[thewisenerd]

> we brought down image creation time to mere seconds, even for images that were multiple GiB in size

this sounds interesting; for e.g., was wondering the other day if we could build images without actually pulling base images.. everytime we compile, we copy artifact(s) onto a multi-hundred MB base image which definitely doesn't need to be pulled everytime.

[ImJasonH]

You absolutely can just build things and put them on base images without building or pulling the base image at any point. This is a central feature of ko, a simple container build tool for most Go applications: https://ko.build

(I am a maintainer)

[wofo]

There are many ways to achieve this. If you are just copying artifacts, and not using RUN, then there is in principle no need to download the base image at all. Feel free to message me if you'd like to chat (email is in my profile).

[a_t48]

Does COPY —-link allow this with BuildKit? In principle it should, in practice I’d guess it ends up pulling the base image.

[wofo]

I can't comment on BuildKit, unfortunately, since I haven't used it. My experience comes from building bespoke systems (an image builder and a custom registry) fully from scratch, because we needed to have full control in order to achieve the performance we were aiming at.

[a_t48]

I haven't gone full custom (yet), just forks (except for the pull client), but totally understand.

[wofo]

Controlling both the builder and the registry is super nice btw, because they can work together. If the builder knows some of the layers already exist in the registry, it merely has to create and push the remaining ones (without downloading any of the other layers, not even those from the base image). That gives you near-instant builds once the biggest layers are cached in the registry!

Since builds usually happen in CI, and pulls happen elsewhere (e.g., a kubernetes node), in the end layers are only downloaded when the resulting container image is actually used.

[a_t48]

Base image independent layers is something I’ve pondered about, but doesn’t feel compatible with things like apt. This conversation is giving me more reason to go implement lazy/fuse base layers though. My exports are already pretty fast due not not using tar+deduplication with similar layers, but pulling the base can still take several minutes.

[a_t48]

You can, by using FUSE and lazy pulling files as they are opened. I'm working on doing this, myself.