Re: Custom CI runners in Kata container

On Fri, Jun 13, 2025 at 10:38:38AM +0200, Jörg Rödel wrote:
>Hi Stefano,
>
>On Wed, May 21, 2025 at 03:17:59PM +0200, Stefano Garzarella wrote:
>> as we discussed some weeks ago, I pinged Fabiano Fidencio of the Kata
>> community about their CI.
>>
>> He told me that for now they use runners on machines provided by
>> AMD/Intel to run GitHub jobs and test SEV-SNP/TDX related features.
>
>Thanks for looking into this! However we decide to proceed here, there 
>will be
>some disadvantages.

Cool.
Do you mean that AMD will plan to provide some machines for the project?

As I mentioned I'm in contact with a professor at University of Pisa 
interested on doing some master thesis on SVSM. Do you think we can 
share the access in some way to those machine or they can just use it 
opening PRs and running CI?

>
>> He recommended we take a look at OVHCloud if we have the ability to
>> expense through CCC, because it might be easier to maintain.
>> Particularly if there's downtime, there should be customer service
>> ready to fix them quickly, which we certainly can't require if the HW
>> vendors donate us access to a few machines.
>>
>> If we still want to ask HW vendors, I can ask the Kata community to
>> share some contacts with us to see if we can get machines.
>
>I am a bit hesitant with the expense-way of getting CI hardware, 
>because it
>requires someone to pay the bill in advance before it can be expensed. 
>On the
>other side you are right about the customer service we could rely on.
>
>This is something we could not rely on for gifted resources, of course. 
>But
>still I'd like to pursue this route first, we can still change gears if 
>it
>turns out too unreliable for us.

Yeah, I see, I agree on going on this route first.

Thanks,
Stefano