New issue
Advanced search Search tips

Issue 814970 link

Starred by 1 user

Issue metadata

Status: Unconfirmed
Owner: ----
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Chrome
Pri: 1
Type: Bug



Sign in to add a comment

Build steps that are aborted because the hit the timeout should not be labelled infra_failure

Project Member Reported by norvez@chromium.org, Feb 22 2018

Issue description

See for example: https://luci-milo.appspot.com/buildbot/chromeos/coral-release/775

Build step 58 (HWTest [bvt-arc] [robo] HWTest [bvt-arc] [robo]) reports:

"
Reason: Suite job failed.
 02-21-2018 [23:53:22] Output below this line is for buildbot consumption:
Will return from run_suite with status: INFRA_FAILURE
"

In fact I think it stopped because it hit a 3h timeout. It's very misleading because failures like that tend to be quickly dismissed and ignored as "infra flake" when it may indicate a legitimate issue (tests suddenly slowing down massively for some reason). Can we report "Hit xxxh timeout" instead of the generic "** HWTest did not complete due to infrastructure issues (code 3) **"
 
Sadly, the vast majority of the times that happens is really is infra flake, such as provision issues or a shortage of DUTs.

Though better error messages would certainly help everyone.

Sign in to add a comment