Olatuak by Nicolas Narbais

Practical Observability Engineering

Better signal, not more telemetry.

OpenTelemetry, architecture, debugging, and production lessons for engineers running real systems. Practical, opinionated guidance to help you build, debug, and scale observability systems using OpenTelemetry.

Explore the Pillars Watch on YouTube Browse All Posts

Content Pillars

Build. Debug. Scale.

Every article, experiment, and video addresses one of three core engineering challenges.

Build 01

How to implement observability systems properly

Collector topology, pipelines, semantic conventions, instrumentation, and Kubernetes architectures.

View Build Guides →

Debug 02

Why observability systems fail and how to diagnose them

Spans disappearing, broken context propagation, backpressure, dropped telemetry, and queue bottlenecks.

View Debug Guides →

Scale 03

Observability across teams, fleets, and high-volume systems

Telemetry governance, fleet management, cardinality controls, sampling at scale, and cost efficiency.

View Scale Guides →

Build • Featured Guide • Guide • 9 min read

Collector to Collector: choose authentication for the hop you have

Two Collectors can authenticate with TLS, mTLS, basic auth, bearer tokens, OIDC, or a service mesh. Each answers a different question, and the right one depends on the boundary the hop crosses.

Published Sep 4, 2026

opentelemetry collector observability kubernetes

Read the Guide →

Collector to Collector: choose authentication for the hop you have

Latest Field Notes & Guides

Browse all posts →

Scale Guide 6 min read

Do not tune every metric: start with a 60-second scrape interval

A practical 60-second metric collection default, with the alerting, query, source-load, and billing checks needed for justified exceptions.

Published Aug 18, 2026

observability metrics opentelemetry

Debug Guide 21 min read

Query-level PostgreSQL evidence with OpenTelemetry: a privilege decision, not a config flag

Query samples, blocking evidence and pg_stat_statements are not config flags. Each one buys diagnostic power for a privilege, for query text leaving the database, or for work the receiver makes PostgreSQL do. What each costs, measured.

Published Aug 12, 2026

postgresql opentelemetry observability

Build Guide 16 min read

A database receiver is not database observability: PostgreSQL with OpenTelemetry

The PostgreSQL receiver reports what the server is doing; your traces report what the request felt. Neither joins the other. A staged guide to server metrics, client spans and pool telemetry, with what each one cannot tell you.

Published Jul 31, 2026

postgresql opentelemetry observability

Build Guide 12 min read

Beyond the defaults: extending Kubernetes enrichment with your own labels and annotations

Add Kubernetes pod and namespace labels or annotations to OpenTelemetry logs with the k8s_attributes processor.

Published Jul 1, 2026

opentelemetry kubernetes observability

Debug Guide 8 min read

A Local Lab for Debugging OpenTelemetry Collector Pipelines

Use Docker Compose to prove that traces, metrics, and logs pass through each OpenTelemetry Collector pipeline before you debug the backend.

Published May 30, 2026

opentelemetry collector prometheus

Build Guide 8 min read

OpenTelemetry on Kubernetes: deploy node, cluster, and identity in that order

How to split node-local and cluster-wide OpenTelemetry Collectors on Kubernetes with Helm, explicit OTLP export to a backend.

Published May 12, 2026

kubernetes opentelemetry observability

Olatuak Video Channel

Watch experiments, stress-tests, and debugging live.

I build, break, debug, and scale observability systems on YouTube. See real Collector pipelines under pressure, reproduce subtle failure modes, and discover architectural lessons.

Watch on YouTube →

Core Topics

Targeted Entry Points

Explore focused guides tailored to specific operational disciplines.