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

Issue 622317 link

Starred by 0 users

Issue metadata

Status: WontFix
Owner: ----
Closed: Mar 2017
Cc:
EstimatedDays: ----
NextAction: ----
OS: Mac
Pri: 2
Type: Bug-Regression



Sign in to add a comment

memory.long_running_idle_gmail_tbmv2.reference occasionally flaky on Mac Retina Perf (2)

Project Member Reported by petrcermak@chromium.org, Jun 22 2016

Issue description

Labels: Performance-Sheriff-BotHealth
Cc: nedngu...@google.com
Happened again: https://build.chromium.org/p/chromium.perf/builders/Mac%20Retina%20Perf%20%282%29/builds/3876
Screenshot: https://console.developers.google.com/m/cloudstorage/b/chrome-telemetry-output/o/profiler-file-id_0-2016-08-04_03-17-2434659.png

Renderer crash again:

(INFO) 2016-08-04 03:17:23,497 browser.DumpStateUponFailure:346  *************** BROWSER STANDARD OUTPUT ***************
(INFO) 2016-08-04 03:17:23,498 browser.DumpStateUponFailure:348  2016-08-04 03:15:38.148 Google Chrome[6666:24413] NSWindow warning: adding an unknown subview: <FullSizeContentView: 0x7fc202f3f5e0>. Break on NSLog to debug.
2016-08-04 03:15:38.148 Google Chrome[6666:24413] Call stack:
(
    "+callStackSymbols disabled for performance reasons"
)
Aug  4 03:15:39  Google Chrome Helper[6670] <Error>: CGAffineTransformInvert: singular matrix.
Aug  4 03:15:39  Google Chrome Helper[6670] <Error>: CGAffineTransformInvert: singular matrix.
Aug  4 03:15:40  Google Chrome Helper[6670] <Error>: CGAffineTransformInvert: singular matrix.

(INFO) 2016-08-04 03:17:23,498 browser.DumpStateUponFailure:351  *********** END OF BROWSER STANDARD OUTPUT ************
(INFO) 2016-08-04 03:17:23,498 browser.DumpStateUponFailure:353  ********************* BROWSER LOG *********************
(INFO) 2016-08-04 03:17:23,498 browser.DumpStateUponFailure:355  No log file
(INFO) 2016-08-04 03:17:23,498 browser.DumpStateUponFailure:358  ***************** END OF BROWSER LOG ******************

I don't think we can do much about this unless we enable logging ( issue 623058 ) in this benchmark. Any thoughts?
Even if we found the culrpit, I think our action item would just be "update the ref build", which is not different from if we don't find the culprit.

I think let just disable the benchmark on ref build Mac with the ShoulDisable() decorator
The flake is very rare (4 times in the last 200 builds: https://build.chromium.org/p/chromium.perf/builders/Mac%20Retina%20Perf%20%282%29?numbuilds=200), so there's no need to disable (sheriffing guide says benchmarks should be disabled if they're failing "more than one out of five runs"). Let's hope updating the ref build will fix this.
Cc: aiolos@chromium.org
sgtm
Cc: -petrcermak@chromium.org
Cc: -petrcermak@chromium.org

Comment 8 by benhenry@google.com, Dec 21 2016

Labels: -Performance-BotHealth
Status: WontFix (was: Untriaged)
This is no longer occurring. 

Sign in to add a comment