Code Red Newsletter #14

Hi there,

Welcome back to your bi-weekly dose of observability geekery. It’s been a busy couple of weeks in OTel-land - config pitfalls uncovered, span naming debates reignited, and yet another round of “is this magic or just auto-instrumentation?” Speaking of which, Dash0 has just contributed a Zig-based injector to the OpenTelemetry project - more on that below. If you’ve ever stared at your telemetry and thought this could really use a naming convention… this issue is for you.

In Focus: Automatic Instrumentation & Best Practices

Let’s be real: observability isn’t just about collecting all the signals. It’s about collecting them well. From naming spans to configuring your SDKs, the difference between clean, actionable telemetry and unreadable spaghetti often comes down to the small stuff. This issue, we’re diving into the practices, pitfalls, and behind-the-scenes magic that make OpenTelemetry actually work in production.

Demystifying Automatic Instrumentation: How the Magic Actually Works

Automatic instrumentation feels like sorcery - add a sidecar, load an agent, and boom: traces everywhere. But under the hood? It’s all about dynamic linking, bytecode weaving, and clever hooks. Causely breaks down the mechanics so you can understand the tradeoffs and troubleshoot when the “magic” misfires.

Read more here.

Dash0 donates Zig Injector to OpenTelemetry

Dash0 has opened a draft PR contributing our Zig-based injector to the OpenTelemetry project as a replacement for the older C-based version. The Zig approach works cleanly across both glibc and musl, making it far more reliable in containerized environments.

It also handles environment variables more gracefully and respects workload settings while still being Kubernetes-aware.

It’s early days, but this is a solid step toward making auto-instrumentation more portable and developer-friendly.

Check out the PR here.

How to Name Your Spans

It sounds trivial - until you’re sifting through thousands of inconsistently named spans at 2 a.m. The OpenTelemetry blog lays down best practices for span naming, why consistency matters, and how to strike a balance between readability and precision.

A must-read if you want traces that tell stories, not riddles.

Read more here.

OpenTelemetry Configuration Gotchas

Configuring OTel should be straightforward… except when it’s not. Nicolas Frankel documents the sneaky pitfalls and “gotchas” that can trip you up: from exporter misconfigurations to environment variable chaos. Consider this your survival guide to not rage-quitting your first Collector setup.

Read more here.

Code RED Podcast: Behind the Dashboards with Chen Harel

Former Coralogix VP of Products and OverOps co-founder Chen Harel joins Dash0’s Mirko Novakovic for a candid look at the observability industry - past, present, and future. They unpack the early days of production debugging, the realities of scaling in a crowded market, and what it’s like competing with the giants. From startup survival tactics to enterprise politics, Chen drops lessons earned the hard way.

Listen here.

Choice Cuts

Because sometimes you need more than just the basics - here are the shiny extras we couldn’t leave out.

OTel in Practice: Alibaba’s OpenTelemetry Journey

One of the world’s largest microservice platforms shares how they rolled out OpenTelemetry across thousands of services. Beyond scale, they’ve also been innovating: a standout highlight is their Go compile-time instrumentation approach, which bakes telemetry directly into services without runtime overhead. It’s proof that OTel isn’t just about adoption at scale - it’s also about pushing the boundaries of how instrumentation can be done. A case study well worth studying.

Read it here.

Can LLMs Replace On-Call SREs Today?

ClickHouse ran the experiment we’ve all been joking about: can an LLM take pager duty? Spoiler: it’s… complicated. The results highlight both the potential and the current limits of AI in observability. A fun read if you’ve ever wished a bot could take your 3 a.m. alerts.

Read the post.

Bonus: More Signal, Less Noise

Prioritizing what to build in a project the size of OpenTelemetry is tough. GitHub issue reactions are becoming a lightweight way for maintainers to gauge interest - cutting through noise and surfacing what the community actually cares about. It’s a small shift, but one that makes open governance feel more responsive and transparent.

Read it here.

See You in Two Weeks

That’s it for this round of Code RED. From span naming sanity to auto-instrumentation magic, one thing is clear: observability isn’t just about more signals - it’s about better ones. Keep naming carefully, configuring wisely, and remember: the real magic happens when your telemetry is clean enough to trust.

We’ll be back in two weeks with more observability insights, stories from the trenches, and maybe a hot take or two on AI and platforms.

Oh, and stay tuned - at Dash0, we’ve got an exciting launch coming next week that you won’t want to miss.

Until then: keep tracing, keep instrumenting, and maybe rename that span before it haunts your next incident review.

Kasper, out.