Repository navigation
fix(orchestrator): use catalog-specified CUDA toolkit version per OS - #5477
Merged
Merged
Conversation
Signed-off-by: Nagachandan-P <Nagachandan.p@dell.com>
Nagachandan-P
requested review from
RvishankarOMnia,
VenkateswaraVatam,
abhishek-sa1,
snarthan and
sujit-jadhav
as code owners
October 7, 2026 10:04
snarthan
approved these changes
Oct 7, 2026
Signed-off-by: Nagachandan-P <Nagachandan.p@dell.com>
Katakam-Rakesh
approved these changes
Oct 7, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Description of the Solution
This PR fixes the CUDA toolkit installation so the version specified in the active catalog is installed instead of the latest available
cuda-toolkit. Theorchestrator now dynamically resolves the CUDA toolkit package key from the functional layer's NVIDIA group components and passes the resolved version into
the rendered
install_cuda_toolkit.shcloud-init script.Related Issue
Changes
Orchestrator Metadata Resolution (
src/orchestrator/roles/provision_common/tasks/)cuda_toolkit_versionby traversing the catalog: functional layer → NVIDIA group → CUDA toolkit package.^cuda_toolkitsupports both legacycuda_toolkitand per-OScuda_toolkit_<version>package keys.CUDA Toolkit Install Script (
src/orchestrator/roles/provision_common/templates/hpc_tools/)install_cuda_toolkit.sh.j2to usecuda-toolkit-${CUDA_TOOLKIT_VERSION}instead of unversionedcuda-toolkit.CUDA_TOOLKIT_VERSIONdefaults to13.3.1if not provided.Metadata Service Templates (
src/orchestrator/roles/provision_common/templates/metadata_svc/)ms-group-*.yaml.j2templates now passcuda_toolkit_versionvialookup('template', ..., template_vars=...)toinstall_cuda_toolkit.sh.j2.Files Changed
src/orchestrator/roles/provision_common/tasks/configure_metadata_svc.ymlsrc/orchestrator/roles/provision_common/templates/hpc_tools/install_cuda_toolkit.sh.j2CUDA_TOOLKIT_VERSIONin `dnf installsrc/orchestrator/roles/provision_common/templates/metadata_svc/ms-group-slurm_node_x86_64.yaml.j2cuda_toolkit_versionto installsrc/orchestrator/roles/provision_common/templates/metadata_svc/ms-group-slurm_node_aarch64.yaml.j2cuda_toolkit_versionto instalsrc/orchestrator/roles/provision_common/templates/metadata_svc/ms-group-login_compiler_node_x86_64.yaml.j2cuda_toolkit_versiontsrc/orchestrator/roles/provision_common/templates/metadata_svc/ms-group-login_compiler_node_aarch64.yaml.j2cuda_toolkit_versionTesting
CUDA_TOOLKIT_VERSIONper node (13.3.1for 10.0,13.4.2for 10.2)dnf repoquery cuda-toolkit-13.3.1andcuda-toolkit-13.4.2resolved correctlynvcc --versionconfirmed13.3.73on RHEL 10.0 and13.4.92on RHEL 10.2cuda_toolkit_10.0andcuda_toolkit_10.2package versions from catalogBackward Compatibility
^cuda_toolkitregex supports both singlecuda_toolkitpackage and per-OScuda_toolkit_<version>package keys.cuda_toolkit_versionis not resolvable, the script defaults to13.3.1.