Issue metadata
Sign in to add a comment
|
Browser startup in Telemetry is flaky when run in parallel on Linux platform |
||||||||||||||||||||||||
Issue descriptionBrowser startup in Telemetry is flaky when run in parallel on Linux platform. This has been making telemetry_perf_unittests super flaky on Linux (https://test-results.appspot.com/dashboards/flakiness_dashboard.html#testType=telemetry_perf_unittests&builder=tryserver.chromium.linux%3Alinux_chromium_rel_ng) & force us to disable parallelization in telemetry_unittest on Linux (issue 676742) The first suspect is we need to disable gpu when running browser in parallel, see: https://github.com/catapult-project/catapult/issues/3074#issuecomment-266176383
,
Mar 2 2017
Looking at one of the failing shards: https://chromium-swarm.appspot.com/task?id=34a8b3b0ef322b10&refresh=10&show_raw=1 the machine is a VM running Mesa and the Gallium/LLVMpipe renderer: OS: linux trusty GPU device 0: VENDOR = 0x0, DEVICE = 0x0 GPU Attributes: amd_switchable : False basic_info_state : 2 context_info_state : 1 direct_rendering : True driver_date : driver_vendor : Mesa driver_version : 10.1.3 ... gl_renderer : Gallium 0.4 on llvmpipe (LLVM 3.4, 256 bits) gl_reset_notification_strategy: 33377 gl_vendor : VMware, Inc. gl_version : 2.1 Mesa 10.1.3 I'm surprised if Chrome goes down its GPU path in this configuration, though I suppose adding --disable-gpu to the command line wouldn't hurt things further.
,
Jul 26 2017
,
Oct 5 2017
|
|||||||||||||||||||||||||
►
Sign in to add a comment |
|||||||||||||||||||||||||
Comment 1 by nedngu...@google.com
, Mar 2 2017