New issue
Advanced search Search tips

Issue 891384 link

Starred by 1 user

Issue metadata

Status: Fixed
Owner:
Closed: Oct 11
Components:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 3
Type: Bug



Sign in to add a comment

build36-b4--device2 still flaky

Project Member Reported by bsheedy@chromium.org, Oct 2

Issue description

 Issue 888092  reflashed the device due to it consistently causing one of the VR tests to flake. However, it looks like even after the reflash, the same issue is still occurring.
 
ack. will source a replacement.

Thanks.

I didn't seem to find any Pixel (1) XL replacements, but we do have one spare Pixel (2) XL. Could that possibly work here? If you need a Pixel (1) XL, we can look into purchasing some.

Thanks.

Yes and no - there's nothing wrong with running the tests on a Pixel 2 XL, but I don't know how we'd reliably trigger tests on both the Pixel 1s and 2s. AFAIK, you can't specify something like "device_type = marlin|taimen" for swarming dimensions, so the best we could do is "device_os = O", but then is too broad and may run on other devices with O.

I think the GPU team may have a workaround for that, as I believe they somehow trigger tests on devices with one of two OS versions during an OS upgrade. I can see if that can be applied to device type, as well.
Thanks. We'll be standing by. For now I've removed the suspect device from the bot all-together. Let us know if you need us to order vs upgrade the platform for --device2.

Status: Fixed (was: Assigned)
So the script does look like it should work for triggering tests on both types of devices, but that's a bit of a hack (it's only meant to be used for brief periods, e.g. when upgrading OS versions). AFAIK there haven't been any capacity issues with the device removed, so I think it's fine to leave it without a replacement for now.

Once 874975 is fixed, we can see about running the tests on Pixel 2s instead of  Pixels.

Sign in to add a comment