Skip to content

[Bug]: Published 580.126.20-rhel10.0 image runs real dnf in DTK mode - predates the dnf stub present on main #985

Description

@aqeelat

Describe the bug

main (rhel10/ocp_dtk_entrypoint L182) stubs dnftrue before the vendored nvidia-driver build, making DTK builds repo/entitlement-free. The published 580.126.20-rhel10.0 image does not contain this stub: DTK-mode builds still run real dnf install kernel-headers/-devel/kernel-$(uname -r), which fails without RHEL entitlements or a custom repoConfig repo.

To Reproduce

DTK-mode deployment with digest sha256:869385f84684eb2b6f24c3fb48fbd468107c34bf589e2809ec9e9e3af7a68b64:

+ dnf -q -y --releasever=10 --setopt=install_weak_deps=False install kernel-6.12.0-233.el10.x86_64
Error: Unable to find a match: kernel-6.12.0-233.el10.x86_64

Real repo-resolution output confirms the stub is absent; newer tags' script layers contain it.

Expected behavior

Published images built from current scripts; ideally clarity on the GitHub-mirror ↔ image-build-source relationship so script verification against main is meaningful. Also: the 580.126.20-rhel10.0 tag has been removed from nvcr.io (digest pins still pull) — consumers of that line have no path forward.

Environment

  • gpu-driver-container source: digest sha256:869385f84684eb2b6f24c3fb48fbd468107c34bf589e2809ec9e9e3af7a68b64 (580.126.20-rhel10.0); comparison tag 580.178.04-rhel10.0
  • NVIDIA Driver Version: 580.126.20
  • Host OS: CentOS Stream CoreOS 10 (SCOS)
  • Kernel Version: 6.12.0-233.el10.x86_64
  • Container Runtime: CRI-O 1.34
  • CPU Architecture: x86_64
  • GPU Model: NVIDIA L4
  • Kubernetes Distro: OKD 4.22.0-okd-scos.6
  • GPU Operator version: v26.3.3

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions