Per Issue 839216, swarm{49..51}-c4 have been allocated for this.

We're going to call this class of GPU trybot "manually triggered" for convention, to distinguish it from the "optional" GPU trybots, which at this point should probably be renamed to something like "extratests".

The following revision refers to this bug:
  https://chrome-internal.googlesource.com/infradata/config/+/014e2f82e9f823de86aeddd6e4f04d0db06c9f07

commit 014e2f82e9f823de86aeddd6e4f04d0db06c9f07
Author: kylechar <kylechar@google.com>
Date: Fri May 04 18:26:59 2018

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/tools/build/+/05327b82f00cf49f3b2c7b24373959eedfefdfa3

commit 05327b82f00cf49f3b2c7b24373959eedfefdfa3
Author: kylechar <kylechar@chromium.org>
Date: Fri May 04 19:27:35 2018

Add gpu_manual_try_win7_nvidia_rel trybot config.

Add trybot config that mirrors the chromium.gpu.fyi "Win7 FYI Release
(NVIDIA)" bot.

Bug:  838895 
Change-Id: I132d6297e986abb574f98713a63f821d1f63f1be
Reviewed-on: https://chromium-review.googlesource.com/1044761
Commit-Queue: kylechar <kylechar@chromium.org>
Commit-Queue: Kenneth Russell <kbr@chromium.org>
Reviewed-by: John Budorick <jbudorick@chromium.org>
Reviewed-by: Kenneth Russell <kbr@chromium.org>

[modify] https://crrev.com/05327b82f00cf49f3b2c7b24373959eedfefdfa3/tests/masters_recipes_test.py
[modify] https://crrev.com/05327b82f00cf49f3b2c7b24373959eedfefdfa3/scripts/slave/recipe_modules/chromium_tests/trybots.py

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/38cb00131ab75ea7e32a7ea6d5fcb377316958b3

commit 38cb00131ab75ea7e32a7ea6d5fcb377316958b3
Author: kylechar <kylechar@chromium.org>
Date: Fri May 04 20:23:18 2018

Add gpu_manual_try_win7_nvidia_rel trybot config.

Add a manually triggered GPU trybot that runs the same configuration
as "Win7 FYI Release (NVIDIA)".

Bug:  838895 
Change-Id: I2a2f87ee7b452200a860282b01c673493adbea00
Reviewed-on: https://chromium-review.googlesource.com/1044767
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Reviewed-by: John Budorick <jbudorick@chromium.org>
Reviewed-by: Andrii Shyshkalov <tandrii@chromium.org>
Commit-Queue: kylechar <kylechar@chromium.org>
Cr-Commit-Position: refs/heads/master@{#556175}
[modify] https://crrev.com/38cb00131ab75ea7e32a7ea6d5fcb377316958b3/infra/config/global/cr-buildbucket.cfg
[modify] https://crrev.com/38cb00131ab75ea7e32a7ea6d5fcb377316958b3/infra/config/global/luci-milo.cfg
[modify] https://crrev.com/38cb00131ab75ea7e32a7ea6d5fcb377316958b3/tools/mb/mb_config.pyl

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/3a8e5c023758c4fad23d7820113351c8fa31e070

commit 3a8e5c023758c4fad23d7820113351c8fa31e070
Author: Kenneth Russell <kbr@chromium.org>
Date: Fri May 04 21:14:49 2018

Document how to add a new manually-triggered GPU trybot.

Bug:  838895 
Change-Id: I4e51e47ce1470ac2fcd8b8a6b465024d9b28b9b4
Reviewed-on: https://chromium-review.googlesource.com/1044750
Reviewed-by: kylechar <kylechar@chromium.org>
Commit-Queue: Kenneth Russell <kbr@chromium.org>
Cr-Commit-Position: refs/heads/master@{#556194}
[modify] https://crrev.com/3a8e5c023758c4fad23d7820113351c8fa31e070/docs/gpu/gpu_testing_bot_details.md

I tried running a patch on gpu_manual_try_win7_nvidia_rel this morning.

https://ci.chromium.org/p/chromium/builders/luci.chromium.try/gpu_manual_try_win7_nvidia_rel/1

It compiled and ran GPU tests so it mostly worked. The patch I tested with added back viz_screenshot_sync_tests which got reverted for failing on this configuration. viz_screenshot_sync_tests ran and failed as expected.

Some tests that ran after viz_screenshot_sync_tests failed. I'll have to test with a patch that's expected to pass and see those tests failed for valid reason or not.

We don't have a lot of Win7 hardware in the pool any more, so that's probably why these jobs failed; they timed out due to lack of capacity.

In infra/config/global/cr-buildbucket.cfg in chromium/src, please try adding an "execution_timeout_secs:" key for gpu_manual_try_win7_nvidia_rel . The default is 60 minutes = 3600 seconds, so try tripling that to 10800 seconds.

