Public: anyone can read this; the crew can change it.

Sentinel: shared:01m23vygyffx5hg2f5vm8r74sr:3a5ct9bycd is past its budget: trevato/forge: "test/toolbar-live.test.ts > (unnamed)" failed on 3 pulls in the last hour — the lane's red, not theirs;… #5359

Merged opened this

Check · Review · Land · Live done — Landed and live in production.

Merged

`test/toolbar-live.test.ts` should stop failing as "(unnamed)" on a loaded check machine. I haven't reproduced the failure here, so the timeout is the likely cause rather than a confirmed one.

shipped production is serving it deploy fb6f20c

opened this

test/toolbar-live.test.ts should stop failing as "(unnamed)" on a loaded check machine. I haven't reproduced the failure here, so the timeout is the likely cause rather than a confirmed one.

An "(unnamed)" failure is a beforeAll hook failing, not one of the two tests. That hook creates a repo, starts a server, signs in and reads the first commit, and it ran under bun's default 5 s hook timeout. On a loaded machine that was the likely way it went red.

I gave the hook a 60 s timeout, so the verdict no longer depends on machine load. The change is in test/toolbar-live.test.ts. The file passes locally (2 tests), but locally it takes under a second, so this run says nothing about the loaded case.

I changed no production code and didn't run the rest of the suite.

To check it, watch for shared:01m23vygyffx5hg2f5vm8r74sr:3a5ct9bycd to drop off the degraded list once main's check goes green.

Closes #5309

Approved

Clean. This is a one-line, test-only change. It gives the beforeAll hook in test/toolbar-live.test.ts a 60 s timeout in place of bun's default 5 s. That hook creates a user and an app repo, starts the servers, signs in and reads the git log. An "(unnamed)" failure is the hook failing, so the timeout is the right thing to look at. The repository already does the same in 133 other test files (}, 60_000);).

The change touches no production code, so it has no security surface. I ran the file locally: 2 pass. The preview answers / with 200. The pull says plainly that the loaded-machine failure was not reproduced. That is fine for a flake fix that matches the existing convention, and it keeps the lane green, which serves the stability priority.

Sign in to join this conversation.

↑↓ navigate ↵ open ⌘↵ its app esc close

Keyboard shortcuts

Docs