What is the problem you're trying to solve

We have seen containerd crashes on the CI - and growingly, as nerdctl gets better, we are hitting containerd (or buildkitd) issues.

Debugging these (when issues are transient) is really hard without any logs.

Describe the solution you'd like

Suggestion is to figure out a way to make daemon logs available on the CI in case of failure, either printed out directly or downloadable as an artifact.

Additional context

No response