Name:      input.Name,
        Framework: input.Framework,
        Config:    input.Config,
    })
    if err != nil {
        span.RecordError(err)
        span.SetStatus(codes.Error, "create failed")
        return nil, err
    }

    span.SetAttributes(attribute.String("agent.id", agent.ID))
    span.SetStatus(codes.Ok, "created")
    
    return agent, nil
}

HTTP Client Tracing

// internal/infrastructure/httpclient/client.go
package httpclient

import (
    "net/http"
    
    "go.opentelemetry.io/contrib/instrumentation/net/http/otelhttp"
)

func NewTracedClient() *http.Client {
    return &http.Client{
        Transport: otelhttp.NewTransport(http.DefaultTransport),
    }
}

Database Tracing

// Using pgx with OTEL
import "github.com/exaring/otelpgx"

poolConfig.ConnConfig.Tracer = otelpgx.NewTracer()

Metrics

Define Custom Metrics

// internal/pkg/telemetry/metrics.go
package telemetry

import (
    "go.opentelemetry.io/otel"
    "go.opentelemetry.io/otel/metric"
)

var meter = otel.Meter("agentstack")

var (
    // Request metrics
    RequestsTotal metric.Int64Counter
    RequestDuration metric.Float64Histogram
    
    // Agent metrics
    AgentsActive metric.Int64UpDownCounter
    AgentDeployDuration metric.Float64Histogram
    
    // Chat metrics
    ChatTokensUsed metric.Int64Counter
    ChatLatency metric.Float64Histogram
    
    // LLM metrics
    LLMCallsTotal metric.Int64Counter
    LLMLatency metric.Float64Histogram
    LLMTokensInput metric.Int64Counter
    LLMTokensOutput metric.Int64Counter
)

func InitMetrics() error {
    var err error

    RequestsTotal, err = meter.Int64Counter("http_requests_total",
        metric.WithDescription("Total HTTP requests"),
        metric.WithUnit("{request}"),
    )
    if err != nil {
        return err
    }

    RequestDuration, err = meter.Float64Histogram("http_request_duration_seconds",
        metric.WithDescription("HTTP request duration"),
        metric.WithUnit("s"),
        metric.WithExplicitBucketBoundaries(0.005, 0.01, 0.025, 0.05, 0.1, 0.25, 0.5, 1, 2.5, 5, 10),
    )
    if err != nil {
        return err
    }

    ChatTokensUsed, err = meter.Int64Counter("chat_tokens_total",
        metric.WithDescription("Total tokens used in chat"),
        metric.WithUnit("{token}"),
    )
    if err != nil {
        return err
    }

    LLMLatency, err = meter.Float64Histogram("llm_call_duration_seconds",
        metric.WithDescription("LLM API call duration"),
        metric.WithUnit("s"),
        metric.WithExplicitBucketBoundaries(0.1, 0.5, 1, 2, 5, 10, 30, 60),
    )
    if err != nil {
        return err
    }

    return nil
}

Record Metrics

// In handler/service
func (h *ChatHandler) Chat(c *fiber.Ctx) error {
    start := time.Now()
    
    // ... process request
    
    duration := time.Since(start).Seconds()
    
    telemetry.RequestsTotal.Add(c.Context(), 1,
        metric.WithAttributes(
            attribute.String("method", "POST"),
            attribute.String("path", "/v1/agents/{id}/chat"),
            attribute.Int("status", c.Response().StatusCode()),
        ),
    )
    
    telemetry.RequestDuration.Record(c.Context(), duration,
        metric.WithAttributes(
            attribute.String("method", "POST"),
            attribute.String("path", "/v1/agents/{id}/chat"),
        ),
    )
    
    telemetry.ChatTokensUsed.Add(c.Context(), int64(usage.TotalTokens),
        metric.WithAttributes(
            attribute.String("agent.id", agentID),
            attribute.String("project.id", projectID),
            attribute.String("model", model),
        ),
    )
    
    return nil
}

Prometheus Endpoint

// internal/api/routes/router.go
import "github.com/prometheus/client_golang/prometheus/promhttp"

func SetupRoutes(app *fiber.App, deps *Dependencies) {
    // Metrics endpoint
    app.Get("/metrics", adaptor.HTTPHandler(promhttp.Handler()))
}

Structured Logging

Logger Setup

// internal/pkg/logger/logger.go
package logger

import (
    "context"
    
    "go.uber.org/zap"
    "go.uber.org/zap/zapcore"
    "go.opentelemetry.io/otel/trace"
)

var log *zap.Logger

func Init(level string, env string) error {
    var cfg zap.Config
    
    if env == "production" {
        cfg = zap.NewProductionConfig()
    } else {
        cfg = zap.NewDevelopmentConfig()
    }
    
    cfg.EncoderConfig.TimeKey = "timestamp"
    cfg.EncoderConfig.EncodeTime = zapcore.ISO8601TimeEncoder
    
    var err error
    log, err = cfg.Build(
        zap.AddCallerSkip(1),
    )
    return err
}

// WithContext adds trace context to log
func WithContext(ctx context.Context) *zap.Logger {
    span := trace.SpanFromContext(ctx)
    if !span.SpanContext().IsValid() {
        return log
    }
    
    return log.With(
        zap.String("trace_id", span.SpanContext().TraceID().String()),
        zap.String("span_id", span.SpanContext().SpanID().String()),
    )
}

func Info(ctx context.Context, msg string, fields ...zap.Field) {
    WithContext(ctx).Info(msg, fields...)
}

func Error(ctx context.Context, msg string, fields ...zap.Field) {
    WithContext(ctx).Error(msg, fields...)
}

func Debug(ctx context.Context, msg string, fields ...zap.Field) {
    WithContext(ctx).Debug(msg, fields...)
}

Usage

func (s *Service) Create(ctx context.Context, input Input) (*Agent, error) {
    logger.Info(ctx, "creating agent",
        zap.String("name", input.Name),
        zap.String("framework", input.Framework),
    )
    
    agent, err := s.repo.Create(ctx, input)
    if err != nil {
        logger.Error(ctx, "failed to create agent",
            zap.Error(err),
            zap.String("name", input.Name),
        )
        return nil, err
    }
    
    logger.Info(ctx, "agent created",
        zap.String("agent_id", agent.ID),
    )
    
    return agent, nil
}

OTEL Collector Configuration

# otel-collector-config.yaml
receivers:
  otlp:
    protocols:
      grpc:
        endpoint: 0.0.0.0:4317
      http:
        endpoint: 0.0.0.0:4318

processors:
  batch:
    timeout: 5s
    send_batch_size: 512
  
  memory_limiter:
    check_interval: 1s
    limit_mib: 1000
    spike_limit_mib: 200
  
  attributes:
    actions:
      - key: environment
        value: production
        action: upsert

exporters:
  otlp/tempo:
    endpoint: tempo:4317
    tls:
      insecure: true
  
  prometheus:
    endpoint: 0.0.0.0:8889
    namespace: agentstack
  
  loki:
    endpoint: http://loki:3100/loki/api/v1/push

service:
  pipelines:
    traces:
      receivers: [otlp]
      processors: [memory_limiter, batch]
      exporters: [otlp/tempo]
    
    metrics:
      receivers: [otlp]
      processors: [memory_limiter, batch]
      exporters: [prometheus]
    
    logs:
      receivers: [otlp]
      processors: [memory_limiter, batch]
      exporters: [loki]

Kubernetes Deployment

apiVersion: apps/v1
kind: Deployment
metadata:
  name: otel-collector
  namespace: observability
spec:
  replicas: 2
  selector:
    matchLabels:
      app: otel-collector
  template:
    spec:
      containers:
        - name: collector
          image: otel/opentelemetry-collector-contrib:0.91.0
          args:
            - --config=/conf/otel-collector-config.yaml
          ports:
            - containerPort: 4317  # OTLP gRPC
            - containerPort: 4318  # OTLP HTTP
            - containerPort: 8889  # Prometheus metrics
          volumeMounts:
            - name: config
              mountPath: /conf
          resources:
            requests:
              cpu: 200m
              memory: 256Mi
            limits:
              cpu: 1000m
              memory: 1Gi
      volumes:
        - name: config
          configMap:
            name: otel-collector-config

Resources

• references/grafana-dashboards.md - Pre-built dashboard templates
• references/alerting-rules.md - Prometheus alerting rules
• assets/dashboards/ - Grafana dashboard JSON files