The Kubernetes project welcomes contributions of test results from organizations that execute e2e test jobs. This ensures we have coverage of Kubernetes on more platforms than just those that the Kubernetes project itself can fund or support.
The process is as follows:
- Designate a point of contact (github handle / e-mail / github team) that we can reach out to if needed (eg: mitigating flood/drought of data, assisting with migrations, etc)
- Create a GCS bucket that is world-readable
- Run e2e tests (we recommend using kubetest)
- Store test results in accordance with Gubernator's GCS Layout
- Unfortunately this part is currently left as an exercise to the reader. We use bootstrap to facilitate this and are in the midst of rewriting it to better support external usage.
- Add the GCS bucket info to buckets.yaml via a PR (use the
previously designated github handle for the
contactfield). - Add jobs and dashboards to the testgrid config via
a PR (use the previously designated point of contact info in a comment next to
added
test_groups, or even better in thedescriptionfield for addeddashboard_tabs)
As of this writing, a good example GCS bucket to grep for in this repo would be the
k8s-conformance-openstack bucket.
We are actively working on improving this process, which means that this document may not be kept exactly up-to-date. Feel free to file an issue against this repo when you run into problems.
We prefer test results that are actively kept up to date and maintained. This is especially true for testgrid, where stale dashboards clutter up the UI. We may periodically identify GCS buckets, jobs, or testgrid dashboards that have become more than 90 days stale and remove them via revertible PR.