Offline/Airgapped Box Caching

[shawngmc]

Is it possible to have an offline cache of box images to point to on your local machine or a local HTTP server?

I see that the box definitions are here and the download URLs are configured here. They use an obs:// URL, and I'm not sure what the OBS URL handler actually does - non-standard HTTP/S port or something completely different?

If not, as an enhancement, this has a few benefits:

• If I can preload images in an build AMI on a metal-class EC2 instance, I can minimize build time on the expensive instance type.
• If I can host them in S3, I can still pull them faster.
• I can reduce costs borne by OBS.
• I can still build locally if my network connection is down, if I'm airgapped, or if the latest image has an issue for me.
• I can do verification of images before I use them.

Make no mistake, I think the ability to just run and pull is excellent for a lot of use cases - there are just odd edge cases, and I'm odd. :)

[shawngmc]

So, from this comment on another ticket, I now know what the obs:// handler does:

Note: If I edit my /usr/local/lib/python3.10/dist-packages/kiwi_boxed_plugin/config/kiwi_boxed_plugin.yml and change the source from using obs:// to the direct http://download.opensuse.org/repositories/, it starts working. Not sure if that is a valid long-term solution.

Originally posted by @neilgierman in #63

So, assuming I pull down the images and host them on an HTTP server, I should be able to self host it by modifying that config file?

And, per this, it looks like I can place the config file in a pre-checked location.