autotests always fails to guest login (socket bind denied once waiting a few minutes after boot) |
||||||||||||||
Issue description
The informational builders (alex, lumpy, tricky) have turned red one or more of the following tests failing:
Chrome
Cr-Commit-Position: refs/heads/master@{#475088}
Revision bc58befabe87e204650c836793204bac887992bd
CHROMEOS_BUILD=9592
security_ProfilePermissions.guest: FAIL: Unhandled WebSocketConnectionClosedException: Connection is already closed., 10 reports
login_CryptohomeIncognito: FAIL: Unhandled WebSocketConnectionClosedException: Connection is already closed.
,
May 27 2017
No repro either with 9592. It looks like caroline is turning red right now.
,
May 27 2017
Setting up a lumpy now.
,
May 27 2017
https://uberchromegw.corp.google.com/i/chromeos.chrome/builders/lumpy-tot-chrome-pfq-informational/builds/22625 Lumpy reproduces already on stock lumpy-release/R60-9591.0, even though the informational builder was green with 9591. The first few times after a boot the test passes. Wait a little, and the test will start failing (unable to login, Chrome goes back to OOBE) each time until next reboot. Chances are this problem was in the build for a while, even though wmatrix says it is green https://wmatrix.googleplex.com/unfiltered?hide_missing=True&releases=tot&tests=login_CryptohomeIncognito 20:02:09 INFO | Websocket at port 23 not yet available: [Errno 111] Connection refused 20:02:14 INFO | Websocket at port 23 not yet available: [Errno 111] Connection refused [...] 20:11:15 INFO | Invoking Oobe.guestLoginForTesting Letting it idle for a few minutes we never see the "Invoking Oobe.guestLoginForTesting" spew.
,
May 27 2017
I think this is coming from here https://cs.chromium.org/chromium/src/third_party/catapult/telemetry/telemetry/internal/backends/chrome/oobe.py?type=cs&q=guestLoginForTesting+package:%5Echromium$&l=55
,
May 27 2017
Identical incognito login problems with security_ProfilePermissions Seems to be connected to ui spew [5097:5164:0526/212939.233244:ERROR:socket_posix.cc(137)] bind() returned an error, errno=13: Permission denied [5097:5164:0526/212939.233360:ERROR:socket_posix.cc(137)] bind() returned an error, errno=13: Permission denied [5097:5164:0526/212939.233398:ERROR:devtools_http_handler.cc(226)] Cannot start http server for devtools. Stop devtools.
,
May 27 2017
Parameters ERROR:socket_posix.cc(137)] bind() returned an error, errno=13, socket_fd_=107, addr=0x7f8c9366f508, len=16: Permission denied ERROR:socket_posix.cc(137)] bind() returned an error, errno=13, socket_fd_=107, addr=0x7f8c9366f508, len=28: Permission denied ERROR:socket_posix.cc(137)] bind() returned an error, errno=13, socket_fd_=112, addr=0x7f7c245e9508, len=16: Permission denied ERROR:socket_posix.cc(137)] bind() returned an error, errno=13, socket_fd_=112, addr=0x7f7c245e9508, len=28: Permission denied
,
May 27 2017
I assume the problem started in the VM with the telemetry roll below. But it is likely it just exposed an existing problem. https://chromium-review.googlesource.com/517285 https://codereview.chromium.org/2904193002
,
May 27 2017
lumpy-release/R60-9500.0.0 is solid, not problems. So this must be a more recent regression. I find the telemetry patch series in issue 720317 interesting. Trying newer images.
,
May 27 2017
lumpy-release/R60-9550.0.0 is good.
,
May 27 2017
I think there are 2 regressions: 1) Making VMTest fail immediately between 9591.0.0 and 9592.0.0 (plus new Chrome) as discussed in #9. 2) Making test on lumpy fail when waiting a few minutes with pure release images between 9589.0.0 (probably good) and 9590.0.0 (verified bad).
,
May 27 2017
You need to find way to put cros test on chromium CQ, otherwise we just keep chasing fires here.
,
May 27 2017
,
May 29 2017
Welp. Where Ned is right he is right.
