Katazens –
There are several patches and discussions in place around support for additional hardware platforms in the Kata Containers project. This is great! The three I’m aware of include AMD’s Epyc servers (leveraging SEV), 64B ARM (though I did talk to folks @ Open Stack Summit who were asking about 32b – can you clarify Kevin?) as well as ppc64le.
To facilitate this happening, we need CI in place for these architectures. If it isn’t tested in our CI, we can’t effectively add support.
While we currently leverage Azure VMs for X86 testing, the CI tooling was created to be generic (we currently use Jenkins to drive the testing, though we are currently working with engineers on the Zuul project to see if this would be a better fit). Graham started an issue specific to ARM at [1].
For ARM: I had discussions with Kevin Zhao while at Open Stack Summit, as well with Mohammed @ Vexxhost and Xu @ Hyper.sh. Each mentioned being able to help with appropriate support (whether it be dedicated machines sitting in an engineering department or baremetal machines offered via a cloud provider). There were a lot of conversations, so mea culpa if I’m miss-remembering. I want to coordinate and make sure we have a clear next step here on who can facilitate, and how/when. Who will drive this for ARM?
For PPC and AMD with SEV: we would have similar requirements. As Graham highlighted in [1], the current configuration using Jenkins can be seen at [2], though this would need adjustment to match the specific scenario (this is specific to our azure setup today). Who’s the best point of contact for IBM? Brent, same question for AMD SEV enabled machines (assuming support is to be added)?
Thanks, Eric
[1] https://github.com/kata-containers/ci/issues/30 [2] https://github.com/kata-containers/ci/tree/master/jenkins