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

Issue 918514 link

Starred by 2 users

Issue metadata

Status: WontFix
Owner: ----
Closed: Jan 11
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Linux
Pri: 3
Type: Bug



Sign in to add a comment

reproducing layout test failure kills entire X session

Project Member Reported by neis@chromium.org, Jan 2

Issue description

When trying to reproduce https://chromium-swarm.appspot.com/task?id=41efa4bb47ef1910&refresh=10&show_raw=1 my entire X sessions gets killed.

I've tracked it down to the use of the --xvfb flag.

Concretely, when running

./../testing/scripts/run_isolated_script_test.py ../../third_party/blink/tools/run_web_tests.py --seed 4 --no-show-results --zero-tests-executed-ok --clobber-old-results --exit-after-n-failures 5000 --exit-after-n-crashes-or-timeouts 100 --debug-rwt-logging --isolated-script-test-output=~/tmp/foo fast/js/try-catch-crash.html

things are fine, but when running

./../testing/scripts/run_isolated_script_test.py --xvfb ../../third_party/blink/tools/run_web_tests.py --seed 4 --no-show-results --zero-tests-executed-ok --clobber-old-results --exit-after-n-failures 5000 --exit-after-n-crashes-or-timeouts 100 --debug-rwt-logging --isolated-script-test-output=~/tmp/foo fast/js/try-catch-crash.html

my X session is killed along with all running applications and I end up at the display manager login prompt.

This is on rodete 201812RD2.
 
Components: -Infra>Platform>Swarming>WebUI
Owner: neis@chromium.org
Status: Untriaged (was: Assigned)
The bug button on the UI is meant for filing bugs against the swarming UI.

Please re-triage.
Components: Blink>Infra
Cc: kbr@chromium.org
Owner: ----
kbr@, could you please help find an owner for this? Thank you.
Cc: dpranke@chromium.org
Components: Test>LayoutTests
Labels: OS-Linux
Built the blink_tests target and ran the two commands above from within out/Release. On my gLinux workstation, both variants run fine.

$ uname -srvo
Linux 4.18.10-1rodete2-amd64 #1 SMP Debian 4.18.10-1rodete2 (2018-10-09) GNU/Linux

The first command line already seems to start the xvfb server even without the --xvfb flag, which seems strange.

How should we make progress on this?

Cc: robertma@chromium.org
Cc: estaab@chromium.org
Both run fine for me as well, unfortunately. I'm on the same version as Ken.

% uname -srvo        
Linux 4.18.10-1rodete2-amd64 #1 SMP Debian 4.18.10-1rodete2 (2018-10-09) GNU/Linux

Are there any system logs we can look at? I'm not sure where else to start with this.
I can't reproduce the issue either; and I'm also running on the same system as kbr & estaab.

I guess the more important information is the versions of Xorg and Xvfb. On my machine:
Package: xorg
Version: 1:7.7+19

Package: xvfb
Version: 2:1.19.6-1


By the way, regarding comment #4, it is actually an unrelated issue: we don't need --xvfb when running web tests; run_web_tests.py starts Xvfb itself. Sent you a PR: https://chromium-review.googlesource.com/c/chromium/src/+/1405694
Running the same versions here.

$ dpkg-query -l xorg
Desired=Unknown/Install/Remove/Purge/Hold
| Status=Not/Inst/Conf-files/Unpacked/halF-conf/Half-inst/trig-aWait/Trig-pend
|/ Err?=(none)/Reinst-required (Status,Err: uppercase=bad)
||/ Name                    Version          Architecture     Description
+++-=======================-================-================-====================================================
ii  xorg                    1:7.7+19         amd64            X.Org X Window System

$ dpkg-query -l xvfb
Desired=Unknown/Install/Remove/Purge/Hold
| Status=Not/Inst/Conf-files/Unpacked/halF-conf/Half-inst/trig-aWait/Trig-pend
|/ Err?=(none)/Reinst-required (Status,Err: uppercase=bad)
||/ Name                    Version          Architecture     Description
+++-=======================-================-================-====================================================
ii  xvfb                    2:1.19.6-1       amd64            Virtual Framebuffer 'fake' X server

Hi, thanks for taking a look. I played around some more and this doesn't seem to affect the default gnome/cinnamon configuration -- the issue occurs only when running i3-cinnamon (https://github.com/jbbr/i3-cinnamon). uname output and xorg & xvfb versions are the same for me as posted above.

Since mine is probably an exotic configuration, it may not be worth spending more time on debugging this.
Status: WontFix (was: Untriaged)

Sign in to add a comment