Skip to content

🐛 Stabilize flaky auth decorator overhead test by using high-resolution timing and safe ratio calculation #9041

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
mergify merged 1 commit into from
Apr 20, 2026
Merged

🐛 Stabilize flaky auth decorator overhead test by using high-resolution timing and safe ratio calculation #9041

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
19 changes: 10 additions & 9 deletions services/web/server/tests/unit/isolated/test_security_web.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3,8 +3,8 @@
# pylint: disable=unused-variable
# pylint: disable=too-many-arguments

import asyncio
import statistics
import time
from collections import OrderedDict
from collections.abc import Awaitable, Callable
from typing import Any
Expand Down Expand Up @@ -414,13 +414,13 @@ async def test_time_overhead_on_handlers_of_auth_decorators(
get_active_user_or_none_dbmock: MagicMock,
is_user_in_product_name_dbmock: MagicMock,
):
async def _req(url_):
start = asyncio.get_event_loop().time()
async def _req(url_: str) -> int:
start_ns = time.perf_counter_ns()
resp = await client.post(url_)
stop = asyncio.get_event_loop().time()
elapsed_ns = time.perf_counter_ns() - start_ns

assert resp.ok, f"error: {await resp.text()}"
return stop - start
return elapsed_ns

num_of_rounds = 100
public_elapsed_times = [await _req("/v0/public") for _ in range(num_of_rounds)]
Expand All @@ -429,11 +429,12 @@ async def _req(url_):
# SEE https://docs.python.org/3/library/statistics.html#function-details
# Note The mean is strongly affected by outliers and is not necessarily a typical example of the data points.
# For a more robust, although less efficient, measure of central tendency, see median().
ref_elapsed = statistics.median(public_elapsed_times)
elapsed = statistics.median(admin_elapsed_times)
ref_elapsed_ns = statistics.median(public_elapsed_times)
elapsed_ns = statistics.median(admin_elapsed_times)
elapsed_ratio = elapsed_ns / max(ref_elapsed_ns, 1)

# NOTE: 150% more wrt reference (basically ~ 2.5x more !!!!!!!!!!!)
# and this is mocking the access to the database!
msg = f"got {elapsed/ref_elapsed=} from medians {elapsed=} and {ref_elapsed=} after {num_of_rounds=}"
msg = f"got {elapsed_ratio=} from medians {elapsed_ns=}ns and {ref_elapsed_ns=}ns after {num_of_rounds=}"
Comment thread
GitHK marked this conversation as resolved.
print(msg.capitalize())
assert elapsed < ref_elapsed * (1 + 1.6), msg
assert elapsed_ns < ref_elapsed_ns * (1 + 1.6), msg
Loading