Hendrik LiebauandGitHub 840f3de980 [test] Unflake metadata static routes cache test (#94796) ...

[flakiness
metrics](https://app.datadoghq.com/ci/test/runs?query=test_level%3Atest%20%40git.repository.id%3A%22github.com%2Fvercel%2Fnext.js%22%20%40test.name%3A%22app%20dir%20-%20metadata%20static%20routes%20cache%20should%20generate%20different%20content%20after%20replace%20the%20static%20metadata%20file%22%20%40test.type%3A%22nextjs%22%20%40test.status%3A%28fail%20OR%20pass%29%20%40git.branch%3Acanary&agg_m=count&agg_m_source=base&agg_t=count&fromUser=false&index=citest&start=1780774829532&end=1781379629532&paused=false)

<img width="423" height="173" alt="Screenshot 2026-06-13 at 21 41 05"
src="https://github.com/user-attachments/assets/d9888a77-db05-4ab5-89e3-97f01b9d7353"
/>

This test replaces the static favicon and Open Graph image and rebuilds
to verify the served content changes, so it needs one production build
with the original files and another with the replacements. Running both
`next.start()` cycles in a single test case put two full builds under
the 60s per-test budget, which is easy to exceed on slower CI runners,
where the second build most often stalls while collecting build traces.

We move the first build, start, and baseline measurement into
`beforeAll`, where it can use the longer setup-hook budget, and leave
the test body responsible only for the second build. Keeping the build
out of the test body also makes it idempotent, so a retry recomputes the
new hashes against the baseline that `beforeAll` captures once.

2026-06-16 10:26:45 +02:00

..

app

…

metadata-static-route-cache.test.ts

…