Maniacs_Mansion/MealQuest
11
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Deflake TestFetchCanceledWaitHoldsNoSlot #88

Merged
approver_agent merged 1 commits from task/deflake-cancel-wait_73e54394-0c57-444d-bd5a-a14ca76dd9e8 into dev 2026-10-06 13:18:27 -04:00
Conversation 1 Commits 1 Files Changed 1
+17 -5
dev_agent commented 2026-10-06 13:07:53 -04:00
Member
Copy Link
Copy Source

Maestro task: 73e54394-0c57-444d-bd5a-a14ca76dd9e8 (Feature 10 release correction).

Why

release/2026-10-06c failed Gitea CI on the pull_request run (run 468 / job 6104) at the release head: --- FAIL: TestFetchCanceledWaitHoldsNoSlot with burst Fetch 2 error = Get "...": EOF, want context.DeadlineExceeded. The same commit's push run passed, so it is a real-clock flake.

Done-condition

TestFetchCanceledWaitHoldsNoSlot is deterministic (no real-clock timing / racing burst vs deadline) while preserving the asserted rate-limit/cancellation behaviour; proven with repeated -race runs; full go test -race passes; incorporated into dev and the release branch.

Change

internal/recipes test deflaked (deterministic clock-based assertions); the production rate limiter is unchanged.

Evidence

  • go test -race -count=1000 -run 'TestFetchCanceledWaitHoldsNoSlot$' ./internal/recipes/ -> ok, 0 failures.
  • go test -race -count=50 ./internal/recipes/ -> ok, 0 failures.
  • gofmt empty; go vet ./...; CGO_ENABLED=0 go build ./...; go test -race -count=1 ./... all pass.

Note for follow-up (not this task): TestReleaseWakesWaiter still uses the real clock with a 500ms margin and could flake; and a caller cancelled mid-wait may surface a network error instead of context.Canceled.

Maestro task: 73e54394-0c57-444d-bd5a-a14ca76dd9e8 (Feature 10 release correction). ## Why release/2026-10-06c failed Gitea CI on the pull_request run (run 468 / job 6104) at the release head: `--- FAIL: TestFetchCanceledWaitHoldsNoSlot` with `burst Fetch 2 error = Get "...": EOF, want context.DeadlineExceeded`. The same commit's push run passed, so it is a real-clock flake. ## Done-condition TestFetchCanceledWaitHoldsNoSlot is deterministic (no real-clock timing / racing burst vs deadline) while preserving the asserted rate-limit/cancellation behaviour; proven with repeated -race runs; full go test -race passes; incorporated into dev and the release branch. ## Change internal/recipes test deflaked (deterministic clock-based assertions); the production rate limiter is unchanged. ## Evidence - `go test -race -count=1000 -run 'TestFetchCanceledWaitHoldsNoSlot$' ./internal/recipes/` -> ok, 0 failures. - `go test -race -count=50 ./internal/recipes/` -> ok, 0 failures. - gofmt empty; go vet ./...; CGO_ENABLED=0 go build ./...; go test -race -count=1 ./... all pass. Note for follow-up (not this task): TestReleaseWakesWaiter still uses the real clock with a 500ms margin and could flake; and a caller cancelled mid-wait may surface a network error instead of context.Canceled.
dev_agent added 1 commit 2026-10-06 13:07:54 -04:00
Deflake TestFetchCanceledWaitHoldsNoSlot
CI / ci (push) Successful in 2m15s
Details
CI / ci (pull_request) Successful in 2m17s
Details
1ea2b6de82 
The burst of canceled waits went through Fetch, where net/http races the
failed first write against its read loop on the torn-down connection, so
the error returned was sometimes EOF or "use of closed network
connection" instead of context.DeadlineExceeded (release/2026-10-06c
pull_request CI, run 468 / job 6104).

Drive the burst through waitTurn, the wait turnConn's first Write makes,
which returns the wait's error directly. Advance the fake clock during
the burst and check the host's slot is still the first send, so a slot
leaked by a canceled wait now fails the test; the request after the
burst still goes out exactly HostInterval after the first send with no
wait. Production code is unchanged.

Maestro: 73e54394-0c57-444d-bd5a-a14ca76dd9e8

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
overseer added 1 commit 2026-10-06 13:07:54 -04:00
Deflake TestFetchCanceledWaitHoldsNoSlot
CI / ci (push) Successful in 2m15s
Details
CI / ci (pull_request) Successful in 2m17s
Details
1ea2b6de82 
The burst of canceled waits went through Fetch, where net/http races the
failed first write against its read loop on the torn-down connection, so
the error returned was sometimes EOF or "use of closed network
connection" instead of context.DeadlineExceeded (release/2026-10-06c
pull_request CI, run 468 / job 6104).

Drive the burst through waitTurn, the wait turnConn's first Write makes,
which returns the wait's error directly. Advance the fake clock during
the burst and check the host's slot is still the first send, so a slot
leaked by a canceled wait now fails the test; the request after the
burst still goes out exactly HostInterval after the first send with no
wait. Production code is unchanged.

Maestro: 73e54394-0c57-444d-bd5a-a14ca76dd9e8

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
review_agent left a comment
Member
Copy Link
Copy Source

Independent Codex gpt-6-sol review: APPROVED for this exact commit.

Independent Codex gpt-6-sol review: APPROVED for this exact commit.
approver_agent merged commit 0c89eef577 into dev 2026-10-06 13:18:27 -04:00
Sign in to join this conversation.
Reviewers
No Reviewers
review_agent
Labels
No items
No labels
Milestone
No items
No Milestone
Projects
Clear projects
No projects
Assignees
Clear assignees
approver_agent (The Approver) chatgpt claude-ai Claude-Code codex (Codex) dev_agent grok opencode orchestrator-agent overseer (Joshua) review_agent (The Reviewer)
No Assignees
3 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: Maniacs_Mansion/MealQuest#88