You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
On the Confidential containers call on 4th August there was a discussion about what we had remaining to cut the first release. This mostly came down to creating an e2e CI/CD process for the operator and the removal of skopeo, which this issue is going to talk about.
Current status
At the moment we have some Kata CI signature verification tests, but they only work when skopeo is installed in the Kata guest. We also have the get_resource endpoint implementation for the sample_kbc in the attestation-agent, but there aren't any integration tests of this yet to show/test how it works with image-rs. It sounds like this kbc implementation, should be very similar to the offline_fs_kbc that we want to implement as that's what our Kata CC CI/CD pipeline uses at the moment, so it's a good base to start from.
There is also the gap of skopeo supporting authenticated container registries, which we can't currently do with image-rs in and e2e scenario, but that's part of kata-containers/kata-containers#4601 and not been agreed and broken down to the same detail yet.
Goal
Our goal for the first release (with respected to signature verification), is to get the Kata CC tests of signature verification using local attestation working without skopeo needing to be installed in the guest image and then remove the skopeo and umoci binaries from the operator payload.
Roadmap for achieving this
This is my proposed path of how we can get to the goal. I'm still fleshing out the detail on some of the issues, but feel free to add comments or questions:
At this point I think we've met the MVP of removing and testing the skopeo - image-rs gap of support signature verification e2e, but there are some further things we might like to do
Think about whether we can merge the offline_fs_kbc and sample_kbc - As a longer term item I think it's worth trying to work out what the difference between the two (or more) local attestation kbcs is and whether we can/should merge them together rather than duplicating the code, but lets get them to work first and worry about this post release.
The text was updated successfully, but these errors were encountered:
stevenhorsman
changed the title
Roadmap from removing skopeo - image-rs signature verification gap for the first release
Roadmap for removing skopeo - image-rs signature verification gap for the first release
Sep 1, 2022
@stevenhorsman is this issue still relevant or can be closed?
If it's still relevant to what release do you think we should map it to (mid-November, end-December, mid-February etc...)?
On the Confidential containers call on 4th August there was a discussion about what we had remaining to cut the first release. This mostly came down to creating an e2e CI/CD process for the operator and the removal of skopeo, which this issue is going to talk about.
Current status
At the moment we have some Kata CI signature verification tests, but they only work when skopeo is installed in the Kata guest. We also have the get_resource endpoint implementation for the
sample_kbc
in theattestation-agent
, but there aren't any integration tests of this yet to show/test how it works with image-rs. It sounds like this kbc implementation, should be very similar to theoffline_fs_kbc
that we want to implement as that's what our Kata CC CI/CD pipeline uses at the moment, so it's a good base to start from.There is also the gap of skopeo supporting authenticated container registries, which we can't currently do with image-rs in and e2e scenario, but that's part of kata-containers/kata-containers#4601 and not been agreed and broken down to the same detail yet.
Goal
Our goal for the first release (with respected to signature verification), is to get the Kata CC tests of signature verification using local attestation working without skopeo needing to be installed in the guest image and then remove the skopeo and umoci binaries from the operator payload.
Roadmap for achieving this
This is my proposed path of how we can get to the goal. I'm still fleshing out the detail on some of the issues, but feel free to add comments or questions:
At this point I think we've met the MVP of removing and testing the skopeo - image-rs gap of support signature verification e2e, but there are some further things we might like to do
offline_fs_kbc
andsample_kbc
- As a longer term item I think it's worth trying to work out what the difference between the two (or more) local attestation kbcs is and whether we can/should merge them together rather than duplicating the code, but lets get them to work first and worry about this post release.The text was updated successfully, but these errors were encountered: