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

Issue 814914 link

Starred by 1 user

Issue metadata

Status: Fixed
Owner:
Closed: Feb 2018
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Fuchsia
Pri: 2
Type: Bug



Sign in to add a comment

Fix Fuchsia test filters for new mojo_unittests target

Project Member Reported by w...@chromium.org, Feb 22 2018

Issue description

Now that Mojo tests have been migrated into a single mojo_unittests target, we have some failing tests, e.g in https://build.chromium.org/p/chromium.fyi/builders/Fuchsia%20%28dbg%29/builds/16563 we see:

[00110.716] 03995.04048> [ RUN      ] FileDataPipeProducerTest.HugeFile
[00110.717] 03995.04048> [       OK ] FileDataPipeProducerTest.HugeFile (45163 ms)

 
I was thinking that was because of the rename too, but I don't see that test in the old filters?
testing/buildbot/filters/fuchsia.mojo_system_unittests.filter was meant to be replaced with testing/buildbot/filters/fuchsia.mojo_unittests.filter, but the removal of fuchsia.mojo_system_unittests.filter somehow got lost during one of the rebases, and then it was changed so fuchsia.mojo_unittests.filter does not reflect it anymore. However, the latest version of fuchsia.mojo_system_unittests.filter doesn't exclude FileDataPipeProducerTest, so not sure what the failure cause could be.

Comment 3 by w...@chromium.org, Feb 22 2018

Re #1: Yes, that is actually a test that I "fixed" myself, to work-around some limit in our tmpfs capacity. I wonder if the new unified test binary is e.g. so big that we don't have enough space for the HugeFile.

The test has now failed for all three runs on Fuchsia Debug since the unified unit-tests binary landed, but is succeeding in Release builds.

When it last passed on Debug, output was:
[00028.587] 03912.03957> [47/56] FileDataPipeProducerTest.HugeFile (21028 ms)

so the test is now taking twice as long to execute, hence the timeout - question is whether that's because the test itself is slower e.g. because the test binaries are bigger, or because there are more/different things running in parallel?

Alas, looks like we should filter the test (again). :-/
Project Member

Comment 4 by bugdroid1@chromium.org, Feb 23 2018

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/6e43881f95391d3ce5a7aa0657c1dceb55efa3e6

commit 6e43881f95391d3ce5a7aa0657c1dceb55efa3e6
Author: Wez <wez@chromium.org>
Date: Fri Feb 23 03:17:02 2018

Filter FileDataPipeProducerTest.HugeFile test under Fuchsia.

This test started taking too long to run after we migrated to a single
monolithic mojo_unittests binary.

Also removes the unused fuchsia.mojo_system_unittests.filter.

Bug:  814914 
Change-Id: I1e666e7706b517806b4e3edad5bee28bdc953dd4
Reviewed-on: https://chromium-review.googlesource.com/933376
Commit-Queue: Scott Graham <scottmg@chromium.org>
Reviewed-by: Scott Graham <scottmg@chromium.org>
Cr-Commit-Position: refs/heads/master@{#538683}
[delete] https://crrev.com/e35fb3af9c9e2ce800e8f5c9bca93cb650c6f738/testing/buildbot/filters/fuchsia.mojo_system_unittests.filter
[modify] https://crrev.com/6e43881f95391d3ce5a7aa0657c1dceb55efa3e6/testing/buildbot/filters/fuchsia.mojo_unittests.filter

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

Owner: w...@chromium.org
Status: Assigned (was: Untriaged)
Closing this out, since it appears to be the only broken test. We'll dig into the _why_ as we burn down the filters.

Comment 6 by w...@chromium.org, Feb 23 2018

Status: Fixed (was: Assigned)

Sign in to add a comment