On Wed, May 30, 2018, at 11:29 AM, Ernst, Eric wrote:
To follow up…
Thanks for proactively setting this up, Clark. It was good to chat with the group. I meant to type up my takeaways last week, but time got away. As I recall, the plan is to have the Zuul folks create an initial setup to track the shim [1] and/or proxy [2] repositories, where an example set of jobs would be created. As discussed, the jobs used in Kata today are quite simple, and are based on kicking off work on Azure.
Clark, can you help fill in the gaps on who we should expect to work with on this (I don’t recall specifically who created the azure portion in zuul, for example).
The first step is to add the openstack zuul github application to these repositories. The process for that is documented at https://docs.openstack.org/infra/manual/drivers.html#hosted-on-github and I've pinged Jonathan asking if he can do that (though I understand he is currently busy with other things so it hasn't happened yet). Once the Github application is installed on the proxy and/or shim repos we can configure Zuul to listen for the web hook events and trigger jobs. I expect that an openstack infra root will work on this. Tristan Cacqueray (tristanC) wrote the nodepool Azure driver and I expect would be happy to have help testing and debugging that. However, I think we can likely start running the first job on vexxhost so that we can work on that in parallel to the nodepool azure driver (this is a cloud we already boot instances on). I'm happy to help coordinate and point people at the other appropriate people to work on $item. I expect this will get easier once we get some of the initial steps done and can point at concrete pieces of configuration and job results.
Thanks!
Eric
[1] https://github.com/kata-containers/shim
[2] https://github.com/kata-containers/proxy
[cid:image001.png@01D3F809.6843E330]
Nice photo :) Hope this helps, Clark