test(agent): deflake progress-extension timing per FLAKY policy (re-review #8)

test_progress_extends_idle_budget_until_success raced wall-clock: the
0.1s-idle/0.04s-tick shape left ~60ms of slack per tick, so one slow
scheduler pass on a loaded CI box lapsed the idle budget mid-loop.
Widened to 0.5s idle / 0.1s ticks (5x per-tick margin, total runtime
still <1s) per the FLAKY policy's minimum-margin guidance.
This commit is contained in:
Teknium 2026-08-02 15:39:43 -07:00
parent b2963f8034
commit 44c362889a
1 changed files with 8 additions and 3 deletions

View File

@ -100,8 +100,13 @@ class TestRunCompressContextWithProgressTimeout:
def worker(fence: CompressionCommitFence):
fence_holder["fence"] = fence
# Keep ticking within each idle window so the waiter extends.
# Round-2 #8 (FLAKY policy): the old 0.1s-idle/0.04s-tick shape
# left only ~60ms of slack per tick — one slow scheduler pass on
# a loaded CI box let the idle budget lapse mid-loop. >=0.5s
# idle with 0.1s ticks keeps a 5x margin per tick while the
# total runtime stays under a second.
for _ in range(6):
time.sleep(0.04)
time.sleep(0.1)
fence.touch_progress()
if not fence.begin_commit():
return (original, "aborted")
@ -114,8 +119,8 @@ class TestRunCompressContextWithProgressTimeout:
worker=worker,
messages=original,
system_prompt_fallback="fallback",
idle_timeout_seconds=0.1,
total_ceiling_seconds=1.0,
idle_timeout_seconds=0.5,
total_ceiling_seconds=5.0,
)
assert result_msgs == compressed