Build succeeded.

• fi-ansible--ansible-review-diff : SUCCESS in 3m 12s

I'm not an ansible expert so maybe what you did is equivalent.. We've been using os_control:os_control_stg for the new hosts so it would be:

hosts: os_masters[0]:os_masters_stg[0]:os_control:os_control_stg

I think the [0] syntax is the right one to use here.
The playbook is only installing objects into OCP, so it should be sufficient (and likely better) to only do that for a single target. As we aren't changing things on the OCP nodes themselves, there is no need to target all the hosts in the group.

There is only one host in both the os_control and os_control_stg groups. :)

So, [0] doesn't hurt anything, but with one host there, it doesn't matter at all. It's always going to be the first host.

+1 - @lucab - so this is fine as-is - but we should update our other projects to use the same syntax:

• coreos-ci
• fedora-coreos-pipeline
• coreos-koji-tagger
• coreos-ostree-importer
• fedora-ostree-pruner

Build succeeded.

• fi-ansible--ansible-review-diff : SUCCESS in 2m 05s

Ack, then it's quicker to just update this one to be aligned with the rest; amended.

Can this be merged now or does it need additional scrutiny/delay due to the freeze?

We can merge now if all you are doing is also deploying to ocp4... we would need a freeze break to modify proxies for change in routes tho, so that probibly should be after release.

Yes, this is only preparing the deployment for now. Routes updates can wait till after F36 release.

I'll rebase/merge/deploy this tomorrow, as it got late today and we aren't in hurry.

LGTM

Build succeeded.

• fi-ansible--ansible-review-diff : SUCCESS in 2m 56s

This needs some conditional adjustments for the difference in the internal registry name, docker-registry.default.svc in OCP3 vs image-registry.openshift-image-registry.svc in OCP4. I'll play around a bit with that.