OpenTelemetry - Quick Reference
Full Reference: See advanced.md for context propagation, detailed metrics API, Collector configuration, Docker Compose stack, sampling strategies, and semantic conventions.
Deep Knowledge: Use mcp__documentation__fetch_docs with technology: opentelemetry for comprehensive documentation.
Concepts
┌─────────────────────────────────────────────────────────────┐
│ OpenTelemetry │
├─────────────────┬─────────────────┬─────────────────────────┤
│ Traces │ Metrics │ Logs │
│ (Distributed) │ (Aggregated) │ (Structured) │
├─────────────────┴─────────────────┴─────────────────────────┤
│ OTLP Protocol │
├─────────────────────────────────────────────────────────────┤
│ Collector (optional) │
├─────────────────────────────────────────────────────────────┤
│ Jaeger │ Zipkin │ Prometheus │ Grafana │ DataDog │ etc. │
└─────────────────────────────────────────────────────────────┘
Node.js Setup
npm install @opentelemetry/sdk-node \
@opentelemetry/auto-instrumentations-node \
@opentelemetry/exporter-trace-otlp-http
Basic Configuration
// tracing.ts - Load FIRST before any other imports
import { NodeSDK } from '@opentelemetry/sdk-node';
import { getNodeAutoInstrumentations } from '@opentelemetry/auto-instrumentations-node';
import { OTLPTraceExporter } from '@opentelemetry/exporter-trace-otlp-http';
import { Resource } from '@opentelemetry/resources';
import { SEMRESATTRS_SERVICE_NAME } from '@opentelemetry/semantic-conventions';
const sdk = new NodeSDK({
resource: new Resource({
[SEMRESATTRS_SERVICE_NAME]: 'my-service',
}),
traceExporter: new OTLPTraceExporter({
url: process.env.OTEL_EXPORTER_OTLP_ENDPOINT || 'http://localhost:4318/v1/traces',
}),
instrumentations: [getNodeAutoInstrumentations()],
});
sdk.start();
process.on('SIGTERM', () => {
sdk.shutdown().finally(() => process.exit(0));
});
Application Entry Point
// index.ts
import './tracing'; // MUST be first import
import express from 'express';
const app = express();
Manual Traces
import { trace, SpanStatusCode } from '@opentelemetry/api';
const tracer = trace.getTracer('my-service', '1.0.0');
async function processOrder(orderId: string) {
return tracer.startActiveSpan('process-order', async (span) => {
try {
span.setAttribute('order.id', orderId);
await processOrderLogic(orderId);
span.setStatus({ code: SpanStatusCode.OK });
} catch (error) {
span.setStatus({ code: SpanStatusCode.ERROR, message: error.message });
span.recordException(error);
throw error;
} finally {
span.end();
}
});
}
Spring Boot Setup
<dependency>
<groupId>io.micrometer</groupId>
<artifactId>micrometer-tracing-bridge-otel</artifactId>
</dependency>
<dependency>
<groupId>io.opentelemetry</groupId>
<artifactId>opentelemetry-exporter-otlp</artifactId>
</dependency>
management:
tracing:
sampling:
probability: 1.0
otlp:
tracing:
endpoint: http://localhost:4318/v1/traces
Environment Variables
OTEL_SERVICE_NAME=my-service
OTEL_EXPORTER_OTLP_ENDPOINT=http://localhost:4318
OTEL_TRACES_SAMPLER=parentbased_traceidratio
OTEL_TRACES_SAMPLER_ARG=0.1 # 10% sampling
When NOT to Use This Skill
- Simple application logging: Use logging frameworks instead
- Monolithic applications: May be overkill
- Vendor-specific APM: DataDog, New Relic have their own SDKs
- Development/debugging only: Standard logging may suffice
- Legacy systems: Migration effort may be high
Anti-Patterns
| Anti-Pattern | Why It's Bad | Solution |
|---|
| Creating spans for every function | Massive overhead | Span only meaningful operations |
| 100% sampling in production | Performance impact, cost | Use 1-10% sampling |
| Not propagating context | Breaks distributed traces | Inject/extract at service boundaries |
| Logging PII in span attributes | Security/compliance violation | Filter sensitive data |
| Forgetting to end spans | Memory leak | Always end spans in finally block |
| Synchronous exporters | Blocks application | Use batch processors |
Quick Troubleshooting
| Issue | Cause | Solution |
|---|
| Traces not appearing | Exporter not configured | Check OTEL_EXPORTER_OTLP_ENDPOINT |
| High memory usage | Too many spans | Enable sampling |
| Missing context propagation | Not injecting headers | Use propagation API |
| Performance degradation | High sampling rate | Reduce to 1-10% |
| SDK initialization error | Import order wrong | Initialize SDK before other imports |
Monitoring Metrics
| Metric | Target |
|---|
| Trace latency (p99) | < 10ms overhead |
| Span drop rate | < 1% |
| Collector memory | < 1GB |
| Export success rate | > 99% |
Checklist
- SDK initialized before other imports
- Resource attributes configured
- Context propagation implemented
- Sampling strategy defined
- Graceful shutdown configured
- Collector deployed
- Dashboards created (Grafana)
- Alerts configured
Reference
Deep Knowledge: Use mcp__documentation__fetch_docs with technology: opentelemetry for comprehensive documentation