Other changes to make:

1) Decommission the "Win7 FYI Debug (NVIDIA)" and "Win7 FYI x64 Debug (NVIDIA)" bots. Prepare a CL in tools/build removing references to it there. Then remove all references to it in Chromium under src/infra/config/global/, src/content/test/gpu/generate_buildbot_json.py , and any other stray references. We need the spare capacity.

2) In src/infra/config/global/cr-buildbucket.cfg, specify execution_timeout_secs for the following bots:
Win7 FYI Release (NVIDIA)
Win7 FYI x64 dEQP Release (NVIDIA)
Win7 FYI x64 Release (NVIDIA)

as 10800 seconds.

Thanks.

What about removing the bots from infradata/config, is that also required?

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/tools/build/+/4a57e9ac698c95b4af43158e116ee1de331401af

commit 4a57e9ac698c95b4af43158e116ee1de331401af
Author: kylechar <kylechar@chromium.org>
Date: Wed May 09 23:02:11 2018

Remove debug Win7 Nvidia GPU bots.

There isn't enough Win7 Nvidia GPU bot capacity to run everything on the
waterfall and allow running manually triggered try jobs. This CL removes
the "Win7 FYI Debug (NVIDIA)" and "Win7 FYI x64 Debug (NVIDIA)" bot
configs.

Bug:  838895 
Change-Id: I7a10eadeba1f6e8f7c596aab04760aeee3cb2627
Reviewed-on: https://chromium-review.googlesource.com/1053109
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Commit-Queue: Kenneth Russell <kbr@chromium.org>

[modify] https://crrev.com/4a57e9ac698c95b4af43158e116ee1de331401af/scripts/slave/recipe_modules/chromium_tests/chromium_gpu_fyi.py

Yes, good point. After the other CLs land, please prepare one which removes them from infradata/config. I'll then help you file an Infra/Labs ticket to reclaim those two VMs.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/5b484f7e8c72518ca8207cccd83cfb1d24bc55e7

commit 5b484f7e8c72518ca8207cccd83cfb1d24bc55e7
Author: Kenneth Russell <kbr@chromium.org>
Date: Fri May 11 01:28:13 2018

Modify Win7 Nvidia GPU bot configurations.

There isn't enough Win7 Nvidia GPU bot capacity to run everything on the
waterfall and allow running manually triggered try jobs. This CL removes
the debug configurations "Win7 FYI Debug (NVIDIA)" and "Win7 FYI x64
Debug (NVIDIA)" from chromium.gpu.fyi waterfall. This CL also increases
the execution timeout for the remaining Win7 Nvidia GPU bots.

Bug:  838895 
Tbr: jbudorick@chromium.org
Cq-Include-Trybots: luci.chromium.try:android_optional_gpu_tests_rel;luci.chromium.try:linux_optional_gpu_tests_rel;luci.chromium.try:mac_optional_gpu_tests_rel;luci.chromium.try:win_optional_gpu_tests_rel
Change-Id: Ia3934d51384de411dbd2e9106213833437b84780
Reviewed-on: https://chromium-review.googlesource.com/1053103
Commit-Queue: Kenneth Russell <kbr@chromium.org>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Cr-Commit-Position: refs/heads/master@{#557751}
[modify] https://crrev.com/5b484f7e8c72518ca8207cccd83cfb1d24bc55e7/content/test/gpu/generate_buildbot_json.py
[modify] https://crrev.com/5b484f7e8c72518ca8207cccd83cfb1d24bc55e7/infra/config/global/cr-buildbucket.cfg
[modify] https://crrev.com/5b484f7e8c72518ca8207cccd83cfb1d24bc55e7/infra/config/global/luci-milo-dev.cfg
[modify] https://crrev.com/5b484f7e8c72518ca8207cccd83cfb1d24bc55e7/infra/config/global/luci-milo.cfg
[modify] https://crrev.com/5b484f7e8c72518ca8207cccd83cfb1d24bc55e7/infra/config/global/luci-scheduler.cfg
[modify] https://crrev.com/5b484f7e8c72518ca8207cccd83cfb1d24bc55e7/testing/buildbot/chromium.gpu.fyi.json

Success! https://ci.chromium.org/p/chromium/builders/luci.chromium.try/gpu_manual_try_win7_nvidia_rel/2

The following revision refers to this bug:
  https://chrome-internal.googlesource.com/infradata/config/+/851281cf97c2c3fb8a4712e8c9feea1fa679123e

commit 851281cf97c2c3fb8a4712e8c9feea1fa679123e
Author: kylechar <kylechar@google.com>
Date: Mon May 14 18:12:47 2018

Fantastic!

Let's call this fixed at this point. kylechar@, thanks for pitching in on the GPU team's infrastructure!