We heard you: Making Compute Errors Easier to Understand

Compute
1

Hi folks,

Dealing with those “Resource Limit Exceeded” or “Edge Code Error” messages on Fastly Compute can be a real pain—it’s often tough to figure out what went wrong or how to fix it. We heard you loud and clear through the community, Reddit, and support: more clarity on when these errors pop up, why, and what to do is essential. So, we’re tackling this head-on by updating our Error Metrics and Observability Dashboards to give you the transparency you asked for!

What’s changing?

We’re moving away from ambiguous “black box” errors and toward a system built for developers. We’ve consolidated and clarified our reporting into three distinct categories:

  1. Instance Resource Limits: No more guessing. If your sandbox is terminated, you’ll see exactly why—whether it was a Timeout, Memory Exhaustion, or vCPU limit.
  2. Runtime & Logic Errors: We’ve grouped code-level issues like stack overflows, unreachable code traps, and circular routing (service chaining) so you can tell at a glance if the issue is in your logic or your configuration.
  3. Platform Errors: If something goes wrong on our end or a request is rejected before it even hits your code, we’re labeling it clearly as a Platform Error. This saves you from hunting for bugs in your own code that don’t exist.

New Docs: Your Troubleshooting Blueprint

Along with the UI update, we’re providing you a brand-new Compute Error Events Reference.

This guide is built for troubleshooting. Every error now includes:

  • The specific trigger - what actually caused the hit.
  • Sandbox behavior - what happens to your Compute sandbox.
  • Actionable steps to fix it - e.g., how to optimize loops for timeouts or manage memory consumption.

Our goal is to eliminate the hours long debug cycle. We want you to be able to identify, diagnose, and deploy a fix in minutes. By providing new revised metrics via our Stats API, you can now build much more intelligent alerting and automated scaling logic.

Timeline

These new metrics and charts are rolling out today. We are keeping the old legacy metrics around until Feb-2027 to give you plenty of time to update your custom monitoring pipelines.

This is part of our ongoing mission to make Fastly the most developer-friendly edge platform. We’re not done yet, so keep the feedback coming—we’re listening.

Happy coding!

Jaskirat

Quick Links:

2 Likes