distribution-spec VS runc

Back in 2013, a little company called Docker made it really easy to start using containers to package up applications. A big key to their success was the OCI (you can learn about that here), an industry wide initiative to have standards around how we package up our applications. Because of OCI standards, we have hundreds (maybe thousands?) of tools that can be combined to manage and deploy applications. So why aren’t we using this for packaging up Notebooks and AI models as well? It would make deploying, sharing, and managing our models easier for everyone involved.

Kubernetes on the backend used to utilize docker for much of its container runtime solutions. One of the modular features of Kubernetes is the ability to utilize a Container Runtime Interface or CRI. The problem was that Docker didn't really meet the spec properly and they had to maintain a shim to translate properly. Instead users could utilize the popular containerd or cri-o runtimes. These follow the Open Container Initiative or OCI's guidelines on container formats.

OCI stands for Open Container Initiative, and its goal as an organization is to define a specification for container formats and runtime.

https://opencontainers.org/

Here is Containerfile from the repo: https://github.com/ublue-os/bazzite/blob/main/Containerfile

apko allows us to build OCI container images from .apk packages.

Flagger's load testing service can be installed via a Kustomization resource based on manifests packaged as an artifact in an Open Container Initiative (OCI) registry

CRI-O is a lightweight container runtime for Kubernetes. It is an implementation of Kubernetes CRI to use Open Container Initiative (OCI) compatible runtimes for running pods. It supports runc and Kata Containers as the container runtimes, but any OCI-compatible runtime can be integrated.

So somebody else created runj (runj is an experimental, proof-of-concept OCI-compatible runtime for FreeBSD jails.) https://github.com/samuelkarp/runj

I can speak to this. Containers, and by extension k8s, break a well known security boundary that has existed for a very long time - whether you are using a real (hardware) server or a virtual machine on the cloud if you pop that instance/server generally speaking you only have access to that server. Yeh, you might find a db config with connection details if you landed on say a web app host but in general you still have to work to start popping the next N servers.

That's not the case when you are running in k8s and the last container breakout was just announced ~1 month ago: https://github.com/opencontainers/runc/security/advisories/G... .

At the end of the day it is simply not a security boundary. It can solve other problems but not security ones.

It's interesting that, in light of things like this, you still see large software companies adding support for new components written in non-memory safe languages (e.g. C)

As an example Red Hat OpenShift added support for crun(https://github.com/containers/crun) this year(https://cloud.redhat.com/blog/whats-new-in-red-hat-openshift...), which is written in C as an alternative to runc, which is written in Go(https://github.com/opencontainers/runc)...

Rabbit hole indeed. That wasn't related to my job at the time, lol. The job change came with a company-provided computer and that put an end to the tinkering.

BTW, I found my hacks to make runc run on Chromebook: https://github.com/opencontainers/runc/compare/main...gabrys...

being the main author of crun, I can clarify that statement: I am not a fan of Go _for this particular use case_.

Using C instead of Go avoided a bunch of the workarounds that exists in runc to workaround the Go runtime, e.g. https://github.com/opencontainers/runc/blob/main/libcontaine...

runc

Not OP, but if I had to guess, a lot of this can be picked up by just observing common security issues in the Linux space, since similar mistakes and oversights have caused quite a few real-world CVEs in the past, e.g. this random example of a TOCTTOU vulnerability in runc.