task: update bundle to v26.4.1 - #3004
Conversation
Signed-off-by: nvidia-ci-cd <svc-cloud-orch-gh@nvidia.com>
Greptile SummaryThe PR refreshes the v26.4.1 OLM bundle's creation timestamp and operator image digest, but the generated CSV now mixes a staging manager image with the previous production related image.
Confidence Score: 4/5This PR should not merge until the stable manager image and relatedImages entry reference the same production pull specification and digest. The CSV directly deploys a new nvstaging image while advertising the old production image for mirroring, so disconnected OLM installations cannot supply the image requested by the manager pod. Files Needing Attention: bundle/manifests/nvidia-network-operator.clusterserviceversion.yaml Important Files Changed
Reviews (1): Last reviewed commit: "task: update bundle to v26.4.1" | Re-trigger Greptile |
| - name: OFED_INIT_CONTAINER_IMAGE | ||
| value: nvcr.io/nvidia/mellanox/network-operator-init-container@sha256:922217cee504f080b2a76c51ec5d17baa562ffe2447bb1e370c01a6b26a7059f | ||
| image: nvcr.io/nvidia/cloud-native/network-operator@sha256:49d8bd43fa4559468ec93a4c52053aaf7f36861d9f802ffe8fd3aa94266f82de | ||
| image: nvcr.io/nvstaging/mellanox/network-operator@sha256:1f2730d93e7f90c007450d164b57d153ab0fe9b7912655cf22d976bcb3da7a2f |
There was a problem hiding this comment.
Related image no longer matches
When this stable bundle is mirrored for a disconnected cluster, relatedImages supplies the old production digest while the manager Deployment requests the new nvstaging digest, causing the operator pod to remain in ImagePullBackOff and preventing installation.
Knowledge Base Used: Deployment and Manifests
Created by the ocp-bundle job in Mellanox/network-operator OCP bundle CI.