What would you like to be added:
As part of https://github.com/kubernetes/k8s.io/issues/158, gcr.io/k8s-staging-csi was created (https://github.com/kubernetes/k8s.io/issues/158#issuecomment-462047593). What we need now are credentials in a secret that the Kubernetes-CSI Prow jobs (https://github.com/kubernetes/test-infra/tree/master/config/jobs/kubernetes-csi) can use to actually push images into that staging area.
Why is this needed:
Currently, images are still being pushed into quay.io from TravisCI while only testing happens in Prow. We want to move entirely to Prow.
/cc
watching since we'd like to do something similar for cluster-api
See slack convo here: https://kubernetes.slack.com/archives/CCK68P2Q2/p1563199619408300?thread_ts=1562332672.163800&cid=CCK68P2Q2
Summary: wg-k8s-infra is onboarding cluster-api as a guinea pig for the promotion process and waiting to see how their adoption goes.
I would love to work on "credentials in a secret" part if no one is working on it currently.
Following up on this thread, there has been work on this area by @Katharine and @thockin to enable GCB builds from Prow.
cloudbuild.yaml file defined similar to: https://github.com/kubernetes-sigs/cluster-api/blob/a2e233cfd8b5a0f4f96f4e9eac3f8e07336483a5/cloudbuild.yamlIssues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.
If this issue is safe to close now please do so with /close.
Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle stale
/cc
/lifecycle frozen
Some more documentation has been added on how to do this:
The question also came up whether "docker buildx" for multiple target architectures can be done there, which is a new requirement (https://github.com/kubernetes-csi/node-driver-registrar/pull/63). So far the answer is "maybe" (https://github.com/kubernetes-csi/node-driver-registrar/pull/63#issuecomment-587728830).
Docker buildx should work fine in GCB or prow.
please do not set qemu / multi-arch / binfmt_misc on prow, as binfmt_misc is kernel global. setting it up on GCB is fine, the underlying VM is disposed after your build.
almost there getting the docker buildx in cloudbuild.yaml in https://github.com/kubernetes-csi/node-driver-registrar to work. still needs https://github.com/kubernetes-csi/node-driver-registrar/pull/76 to merge
Discussion continues in https://github.com/kubernetes-csi/csi-release-tools/issues/86
I think this was done? Poke.
Yes. We are keeping that csi-release-tools issue open because we also started tracking multi-architecture image publishing which still needs to be done for some sidecars, but as far as test-infra is concerned, everything is working.
/close
@pohly: Closing this issue.
In response to this:
Yes. We are keeping that csi-release-tools issue open because we also started tracking multi-architecture image publishing which still needs to be done for some sidecars, but as far as test-infra is concerned, everything is working.
/close
Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.
Most helpful comment
almost there getting the docker buildx in cloudbuild.yaml in https://github.com/kubernetes-csi/node-driver-registrar to work. still needs https://github.com/kubernetes-csi/node-driver-registrar/pull/76 to merge