New issue
Advanced search Search tips
Note: Color blocks (like █ or █) mean that a user may not be available. Tooltip shows the reason.

Issue 849571 link

Starred by 1 user
Status: WontFix
Owner:
liaoyuke@chromium.org
Closed: Jan 8
Cc:
█ mgiuca@chromium.org
█ robert...@chromium.org
chanli@chromium.org
lijeffrey@chromium.org
liaoyuke@chromium.org
st...@chromium.org
Components:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 2
Type: Bug

Blocking:
issue 837855



Sign in to add a comment

TryFlakes is not good at avoiding duplicates

Project Member Reported by mgiuca@chromium.org, Jun 5 2018 
TryFlakes just reported two bugs in the same minute with two different parameterized versions of the exact same test failing. See screenshot.

Bugs:
https://bugs.chromium.org/p/chromium/issues/detail?id=849556
https://bugs.chromium.org/p/chromium/issues/detail?id=849557

This is a particularly egregious example (perhaps consider parameterized versions of a test to be equivalent?). But in general, I have noticed that TryFlakes and FindIt report obviously-the-same-problem as multiple bugs quite frequently.
duplicate.png
26.4 KB View Download

Comment 1 by mgiuca@chromium.org, Jun 5 2018 

Also (kind of a separate issue), it's really late.

These issues were reported at 2018-06-05 05:46:33 UTC.

Inspecting the bot, these failures happened in a contiguous block, then stopped. The last failure was at 2018-06-04 18:03:16 UTC. Why did it take over 11 hours from when it *stopped being a problem* for the issue to be reported? (And if it's going to be that late, why didn't it figure out that it's no longer a problem, and cancel the report?)
last-failure.png
268 KB View Download

Comment 2 by liaoyuke@chromium.org, Jun 5 2018

Blocking: 837855
Owner: liaoyuke@chromium.org
Status: Assigned (was: Unconfirmed)
Thank you very much for the feedback mgiuca@!

TryFlakes was not designed to deal with duplicates from the first place (along with many other issues), so it's non-trivial to refactor and fix this issue in TryFlakes. Instead, I'm working on a new flake detection that is designed to correctly handle the cases you mentioned, and I will use the examples you provided to verify the new detection.

I have an umbrella bug if you're interested: crbug.com/837855.

Comment 3 by mgiuca@chromium.org, Jun 6 2018 

Thanks. Feel free to close this as a duplicate of that (or not, if it helps you to leave it open).

Comment 4 by liaoyuke@chromium.org, Jan 8

Status: WontFix (was: Assigned)
try-flakes is deprecated, and a solution was implemented in the new flake detector to avoid such duplicates.

https://findit-for-me.appspot.com/ranked-flakes
►

Sign in to add a comment