runner-images VS Moby

Yeah this is a good option if you'd like something to deploy yourself! You can also build an AMI from GitHub's upstream image definition (https://github.com/actions/runner-images/tree/main/images/ub...) if you'd like it to match what's available in GitHub-hosted Actions.

With Depot, we're moving towards deeper performance optimizations and observability than vanilla GitHub runners - we've integrated the runners with a cache storage cluster for instance, and we're working on deeper integration with the compute platform that we built for distributed container image builds - as well as expanding the types of builds we can process beyond Actions and Docker, for instance.

But different options will be better for different folks, and the `philips-labs` project is good at what it does.

Whoa, there's a lot of stuff in there [1] that gets installed straight from vendors, without pinning content checksums to a value known-good to Github.

I get it, they want to have the latest versions instead of depending on how long Ubuntu (or, worse, Debian) package maintainers take to package stuff into their mainline repositories... but this attack surface is nuts.

[1] https://github.com/actions/runner-images/tree/main/images/ub...

I had a similar experience with ARC (actions-runner-controller).

One of the machines in the fleet failed to sync its clock via NTP. Once a job X got scheduled to it, the runner pod failed authentication due to incorrect clock time, and then the whole ARC system started to behave incorrectly: job X was stuck without runners, until another workflow job Y was created, and then X got run but Y became stuck. There were also other wierd behaviors like this so I eventually rebuilt everything based on VMs and stopped using ARC.

Using VMs also allowed me to support the use of the official runner images [0], which is good for compatibility.

I feel more people would benefit from managed "self-hosted" runners, so I started DimeRun [1] to provide cheaper GHA runners for people who don't have the time/willingness to troubleshoot low-level infra issues.

[0]: https://github.com/actions/runner-images

Reminds me: Still waiting for native ARM support on GitHub Actions https://github.com/actions/runner-images/issues/5631

Used https://github.com/actions/runner-images to get the packages needed for Ubuntu 22.04 As the packer requires a builder, I used "null" builder to set it as localhost ref: https://developer.hashicorp.com/packer/docs/builders/null (It was way difficult to figure it out the 1st time) I had to modify the .pkr.hcl file to pick my provisioners. I could not understand the use of /opt/hostedtoolcache folder (which I did later)

Having been featured in our February 2023, and January 2024 Release Radars, Moby is the original Linux Container runtime. This new version adds a bunch of changes to the Docker CLI and Moby itself with additional features. There's bug fixes and enhancements, with the main thing for users to be on the look out for containers that were created using Docker Engine 25.0.0. These containers might have duplicate MAC addresses, and thus must be recreated. The same goes for those containers created with Moby 25.0+ and with user defined MAC addresses. Read up on all these changes in the release notes.

Formlabs does this as well for their 3d printers, my earliest encounter of this was when Docker started getting popular: https://github.com/moby/moby/blob/master/pkg/namesgenerator/...

Try to use moby instead since that is the engine in Docker.

https://github.com/moby/moby

> Podman is designed to help with this by providing stronger default security settings compared to Docker. Features like rootless containers, user namespaces, and seccomp profiles, while available in Docker, aren't enabled by default and often require extra setup.

Seccomp has been enabled by default since 2015: https://github.com/moby/moby/pull/18780

It is true that Rootless isn't enabled by default but its "extra setup" can be done with a single command (`dockerd-rootless-setuptool.sh install`)

Perhaps.

Thing is, https://github.com/moby/moby/blob/670bc0a46c4ca03b75f1e72f73... is using https://github.com/mistifyio/go-zfs which features code like `out, err := zfsOutput("get", "-H", key, d.Name)` (Source: https://github.com/mistifyio/go-zfs/blob/master/zfs.go#L315) to get a single zfs property.

Somebody chose to use a library as abstraction that looks good but is implemented as a MVP (nothing wrong with that). "In the future, we hope to work directly with libzfs" should have raised an alarm somewhere, though.

AppArmor can restrict /proc and this is even used by docker: https://github.com/moby/moby/blob/master/contrib/apparmor/te...