Yes, that's possible :). Vince, can we pull together a couple of bare-metal builders from all of the win_chromium_rel_ng VMs we returned?

How many beefy VMs would you need?

I think two should be fine. We can get by with one if need be. Did you mean to write "VMs"? I wrote bare metal, but if we can get a big VM that's fine, too.

I can spin up bm's.  You wanted to replace all 4 current win_upload_clang slaves?

Sorry, I got distracted by the title.  Yes, baremetal.

Ah, my mistake then, really :).

As I said, I think we only need two machines for this bot (just for redundancy), but the 4 we have today also are used for a couple of other optional builders.

However, looking at master.tryserver.chromium.win/slaves.cfg, it's way more complicated than it needs to be, so I can probably move builders around and prune a bunch of machines out one way or another.

The following revision refers to this bug:
  https://chrome-internal.googlesource.com/chrome-golo/chrome-golo/+/4b167f8f3508b96d9c513107eb5d921c0b137894

commit 4b167f8f3508b96d9c513107eb5d921c0b137894
Author: Peter Schmidt <pschmidt@google.com>
Date: Fri Oct 13 00:13:35 2017

Does that mean I can start using build{4,6}-m4 ?

Can we try a beefier GCE win10 machine for this or will that not work?

Yes, GCE would work also.

However, as I discussed in the staff mtg yesterday, if we have bare metal machines sitting around doing nothing, which we might since we've moved win_chromium_rel_ng to the cloud, it's probably better to put them to use than it is to spin up new cloud instances. If we have a need for more bare metal testers, that's a better use for them, of course, and if we'd rather retire them and free up the rack space, that's fine, too. Up to y'all.

The following revision refers to this bug:
  https://chrome-internal.googlesource.com/chrome-golo/chrome-golo/+/1c392874f1865058f3c53c2d2e6bb963fd3e41bb

commit 1c392874f1865058f3c53c2d2e6bb963fd3e41bb
Author: Peter Schmidt <pschmidt@google.com>
Date: Mon Oct 16 17:09:29 2017

o.k build{4,6}-m4 are ready.

The following revision refers to this bug:
  https://chrome-internal.googlesource.com/chrome-golo/chrome-golo/+/76a4008858a2ff92033d12f28227067e36f12b9e

commit 76a4008858a2ff92033d12f28227067e36f12b9e
Author: Peter Schmidt <pschmidt@google.com>
Date: Mon Oct 16 19:17:10 2017

Thanks! Hooking them up here: https://chromium-review.googlesource.com/#/c/chromium/tools/build/+/721968

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/tools/build/+/4156b5797de3e05e913b4868927aac82f9fc8221

commit 4156b5797de3e05e913b4868927aac82f9fc8221
Author: Hans Wennborg <hans@chromium.org>
Date: Mon Oct 16 22:00:26 2017

Use build{4,6}-m4 for the win_upload_clang bot

Bug:  773902 
Change-Id: I5df6ca12e79ef4bf55b552f12d18b59616bc977f
Reviewed-on: https://chromium-review.googlesource.com/721968
Reviewed-by: Dirk Pranke <dpranke@chromium.org>
Commit-Queue: Hans Wennborg <hans@chromium.org>

[modify] https://crrev.com/4156b5797de3e05e913b4868927aac82f9fc8221/masters/master.tryserver.chromium.win/master.cfg
[modify] https://crrev.com/4156b5797de3e05e913b4868927aac82f9fc8221/masters/master.tryserver.chromium.win/slaves.cfg

Noted that these were not connected to the master.  Just fixed that.