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

Issue 815004 link

Starred by 1 user
Status: Untriaged
Owner: ----
Cc:
perezju@chromium.org
█ kbr@chromium.org
Components:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 3
Type: Bug

Blocking:
issue 814791



Sign in to add a comment

Telemetry fails to shutdown debug Chrome on Window

Project Member Reported by nednguyen@chromium.org, Feb 23 2018 
Log https://chromium-swarm.appspot.com/task?id=3bd8b63be2791010&refresh=10&show_raw=1

Within the swarming log, we can find:

Failed to cooperatively shutdown. Proceeding to terminate: Timed out while waiting 5s for py_utils.WaitFor(lambda: not self.IsBrowserRunning(), timeout=5)



This seems to lead to leaky browser processes, e.g: 
- pid 7668; Handles: 286; Exe: e:\b\s\w\ir\out\Debug\chrome.exe; Cmd: "e:\b\s\w\ir\out\Debug\chrome.exe" --type=renderer --autoplay-policy=no-user-gesture-required --enable-logging --trace-config-file="e:\b\s\w\itiieftv\tmpodckl1\chrome-trace-config.json" --v=0 --enable-gpu-benchmarking --field-trial-handle=1564,8170620737471744229,6519055701058839744,131072 --service-pipe-token=F417D420F5D3EF358C2B5CCB9C5AB072 --lang=en-US --noerrdialogs --user-data-dir="e:\b\s\w\itiieftv\tmpde9j08" --enable-offline-auto-reload --enable-offline-auto-reload-visible-only --blink-settings=disallowFetchForDocWrittenScriptsInMainFrame=false,disallowFetchForDocWrittenScriptsInMainFrameOnSlowConnections=true,cssExternalScannerNoPreload=false,cssExternalScannerPreload=true --enable-net-benchmarking --device-scale-factor=1 --num-raster-threads=4 --enable-main-frame-before-activation --enable-checker-imaging --enable-compositor-image-animations --service-request-channel-token=F417D420F5D3EF358C2B5CCB9C5AB072 --renderer-client-id=3 --mojo-platform-channel-handle=2444 /prefetch:1
-

Comment 1 by kbr@chromium.org, Feb 23 2018 

How often is this happening?

If the cooperative shutdown of the browser fails, we will leak processes. It's expected that this generally won't happen. This was needed on Windows because killing the browser process hard, via Python, on that platform does not cause the entire process tree to get torn down.

Comment 2 by kbr@chromium.org, Feb 23 2018 

Note: if this is happening only on the Debug bots then it's probably because the build is much slower than a Release build, and that timeout probably needs to be increased to compensate.

Or, we could add a DevTools command to cooperatively shut down the browser.

Comment 3 by nednguyen@chromium.org, Feb 23 2018 

Ken: this is happening very often in Debug bots: https://uberchromegw.corp.google.com/i/chromium.win/builders/Win7%20Tests%20%28dbg%29%281%29

I just sent out CL to increase the timeout (TBR'ed you :-P)

Comment 4 by kbr@chromium.org, Feb 23 2018 

Aha, I see. Sounds fine.
Project Member

Comment 5 by bugdroid1@chromium.org, Feb 23 2018 

The following revision refers to this bug:
  https://chromium.googlesource.com/catapult/+/13a8b379801a2265f1db14309816951fc2710c18

commit 13a8b379801a2265f1db14309816951fc2710c18
Author: Nghia Nguyen <nednguyen@google.com>
Date: Fri Feb 23 01:51:38 2018

Increase the timeout to cooperatively shutdown browser from 5 to 15s

Bug: chromium:815004
Change-Id: I6140c0b2a6b8792adc96b457eae032d88d35051f
TBR=perezju@chromium.org, kbr@chromium.org
Reviewed-on: https://chromium-review.googlesource.com/933541
Commit-Queue: Ned Nguyen <nednguyen@google.com>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Reviewed-by: Ned Nguyen <nednguyen@google.com>

[modify] https://crrev.com/13a8b379801a2265f1db14309816951fc2710c18/telemetry/telemetry/internal/backends/chrome/desktop_browser_backend.py

Comment 6 by benhenry@google.com, Jan 16 (6 days ago)

Components: Test>Telemetry

Comment 7 by benhenry@google.com, Jan 16 (6 days ago)

Components: -Speed>Telemetry
►

Sign in to add a comment