To troopers to triage.

That builder used to run on https://chromium-swarm.appspot.com/bot?id=build7-b9

That machine was previously misconfigured, and was fixed in bug 887759.

Consequently, the builder's target_dimensions needs to change. It needs to drop the "cores" and "cpu" dimensions, and change "os" from "Ubuntu-14.04" to "Android". (ie: just remove this line https://boringssl.googlesource.com/boringssl/+/5b21ed6d3975f8b3f786778f43b4910fa7e7a1ef/cr-buildbucket.cfg#11)

davidben@, can you make that change?

The following revision refers to this bug:
  https://boringssl.googlesource.com/boringssl/+/ee859a8c1d3664288c342beb3f862f7b57bd1c99

commit ee859a8c1d3664288c342beb3f862f7b57bd1c99
Author: David Benjamin <davidben@google.com>
Date: Mon Oct 01 20:36:09 2018

Remove "linux" mixin for Android.

In theory this will fix it.

Bug:  chromium:890955 
Change-Id: I8110a89d41d465e806672b365bd9a6c235f1a7a4
Reviewed-on: https://boringssl-review.googlesource.com/32264
Reviewed-by: Adam Langley <agl@google.com>

[modify] https://crrev.com/ee859a8c1d3664288c342beb3f862f7b57bd1c99/cr-buildbucket.cfg

Thanks!

Between this and  issue #890351 , it seems infra changes have been breaking us a bunch of late. Do you have some story for checking whether, say, a team's CI would be left with no bots after a particular change?

Still doesn't seem to be working.
https://ci.chromium.org/p/boringssl/builders/luci.boringssl.ci/android_aarch64/b8933828084233745424

Any ideas what might be going wrong? The swarming page says we're querying for caches: builder_blah which appears to match zero bots. If I remove that, the site says bots exist, but I don't see where that filter is coming from.

Oh or maybe it's something else. There's a default "cpu:" and "cores:" filter in builder_defaults. Maybe I need to clear those...

(Someone from infra wrote this file for us.)

Speculative fix.
https://boringssl-review.googlesource.com/c/boringssl/+/32305

> The swarming page says we're querying for caches: builder_blah which appears to match zero bots.

tandrii points out that there's two tasks and only one of them require's the cache. So it's probably the cpu/cores issue.

The following revision refers to this bug:
  https://boringssl.googlesource.com/boringssl/+/c9ef13fbfb6848c81d5efd3bd05cfb4aae4c6bdf

commit c9ef13fbfb6848c81d5efd3bd05cfb4aae4c6bdf
Author: David Benjamin <davidben@google.com>
Date: Mon Oct 01 23:41:10 2018

Remove cpu/cores filter from Android bots.

This is a speculative fix. I thought it was the cache filter, but
there's a default cpu and cores filter being applied and the Android
bots don't have core and cores set at all (is that a mistake?).

An earlier change moved the cpu filter into the mixins to align the CI
and CQ, so we can just remove that. Remove the cores filter at all, to
further align them. The luci.flex.ci pool doesn't have variations in the
number of cores (aside from Mac which already cancelled the filter), and
we don't particularly care how many cores.

Bug:  chromium:890955 
Change-Id: I9da9b6513da93ea861bb0ab4aa1a6daa8fa35eba
Reviewed-on: https://boringssl-review.googlesource.com/32305
Reviewed-by: Andrii Shyshkalov <tandrii@google.com>
Reviewed-by: David Benjamin <davidben@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>

[modify] https://crrev.com/c9ef13fbfb6848c81d5efd3bd05cfb4aae4c6bdf/cr-buildbucket.cfg

Heading home now. I'll close this tomorrow if it cycles green.

Looks like it's the cpu & cores dims, yeah. Sorry, I misread the buildbucket config and thought that the linux mixin would cover both those as well.

No worries. Thanks for the help!