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

Issue 672846 link

Starred by 2 users

Issue metadata

Status: Available
Owner: ----
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 2
Type: Feature



Sign in to add a comment

Decrease false positives in CQ due to re-use of 1 day old stale tryjobs hiding compile&test failures

Project Member Reported by martiniss@chromium.org, Dec 9 2016

Issue description

See https://codereview.chromium.org/2555923003#msg36 for context.

"
Mini post mortem:
The flaky failures of android_n5x_swarming_rel delayed this patch from committing for ~1 day. In the meantime another change landed that, together with this patch, fails to compile. The patch landed anyway because ~1 day old try bot results were sufficient to pass CQ checks."

It seems strange that CQ do this, but it doesn't sound like good behavior.

CQ team, can you take a look?
 
I don't really know why we picked 1 day, but what would be your suggestion?
My guess is that 1 day makes cross timezone reviews more bearable.
Submit patch, send to try bots, wait for review overnight, land in the morning if lgtm'd and trybots pass.
Labels: -Pri-1 -Type-Bug Pri-2 Type-Feature
Summary: Decrease false positives in CQ due to re-use of 1 day old stale tryjobs hiding compile&test failures (was: CQ landed a patch 1 day after tryjobs finished, causing compile to faile)
So, we (CQ) don't intend to do anything about this in the short to medium term. Since this "bug report" is WAI given current constraints, I mark this as feature request.

It could and should be revisited if:


Option A: we have more capacity OR some new service can "sense" empty resources and re-trigger CQ dry runs on Cls with old (say >12hr) green tryjobs. In short, re-running morning MTV tryjobs the following night. This might actually even save some capacity during MTV peaks.

Option B: we have much faster cycle times, so we can change the 1 day limit to lower number.
Components: -Infra>CQ Infra>Client>Chrome
Cc: tandrii@chromium.org
Components: -Infra>Client>Chrome Infra>Platform>CQdaemon
tandrii, is this configurable on the chromium side? I was under the impression that this was an intrinsic part of the CQ.
Labels: -OS-Chrome
No, it's not, but there is already a feature request to make it so: see issue 753103.

That said, are you actually disagreeing with what I wrote in #3 above?
Nope, no disagreement there.
Components: -Infra>Platform>CQdaemon Infra>Client>Chrome
Then, I think this bug is to be kept outside of CQ daemon component, as there is nothing that CQ daemon needs to do, at least until Option A above becomes feasible. It is clearly a valid bug, so needs to be kept open.
Project Member

Comment 9 by sheriffbot@chromium.org, Aug 21

Labels: Hotlist-Recharge-Cold
Status: Untriaged (was: Available)
This issue has been Available for over a year. If it's no longer important or seems unlikely to be fixed, please consider closing it out. If it is important, please re-triage the issue.

Sorry for the inconvenience if the bug really should have been left as Available.

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot
Status: Available (was: Untriaged)
Neither options in c#3 are yet true, so throwing back into the queue.

Sign in to add a comment