Observability in Cloud Native apps using OpenTelemetry

Meet the demo app with auth and to-do services built in Node.js using TypeScript and Express, backed by Redis orchestrated by Docker Compose. Prepare to implement OpenTelemetry in this repository.

Install OpenTelemetry in your demo app, initialize the SDK with a service name and OTLP exporter, run Jaeger locally via Docker, and enable instrumentation across services.

OpenTelemetry builds distributed traces across services by sharing trace IDs, span IDs, and parent IDs, with context automatically injected into http headers for cross‑process correlation and Jaeger visualization.

Learn to add metrics with OpenTelemetry by configuring the SDK and Prometheus exporter, expose a metrics endpoint, and instrument HTTP calls to capture latency, routes, and status.

Correlate logs with traces by capturing exceptions in the active span and enriching logs with trace id and span id, enabling jump from logs to traces in Jaeger with OpenTelemetry.

Create manual spans with the OpenTelemetry API to instrument custom logic, starting and ending a root span set default items with four Redis child spans.

Add custom attributes to spans in OpenTelemetry using tags, capture user IDs and other app-specific data, and verify traces via active spans and search in Jaeger.

Explore advanced configurations of the OpenTelemetry SDK to deepen observability in cloud native apps, building on the basic implementation covered earlier.

Configure the OpenTelemetry SDK by exploring instrumentations in the registry, selecting and enabling or disabling them, and tuning per-instrumentation settings to tailor spans and metrics.

Turn on OpenTelemetry logs to diagnose issues, using the OTEL_LOG_LEVEL environment variable to emit debug, warning, or error messages that reveal spans and traces in Jaeger.

Learn how to describe the environment and metadata for all spans using predefined and custom resources in OpenTelemetry, including cloud environment, region, pod id, deployment number, and team owner.

Learn how sampling traces in OpenTelemetry controls data volume and costs, with head sampling, downstream propagation, and parent-based versus trace-id ratio samplers.

Explore how opentelemetry propagates context via http headers, using W3C trace context and baggage to pass data like enterprise user plan across downstream services for distributed tracing.

Configure the OpenTelemetry SDK via environment variables to enable quick changes without redeploys, control debug levels and exporters, and tailor context propagation, resources, and samplers.

Explore how spans are created, processed, and exported in OpenTelemetry, and tune batch span processing for performance using defaults like max export batch size, queue size, and schedule delay.

Discover how the opentelemetry collector acts as a scalable backend that ingests, processes, and exports traces, metrics, and logs through customizable receivers, processors, and exporters.

Learn to integrate the OpenTelemetry collector with the SDK, sending traces and metrics to the collector, and have Prometheus scrape the collector while Jaeger receives traces.

Explore how opentelemetry collector processors sit between receive and export, enabling batch processing, memory limiting, and resource enrichment of traces and metrics.

Use the tail sampling processor to keep high-latency or error traces with a 10-second decision window and http status code 500 policy; consider a load balancer exporter for consistent routing.

Debug the collector by using two logging options: service telemetry logs with adjustable verbosity and exporter logging to diagnose issues with traces, spans, the SDK, the receiver, and data flow.

Choose between running your own OpenTelemetry collector or sending data to a vendor, enabling flexible routing of logs, traces, and metrics.