← All notes

Production engineering

[Test] Debugging a slow production release

A dummy engineering note for checking the blog index, topic filters, reading metadata and first-party article layout.

Rows of servers in a production data centre
Release speed is an end-to-end systems property.Taylor Vick
On this page

This is a dummy article created to verify the complete Sanity-to-Next.js publishing path. It is safe to edit or delete after testing.

Start with the release boundary

When a release feels slow, first separate build time, deployment orchestration and application startup. Treating them as one number hides the part of the system that actually needs attention.

Measure before changing the pipeline

Capture a small baseline, change one constraint and compare the same stages again. The goal is not a dramatic dashboard. It is a repeatable explanation for where time goes.

What this test covers

  • Featured treatment on the blog index
  • Topic filtering and reading-time metadata
  • Portable Text headings, paragraphs and lists
release-check.sh
time pnpm build
curl -fsS https://example.com/health
SignalBeforeAfter
Build time18 min7 min
Feedback loopManualAutomated
Example release-system measurements used for the test article.

How this was informed

Evidence and method

This test article uses representative release measurements and intentionally synthetic outcomes to validate the publishing system.

FAQ

Questions, answered

What should you measure first in a slow release?

Start with elapsed time at every stage and the time spent waiting between stages. Queue time often reveals more than execution time.

Should every release step run in parallel?

No. Parallelise independent checks, but preserve ordering where one result changes the safety of the next action.

Continue the conversation

Working through a similar engineering decision?

Share the context. I am always interested in the constraints behind the work.

Get in touch