Skip to content

agent_throughput_multiple_responses uses same flaky wall-clock defect class as #6687 #6690

Description

@bug-ops

Description

Found during review of issue #6687 (PR #6688). tests/performance_agent_integration.rs::agent_throughput_multiple_responses (around line 443) uses a fixed wall-clock assertion around a fully-mocked Agent::run() call — the same defect class fixed for agent_integration_with_safe_bash_blocks and agent_integration_no_bash_blocks in PR #6688 (issue #6687), and previously for 3 zeph-core unit tests in PR #6685 (issue #6679).

This instance was left out of PR #6688's scope because, unlike the other two tests in that file, it currently has ample headroom against observed CI load and had not been reported as flaky. It is being tracked separately rather than silently left as a latent flake risk.

Reproduction Steps

Not yet reproduced as flaky — this is a preventive tracking issue, not a confirmed CI failure.

Expected Behavior

Same treatment as the other tests in this defect class: drop the wall-clock assertion (or replace with a structural proof) if the timing check adds no coverage beyond "didn't hang", which a bounded top-level test timeout / CI job timeout already covers.

Actual Behavior

Test still asserts a fixed wall-clock bound around a mocked, no-I/O Agent::run() call, same latent flake risk as #6679/#6687.

Environment

Logs / Evidence

Flagged by adversarial critique during the #6687 flaky-test fix (PR #6688) as out of scope for that fix but worth tracking separately.

Metadata

Metadata

Assignees

Labels

P3Research — medium-high complexitybugSomething isn't working

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions