Re-enable `test_platform_with_runcmd` in Azure DevOps

[stuartleeks]

As part of #140 the AzDO build is now triggered, but there is a failure in the execution of the test_platform_with_runcmd test, so it has temporarily been commented out in the pipeline. This issue is tracking fixing and re-enabling it.

[2023-01-24T14:42:36.350Z] #13 0.141 sudo: effective uid is not 0, is /usr/bin/sudo on a file system with the 'nosuid' option set or an NFS file system without root privileges?

build log

[stuartleeks]

Updating the Dockerfile for test_platform_with_runcmd to set USER root before the RUN steps that use sudo seemed to workaround the effective uid is not 0 error.

With that in place, there is an error in the skopeo copy stage:

time="2023-01-24T20:59:45Z" level=fatal msg="creating an updated image manifest: preparing updated manifest, layer \"sha256:d699ae6cb5d892663268ad8f5dff7f7122e0d32890571599485779b7596f0e0d\": unsupported MIME type for compression: application/vnd.in-toto+json"

According to this issue, it may be resolved in later versions of skopeo. Next steps would be to check the version installed

[stuartleeks]

I was looking at the installed software on GitHub runners and AzDO runners, and both have the same version 1.4.1 of skopeo. This is lower than the 1.9.0 that is suggested to work in podman-container-tools/skopeo#1874.

This confused me as the error only shows on the AzDO PR builds. However, I spotted that the issue does show in the GH builds on merge to main (it turns out the images aren't pushed on the PR builds).

Example: https://github.com/devcontainers/ci/actions/runs/4016076334/jobs/6898731939

So it looks like the skopeo issue would likely be resolved in GH & AzDO by installing a later skopeo version

[natescherer]

Just ran into this problem myself. actions/runner-images#6180 (comment) works for installing an updated skopeo version until the runner images can be updated.