This was requested as a possible response to capacity crunches, e.g. last week when two cloudvirts died at once.
I'm looking at this again, and I really think that emptying the scheduler pool is the right way to do this. It's easy and avoids the complexity of adding yet another hiera setting. Here's what it looks like:
diff --git a/hieradata/eqiad/profile/openstack/eqiad1/nova.yaml b/hieradata/eqiad/profile/openstack/eqiad1/nova.yaml index c68e789..1394efd 100644 --- a/hieradata/eqiad/profile/openstack/eqiad1/nova.yaml +++ b/hieradata/eqiad/profile/openstack/eqiad1/nova.yaml @@ -35,10 +35,4 @@ profile::openstack::eqiad1::nova::physical_interface_mappings: # cloudvirtanXXXX: reserved for gigantic cloud-analytics worker nodes # # -profile::openstack::eqiad1::nova::scheduler_pool: - - cloudvirt1013 - - cloudvirt1025 - - cloudvirt1026 - - cloudvirt1027 - - cloudvirt1028 - - cloudvirt1029 +profile::openstack::eqiad1::nova::scheduler_pool: 
@Bstorm can you live with that? If so, I'll document in the nova troubleshooting runbook and declare this done.
If it's well-documented, I think it's fine. It's not something we should do often or lightly, but it will be a high pressure situation. If it's all-but copy-paste, that's great.
Gerrit makes the undo step pretty easy with the revert button.