WIP: Initial workflow (OWNERS, etc.) only tracking secondary metadata #2
Conversation
Release-admins are from [1]. Cincinnati graph approvers are seeded with [2], although we may diverge because data-tooling maintenance will be fairly distinct from developing Cincinnati itself. [1]: https://github.com/openshift/release/blob/cdafdc7392068d5925f0f76fa16b14a9dd20c8c4/cluster/ci/config/roles.yaml#L54-L63 [2]: https://github.com/openshift/cincinnati/blob/fb49dc5c9642889e0d87e74ddc13f8685d7e70c9/OWNERS_ALIASES#L5-L9
openshift-ci-robot
added
the
size/L
wking
changed the title
openshift-ci-robot
added
the
do-not-merge/work-in-progress
|
Thanks @wking, this is much like what I originally had in mind. Ideally I would've liked if pushing the tags could've happened from CI which we have deemed to risky for now in an out-of-band discussion. Which brings me to the most interesting part of the validation rules when to enforce them.
We have a quay v1 API client over at cincinnati. It also has support for token authentication and fetching the labels. It'd be great to extend that with pushing labels. |
wking
force-pushed
the
secondary-metadata-only
branch
from
47d4696
to
0bbd51d
Compare
Alex did this ~two hours ago. Generated with: $ hack/pull
Generated with: $ hack/pull
wking
force-pushed
the
secondary-metadata-only
branch
from
056b1f5
to
5510b7a
Compare
Alex did this ~two hours ago. Generated with: $ hack/pull
wking
force-pushed
the
secondary-metadata-only
branch
from
5510b7a
to
ef564ba
Compare
Alex just did this. Generated with: $ hack/pull
Alex just did this. Generated with: $ hack/pull
Someone did this within the past few hours ;). Generated with: $ hack/pull
Clayton just did this. Generated with: $ hack/pull
I'm catching up after a few days; I'm not actually sure when this happened. Generated with: $ hack/pull
Clayton tagged these in 20 minutes ago.
I haven't tracked down when it was tagged into prerelease, but Jessica tagged it into stable on 2019-10-15T17:22Z.
I haven't tracked down the earlier channels, but the stable-4.2 tag was Jessica eight hours ago. Generated with: $ hack/pull
…te-4.2 We left 4.2.0->4.2.1 out of the 4.2.1 metadata, so cut 4.2.2 with the same payload excepting the fixed previous-version metadata.
Lots of changes. I haven't been paying enough attention to comment on who/when/why. Generated with: $ hack/pull
4.2.11 failed promotion, and we couldn't figure out how to retest. We rebuilt it as 4.1.12, but accidentally tagged 4.2.11 into channels. Remove the incoming edges, and we'll add an off-ramp for 4.2.11->4.2.14.
|
replaced by #4. /close |
|
@wking: Closed this PR. In response to this:
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. |
|
@wking: PR needs rebase. 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. |
openshift-ci-robot
added
the
needs-rebase
While #1 is currently only updating Quay lables, there has been some concern around it positioning itself to be the single source of Cincinnati graph-builder data at some future point. Since my main goal is to get the release-admin workflow into version control, here's a lighter-weight alternative that only manages secondary metadata. Cincinnati would continue to pull primary metadata dynamically from image
release-metadataon Quay, and it would get overrides from this repository.For at least the short term, those overrides would be set in Quay labels just like they are today; we might change that later, but it wouldn't affect the release-admin workflow.
There's also been pushback against hand-edited JSON in #1. But I'm lazy and want these scripts to work even for folks without PyYAML installed ;). We can switch to YAML easily later if folks are comfortable with the workflow.