ProcessHarness: HTTP health polling #3

New issue

Closed

opened 2026-04-11 10:55:37 +00:00 by charles · 0 comments

charles commented

2026-04-11 10:55:37 +00:00

Owner

User story

As a test author, I want the harness to block until my app responds 200 on its health endpoint, so that no test runs against a half-started process and flakes from race conditions.

Acceptance criteria

After spawning, ProcessHarness::start() polls health_url via reqwest GET at health_poll_interval (default 100 ms).
A 2xx status marks the process ready and resolves start() with Ok(()).
If startup_timeout elapses without success:
- The harness kills the child.
- Returns TestError::Connection whose message contains:
  - The URL polled
  - The last HTTP status or transport error seen
  - The last 50 lines of stderr captured from the child
If the child exits before becoming ready, the poll loop short-circuits and reports the exit code along with stderr (no waiting out the full timeout).
Health polling honours cancellation: if the harness is dropped mid-poll, no tasks leak.
Polling uses a single reusable reqwest::Client (not one per request).
Connection errors during polling are silently retried until the timeout (they're expected during startup).

Out of scope

Configurable success status range — defaults to 2xx for now.
gRPC or custom protocol health checks.

References

Spec §1.2 (Health wait), §5.1 (Startup sequence)

## User story As a **test author**, I want the harness to block until my app responds 200 on its health endpoint, so that no test runs against a half-started process and flakes from race conditions. ## Acceptance criteria - [ ] After spawning, `ProcessHarness::start()` polls `health_url` via reqwest GET at `health_poll_interval` (default 100 ms). - [ ] A 2xx status marks the process ready and resolves `start()` with `Ok(())`. - [ ] If `startup_timeout` elapses without success: - The harness kills the child. - Returns `TestError::Connection` whose message contains: - The URL polled - The last HTTP status or transport error seen - The last 50 lines of stderr captured from the child - [ ] If the child exits before becoming ready, the poll loop short-circuits and reports the exit code along with stderr (no waiting out the full timeout). - [ ] Health polling honours cancellation: if the harness is dropped mid-poll, no tasks leak. - [ ] Polling uses a single reusable `reqwest::Client` (not one per request). - [ ] Connection errors during polling are silently retried until the timeout (they're expected during startup). ## Out of scope - Configurable success status range — defaults to 2xx for now. - gRPC or custom protocol health checks. ## References - Spec §1.2 (Health wait), §5.1 (Startup sequence)

charles added this to the v0.1.0 milestone

2026-04-11 10:55:37 +00:00

charles added the

area:harness

type:user-story

labels

2026-04-11 10:55:37 +00:00

charles referenced this issue from a commit

2026-04-11 12:04:20 +00:00

feat(harness): builder, lifecycle, stdout/stderr ring buffers (closes #2)

charles referenced this issue

2026-04-11 12:04:45 +00:00