fix: update digest lookup to use versioned RepoTags instead of latest#1935
fix: update digest lookup to use versioned RepoTags instead of latest#1935
Conversation
deb4294 to
e789ee7
Compare
|
PTAL @cdesiniotis @tariq1890 |
bundle/manifests/gpu-operator-certified.clusterserviceversion.yaml
Outdated
Show resolved
Hide resolved
bundle/manifests/gpu-operator-certified.clusterserviceversion.yaml
Outdated
Show resolved
Hide resolved
a8eb42b to
6f8277b
Compare
rajathagasthya
left a comment
There was a problem hiding this comment.
This largely looks good to me. Just a couple of questions.
bundle/manifests/gpu-operator-certified.clusterserviceversion.yaml
Outdated
Show resolved
Hide resolved
6f8277b to
f3e6ed9
Compare
|
PTAL @tariq1890 Regarding this section: I have only added TODO comments. |
|
@shivakunv I took a look at some of the sample PRs raised in your fork, and I noticed that only one line gets updated in the |
6e19949 to
c769a2d
Compare
@cdesiniotis Thanks!! Please check my repo PRs opened by renovate: |
|
I see two PRs raised today, https://github.com/shivakunv/gpu-operator-shiva/pull/233 and https://github.com/shivakunv/gpu-operator-shiva/pull/232. Both of these PRs only update the helm values, and not the CSV file. |
@cdesiniotis Please check the snippets below for PR: https://github.com/shivakunv/gpu-operator-shiva/pull/220
|
c769a2d to
eaa6081
Compare
eaa6081 to
2061238
Compare
| # TODO: update to a k8s-driver-manager image from nvcr.io once the next version is published | ||
| # TODO: Add Renovate rule to update this image to the values.yaml image | ||
| - name: vfio-manager-image | ||
| image: ghcr.io/nvidia/k8s-driver-manager:69eec274 |
There was a problem hiding this comment.
Can we use the k8s-driver-manager image and tag from values.yaml here? Not sure why we want to keep it as TODO here.
There was a problem hiding this comment.
@cdesiniotis has intentionally added this for k8s-driver-image.
We need clarification from Chris.
should I update it @cdesiniotis ?
There was a problem hiding this comment.
Right now, the vfio-manager image in both the CSV and values file is pointing to a pre-release image on ghcr.io. Once the next version of the k8s-driver-manager is released, we will update the CSV and values file to point to the released image on nvcr.io.
| # TODO: update to a k8s-driver-manager image from nvcr.io once the next version is published | ||
| # Add Renovate rule to update this image to the values.yaml image | ||
| - name: "VFIO_MANAGER_IMAGE" | ||
| value: "ghcr.io/nvidia/k8s-driver-manager:69eec274" |
There was a problem hiding this comment.
refer last comment
|
Other than few minor comments, the PR looks good to me. |
2061238 to
0c22e75
Compare
0c22e75 to
b875ade
Compare
4c50f01 to
a63305b
Compare
2587090 to
f137ff6
Compare
Signed-off-by: Shiva Kumar (SW-CLOUD) <shivaku@nvidia.com> Signed-off-by: Rajath Agasthya <ragasthya@nvidia.com>
f137ff6 to
2f55809
Compare
| repository: nvcr.io/nvidia/cloud-native | ||
| image: dcgm | ||
| version: 4.5.2-1-ubuntu22.04 | ||
| version: 4.5.2-1-ubi9 |
There was a problem hiding this comment.
Why was this change made?
There was a problem hiding this comment.
Is this so that the image in the CSV file and helm values are identical, and therefore, renovate can update them simaltaneously?

Current Issue:
Renovate always pulls the digest from the latest tag, which results in a SHA difference.
However, the version in the
values.yamlfile is not updated in this scenario, so the digest in the manifests becomes mismatched and stale.Solution:
Open a PR only when the RepoTags value in values.yaml changes.
Then search for that exact digest in
bundle/manifests/gpu-operator-certified.clusterserviceversion.yamlinstead of using the digest from latest.
check open PRs:
https://github.com/shivakunv/gpu-operator-shiva/pulls