,
May 30 2017
I need to pass this to the next gardener. I know next to nothing about login.
,
May 31 2017
I'm inclined to revert [1] mid-day tomorrow (even if it isn't the root cause) simply to get the tree green, especially given that the patch is a small code cleanup. Reverting locally fixes the failing tests. I haven't yet been able to determine the cause of the failures. 1: https://chromium.googlesource.com/external/github.com/catapult-project/catapult.git/+/02e58de92d78a5d8455a5973ef43b6e7c3cca48b
,
May 31 2017
Ok to revert. But please help me find which assumption in CrOS test that CL is breaking so I can reland it. Reducing the overhead between Telemetry & browser in general is good for perf testing.
,
May 31 2017
Revert didn't apply cleanly so I put together a small patch that does the same thing at [1]. 1: https://codereview.chromium.org/2913213002
,
May 31 2017
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/e3db9988b2c7e8b41f75b9d326edad68cdb0b184 commit e3db9988b2c7e8b41f75b9d326edad68cdb0b184 Author: jdufault <jdufault@chromium.org> Date: Wed May 31 23:18:44 2017 Roll src/third_party/catapult/ 24245dde6..e6ad06342 (3 commits) https://chromium.googlesource.com/external/github.com/catapult-project/catapult.git/+log/24245dde6fc8..e6ad06342efb $ git log 24245dde6..e6ad06342 --date=short --no-merges --format='%ad %ae %s' 2017-05-31 dtu [pinpoint] Move Change.Validate() into Change.FromDict(). 2017-05-31 jdufault cros: Fix bots by using a different websocket to get system info. 2017-05-31 rnephew [Telemetry] Pass browser finder options to story expectations methods to determine disabling. Created with: roll-dep src/third_party/catapult TBR=nednguyen@chromium.org BUG=726917 Review-Url: https://codereview.chromium.org/2920583002 Cr-Commit-Position: refs/heads/master@{#476085} [modify] https://crrev.com/e3db9988b2c7e8b41f75b9d326edad68cdb0b184/DEPS
,
May 31 2017
Ned - steps to reproduce this bug: Launch a cros VM using the instructions in https://go/cros-vm From your catapult checkout: telemetry/bin/run_tests --browser=cros-chrome-guest --remote=localhost --remote-ssh-port=9222 testBrowserCreation
,
May 31 2017
I will wait for the test to green again before working on validating this.
,
May 31 2017
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/f471b73c30b2b9bc7f2684c850e30299d0abc366 commit f471b73c30b2b9bc7f2684c850e30299d0abc366 Author: Jacob Dufault <jdufault@google.com> Date: Wed May 31 23:55:47 2017 Roll src/third_party/catapult/ 24245dde6..e6ad06342 (3 commits) https://chromium.googlesource.com/external/github.com/catapult-project/catapult.git/+log/24245dde6fc8..e6ad06342efb $ git log 24245dde6..e6ad06342 --date=short --no-merges --format='%ad %ae %s' 2017-05-31 dtu [pinpoint] Move Change.Validate() into Change.FromDict(). 2017-05-31 jdufault cros: Fix bots by using a different websocket to get system info. 2017-05-31 rnephew [Telemetry] Pass browser finder options to story expectations methods to determine disabling. Created with: roll-dep src/third_party/catapult TBR=nednguyen@chromium.org BUG=726917 Review-Url: https://codereview.chromium.org/2920583002 Cr-Original-Commit-Position: refs/heads/master@{#476085} Review-Url: https://codereview.chromium.org/2921533002 . Cr-Commit-Position: refs/branch-heads/3116@{#3} Cr-Branched-From: 64a9be37ccf1e29e9e9cc302da2297f2d3661b5d-refs/heads/master@{#475721} [modify] https://crrev.com/f471b73c30b2b9bc7f2684c850e30299d0abc366/DEPS
,
Jun 1 2017
Tests are now green - passing to nednguyen@ to eventually re-land.
,
Jun 1 2017
,
Sep 21 2017
,
Aug 2
,
Aug 10
|
||||||||||||||
►
Sign in to add a comment |
||||||||||||||
Comment 1 by ihf@chromium.org
, May 26 2017