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

Issue 691627 link

Starred by 1 user

Issue metadata

Status: Archived
Owner: ----
Closed: Oct 2017
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Android
Pri: 1
Type: Bug-Regression



Sign in to add a comment

system_health.memory_mobile flakily failing for unknown reason

Project Member Reported by charliea@chromium.org, Feb 13 2017

Issue description

Revision range first seen: ?????
Link to failing step log: https://luci-logdog.appspot.com/v/?s=chrome%2Fbb%2Fchromium.perf%2FAndroid_Nexus5_Perf__1_%2F5239%2F%2B%2Frecipes%2Fsteps%2Fsystem_health.memory_mobile%2F0%2Fstdout

This failure appears to be happening for at least a couple different stories, but may be happening for all stories.

Example stories are:

load:news:sohu - https://luci-logdog.appspot.com/v/?s=chrome%2Fbb%2Fchromium.perf%2FAndroid_Nexus5_Perf__1_%2F5239%2F%2B%2Frecipes%2Fsteps%2Fsystem_health.memory_mobile%2F0%2Fstdout

Failure screenshot: https://console.developers.google.com/m/cloudstorage/b/chrome-telemetry-output/o/profiler-file-id_149-2017-02-12_18-56-5359011.png

load:news:nytimes - https://luci-logdog.appspot.com/v/?s=chrome%2Fbb%2Fchromium.perf%2FAndroid_Nexus5_Perf__1_%2F5240%2F%2B%2Frecipes%2Fsteps%2Fsystem_health.memory_mobile%2F0%2Fstdout

Failure screenshot: https://console.developers.google.com/m/cloudstorage/b/chrome-telemetry-output/o/profiler-file-id_145-2017-02-13_00-10-4218101.png

What's interesting here is that, in both cases, the tests appear to be failing due to Chrome crashes (e.g. WebSocketConnectionClosedException: Connection is already closed.). However, in both cases, the failure screenshots indicate that the browser is still alive.

nednguyen@, can you think of any situations where we have WebSocketConnectionClosedExceptions where the browser screenshot seems to indicate that the browser is still alive?
 

Comment 1 Deleted

(swapping out nednguyen@'s accounts)
Cc: primiano@chromium.org hjd@chromium.org
Could this be a bug in how memory dump command is implemented? What happened here is the socket connection between Telemetry & Chrome is closed before Telemetry receives the response from Chrome after it sends 'Tracing.requestMemoryDump'

memory-infra folks: any opinion?
I'm going to kick off a bisect on this: it seems like load:news:sohu started failing much more frequently recently, and it seems like there's a particular revision range where that may have started.
Project Member

Comment 6 by 42576172...@developer.gserviceaccount.com, Feb 13 2017


=== BISECT JOB RESULTS ===
NO Test failure found

Bisect Details
  Configuration: android_nexus5_perf_bisect
  Benchmark    : system_health.memory_mobile
  Metric       : memory:chrome:all_processes:process_count_avg/load_news/load_news_sohu

Revision             Exit Code      N
chromium@449736      0 +- N/A       5      good
chromium@449841      0 +- N/A       5      bad

Please refer to the following doc on diagnosing memory regressions:
  https://chromium.googlesource.com/chromium/src/+/master/docs/memory-infra/memory_benchmarks.md

To Run This Test
  src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=load.news.sohu system_health.memory_mobile

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8987747008189785616

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=6669481088122880


| O O | Visit http://www.chromium.org/developers/speed-infra/perf-bug-faq
|  X  | for more information addressing perf regression bugs. For feedback,
| / \ | file a bug with component Speed>Bisection.  Thank you!

Comment 7 by zh...@chromium.org, Oct 16 2017

Status: Archived (was: Untriaged)

Sign in to add a comment