Telemetry seems to undercount story times in benchmarks |
|||||
Issue descriptionIt appears that our logic counting story runtimes is incorrect somehow. I've been gather story timing data, as part of my efforts to better shard our benchmarks. I noticed that we seem to be miscounting our story timing. Two examples: system_health.memory_desktop (https://chromium-swarm.appspot.com/task?id=37498c69d30d8610&refresh=10). It takes about 10 minutes longer to execute than the summation of story times says it should. We'd expect maybe a minute due to python startup or something like that, but 10 minutes seems a bit excessive. service_worker.service_worker_micro_benchmark (https://chromium-swarm.appspot.com/task?id=374abd00973e6610&refresh=10&show_raw=1). The benchmark takes 15 seconds, but the test itself takes about 8 seconds to run. It's unclear what's causing this, but I'm filing this bug just to write it down.
,
Jul 12 2017
I guess this could be due to code running in between stories? Not sure exactly how much happens there, but I think some platform setup does.
,
Jul 12
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
,
Jul 12
I think Ashley addressed this a while ago
,
Jan 16
,
Jan 16
|
|||||
►
Sign in to add a comment |
|||||
Comment 1 by nedngu...@google.com
, Jul 12 2017Components: Speed>Telemetry