New issue
Advanced search Search tips

Issue 745033 link

Starred by 1 user
Status: WontFix
Owner: ----
Closed: Dec 2017
Cc:
jeffcarp@chromium.org
robertma@chromium.org
Components:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 2
Type: Bug



Sign in to add a comment

win_chromium_rel_ng timed out after 4 hours (in no-op "compile (without patch)" step)

Project Member Reported by e...@chromium.org, Jul 17 2017 
In run 490840 the win_chromium_rel_ng bot failed the 
"compile (without patch)" step with the error "confirm no-op compile" and no further details.

This is after having been processing the job for over four hours already and triggered a second run thereby requiring at least a full work day for a single run on a single bot. This is really hurting productivity.

1: https://build.chromium.org/p/tryserver.chromium.win/builders/win_chromium_rel_ng/builds/490840

Comment 1 by qyears...@chromium.org, Jul 18 2017

Cc: jeffcarp@chromium.org
Labels: -Pri-3 Pri-1
Summary: win_chromium_rel_ng timed out after 4 hours (in no-op "compile (without patch)" step) (was: win_chromium_rel_ng fails no-op "compile (without patch)" step)
That is pretty frustrating.

Note at the end of the logs for the entire job, it has:

   command timed out: 14400 seconds elapsed, attempting to kill

Not sure whether something was hung, or if something took an especially long time and was actually still running... it looks like some other jobs may be taking a long time (https://build.chromium.org/p/tryserver.chromium.win/builders/win_chromium_rel_ng/) so this may be a general problem now that should be looked at tomorrow.

Comment 2 by qyears...@chromium.org, Jul 18 2017 

More recent jobs seem to be OK -- eae, let me know if this is still a problem; maybe it's something that happens for certain CLs.

Comment 3 by e...@chromium.org, Jul 18 2017 

Will do. The second try succeeded after 2.5h. The problem seems to be that if the windows bot fails any layout tests there isn't enough time for it to retry them before hitting the 4h limit. Perhaps we should fail "quickly" if any layout tests fails (with retries) instead of attempting to run without the patch?

Comment 4 by foolip@chromium.org, Dec 15 2017

Cc: robertm@chromium.org
robertma@, you've mention the layers of retry logic before, do you think the suggested quick fail would make sense? Might it make wpt-import faster?

Comment 5 by qyears...@chromium.org, Dec 22 2017

Cc: -robertm@chromium.org robertma@chromium.org
Labels: -Pri-1 Pri-2
Status: WontFix (was: Untriaged)
Doesn't seem to be an issue recently.

Attempting to run without the patch and only considering tests that only failed with patch is useful whenever there are failing tests at HEAD, which I think is pretty common for layout tests...
►

Sign in to add a comment