https://github.com/ostreedev/ostree/pull/1462
Some things I like specifically: - The qcow2 inventory is useful (though I want it to be more configurable) - Ansible is mostly better than shell script - It's a layer of stuff that's abstract across "execution environments" like Jenkins vs whatever else - The concept that playbooks can in theory be more easily shared
And of course that it makes things a bit more obvious how to reuse *upstream* tests stored in upstream git in the downstream dist-git (though I need to actually plumb that through).
BTW we're also in the process of completely revamping our "CI execution" or https://github.com/projectatomic/papr to use Kubernetes/OpenShift more. In our case then the upstream tests using STI would actually execute as Kubernetes Jobs; in our particular case then we'd be moving strongly towards VM-in-container. See: https://github.com/projectatomic/papr/pull/70 https://github.com/projectatomic/paci/pulls?q=is%3Apr+is%3Aclosed etc.; also part of this is moving our CI into CentOS CI's OpenShift instance.