ANGLE Wrangling

As an ANGLE Sheriff. Your job is to:

  1. Keep the ANGLE Try Waterfall in good working order.
  2. Monitor the Chromium GPU FYI Waterfall and respond to ANGLE bugs.
  3. Control and monitor the ANGLE auto-rollers.
  4. Keep the ANGLE Standalone Testers in good working order.
  5. Keep the SwANGLE Try Waterfall in good working order.
  6. Monitor and respond to ANGLE's Perf alerts

If you're not an ANGLE team member, you can contact us on the public ANGLE project Google group.

Note: It‘s highly recommend that all wranglers install the Chromium Flake Linker extension for inspecting bot builds. It’ll save you a lot of time.

Task 1: Monitor ANGLE CI and Try Testers

Your first job is to keep the ANGLE Try Waterfall healthy. Some failures are expected on this waterfall. Developers might be testing expectedly buggy code. But persistent flakiness and failures should be reported and reverted.

When encoutering an unexpected failure in your CLs please file an ANGLE bug and cc the current ANGLE wrangler. If the failure is unrelated to ANGLE file a Chromium bug and mark the bug as Hotlist-PixelWrangler. Refer to build.chromium.org to find the current ANGLE wrangler and GPU Pixel Wrangler.

NOTE: When all builds seem to be purple or otherwise broken:

This could be a major infrastructure outage. File a high-priority bug using g.co/bugatrooper.

Task 2: Respond to Bugs

ANGLE bugs sometimes make it past the commit queue testing and into the master branch. This can be because of flaky tests or because the failures are specific to system configurations for which we lack full pre-commit testing support.

The Chromium GPU FYI Waterfall waterfall includes a number of these one-off specialized configurations. Monitor this console for persistent breakage that could be related to ANGLE. Also follow the Internals>GPU>ANGLE component on the Chromium issue tracker to be alerted to reports of breakage on the GPU.FYI waterfall. Googlers can use sheriff-o-matic to monitor the health of the GPU.FYI waterfall.

Note that the GPU Pixel Wrangler is responsible for the Chromium bugs. Please file issues with the tag Hotlist-PixelWrangler for bugs that aren't caused by ANGLE regressions.

IMPORTANT: Info to include in bug reports:

  • Links to all first failing builds (eg first windows failure, first mac failure, etc).
  • Related regression ranges. See below on how to determine the ANGLE regression range.
  • Relevant error messages.
  • Set components: Internals>GPU and/or Internals>GPU>ANGLE.
  • cc relevant sheriffs or blame suspects.
  • Set the Hotlist-PixelWrangler label.

How to determine the ANGLE regression range on the GPU.FYI bots:

  1. Open the first failing and last passing builds.
  2. For test failures: record parent_got_angle_revision in both builds.
  3. For compile failures record got_angle_revision.
  4. Use this URL: https://chromium.googlesource.com/angle/angle.git/+log/<last passing revision>..<first failing revision>

Task 3: The Auto-Rollers

The ANGLE auto-roller automatically updates Chrome with the latest ANGLE changes.

  1. Roller health: You will be cc'ed on all rolls. Please check failed rolls to verify there is no blocking breakage.
  2. Chrome Branching: You are responsible for pausing the roller 24h before branch days, and resuming afterwards. See the Chrome Release Schedule.

We also use additional auto-rollers to roll third party libraries into ANGLE once per day:

Please ensure these rollers are also healthy and unblocked. You can trigger manual rolls using the dashboards to land high-priority changes.

NOTE: When Vulkan-Headers roll is broken:

The Vulkan-Tools, Vulkan-Loader, and Vulkan-ValidationLayers repos all depend on the Vulkan-Headers repo. When Vulkan-Headers updates, all of those repos have a roll process managed by LunarG to update them for the new Vulkan-Headers. This usually takes 2-3 business days after the Vulkan-Headers update. If Vulkan-Headers roll fails, pause the roller with a note that it should be re-enabled when the dependent repos have been updated for the latest Vulkan-Headers changes. This will require a manual roll if the two repos have to be rolled in unison. To perform a manual roll, create a CL that updates the DEPS file with the new SHA1 value from the desired checkout of each dependent repo. Once the manual roll lands, re-enable the auto-rollers for the relevant repos.

The autoroller configurations live in the skia/buildbot repository in the autoroll/config folder.

Task 4: ANGLE Standalone Testing

See more detailed instructions on by following this link.

Task 5: Monitor SwANGLE CI and Try Testers

Most important task here is to keep healthy the 2 SwANGLE bots on ANGLE CQ, linux-swangle-try-tot-angle-x64 and win-swangle-try-tot-angle-x86. As well as the 2 SwANGLE bots used for ANGLE rolls on Chromium CQ, linux-swangle-try-x64 and win-swangle-try-x86.

Same instructions as for Task 1 apply here. Some failures on these bots may be due to SwiftShader changes, however. The possible ways to handle these failures are:

  1. If possible, suppress the failing tests in ANGLE, opening a bug to investigate these later.
  2. If it is clear that an ANGLE CL caused a regression, consider whether reverting it or suppressing the failures is a better course of action.
  3. If a SwiftShader CL is suspected, and the breakage is too severe to be suppressed, (a lot of tests fail in multiple suites), it is possible to revert the responsible SwiftShader roll into Chromium and open a SwiftShader bug. SwiftShader rolls into Chromium should fail afterwards, but if the bad roll manages to reland, the autoroller needs to be stopped.

A lower priority task here is to keep healthy all the SwANGLE CI and Try bots.

Task 6: Monitor and respond to ANGLE's perf alerts

Any large regressions should be triaged with a new ANGLE bug linked to any suspected CLs that may have caused performance to regress. If it's a known/expected regression, the bug can be closed as such. The tests are very flaky right now, so a WontFix resolution is often appropriate.