Skip to content

fix(test): give the SEA binary e2e test its own startup budget #29

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
Mearman merged 1 commit into from
Sep 13, 2026
Merged

fix(test): give the SEA binary e2e test its own startup budget #29

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
5 changes: 5 additions & 0 deletions src/test/timeouts.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7,3 +7,8 @@ export const REAL_PROCESS_SPAWN_TEST_TIMEOUT_MS = 20_000;
* A test that creates, times, stats and removes on the order of hundreds of individual files sequentially (e.g. exercising sweepSpool's own batch cap) issues far more filesystem syscalls than vitest's default test timeout was ever sized for, on any platform — NTFS under CI disk contention has been observed pushing this over 5000ms on a windows-latest runner even though the identical code path stayed well within it on the same run's ubuntu and windows-11-arm legs. Real I/O volume to accommodate, not a correctness concern.
*/
export const HEAVY_FS_IO_TEST_TIMEOUT_MS = 20_000;

/**
* Starting the packaged SEA binary is a heavier cold start than a plain CcPeer.create(): it loads and JIT-compiles one large single-file bundle (every dependency inlined, no node_modules resolution to spread the cost across smaller files) rather than a normal small script. A real macos-26-intel CI runner — the first run to ever actually reach this check, since every earlier attempt queued forever against the since-retired macos-13 label — took longer than REAL_PROCESS_SPAWN_TEST_TIMEOUT_MS's own 20s here, confirming this is genuine startup cost on that runner class rather than a hang.
*/
export const SEA_BINARY_STARTUP_TIMEOUT_MS = 60_000;
6 changes: 3 additions & 3 deletions test/sea.e2e.test.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5,7 +5,7 @@ import { tmpdir } from "node:os";
import { join } from "node:path";
import { createServer } from "node:net";

import { REAL_PROCESS_SPAWN_TEST_TIMEOUT_MS } from "../src/test/timeouts.js";
import { SEA_BINARY_STARTUP_TIMEOUT_MS } from "../src/test/timeouts.js";

/**
* Exercises the actual packaged single-executable artifact CI ships, not the in-process TS modules every other tier tests: a real, separate process, started the way an end user would, over the network. Requires CC_PEER_SEA_BINARY (the path to a binary already built by scripts/build-sea.ts) — this test does not build one itself, since a SEA build needs a Node distribution with the feature enabled and takes real time, neither of which belongs in a test's own setup.
Expand Down Expand Up @@ -82,7 +82,7 @@ describe.skipIf(BINARY === undefined)("packaged SEA binary", () => {
stderr += chunk.toString("utf8");
});
try {
await waitForPort(port, REAL_PROCESS_SPAWN_TEST_TIMEOUT_MS);
await waitForPort(port, SEA_BINARY_STARTUP_TIMEOUT_MS);

const health = await fetch(
`http://127.0.0.1:${port.toString()}/healthz`,
Expand Down Expand Up @@ -110,6 +110,6 @@ describe.skipIf(BINARY === undefined)("packaged SEA binary", () => {
child.kill();
}
},
REAL_PROCESS_SPAWN_TEST_TIMEOUT_MS,
SEA_BINARY_STARTUP_TIMEOUT_MS,
);
});
Loading