Remove mode label and do not allow ClusterPolicy & GPUCluster co-existence - #2696
Merged
Merged
Conversation
cdesiniotis
force-pushed
the
remove-mode-label
branch
8 times, most recently
from
July 30, 2026 20:49
3e55040 to
8fb2f3f
Compare
cdesiniotis
marked this pull request as ready for review
July 30, 2026 21:10
cdesiniotis
requested review from
karthikvetrivel,
rahulait,
rajathagasthya,
shivamerla and
tariq1890
as code owners
July 30, 2026 21:10
cdesiniotis
force-pushed
the
remove-mode-label
branch
from
July 30, 2026 21:39
8fb2f3f to
985ac56
Compare
cdesiniotis
commented
Jul 30, 2026
| # clusterPolicy controls whether the chart deploys the ClusterPolicy CR, which | ||
| # manages the classic device-plugin GPU enablement stack. Disable it to run a | ||
| # GPUCluster-only (DRA) installation. | ||
| # The ClusterPolicy CR stores the desired state of the GPU |
Contributor
Author
There was a problem hiding this comment.
The updates to the helm values.yaml file here are the only "user-facing" change in this PR. Let's review this carefully.
rahulait
reviewed
Jul 30, 2026
tariq1890
reviewed
Jul 30, 2026
tariq1890
reviewed
Jul 30, 2026
cdesiniotis
force-pushed
the
remove-mode-label
branch
from
July 31, 2026 00:47
985ac56 to
32712e1
Compare
cdesiniotis
commented
Jul 31, 2026
Comment on lines
+5
to
+7
| {{- if and .Values.gpuCluster.deployCR (and .Values.driver.enabled (not .Values.driver.nvidiaDriverCRD.enabled)) }} | ||
| {{ fail "the NVIDIADriver CRD must be enabled when deploying a GPUCluster CR, set driver.nvidiaDriverCRD.enabled=true" }} | ||
| {{- end }} |
Contributor
Author
There was a problem hiding this comment.
I added this helm validation to ensure the NVIDIADriver CRD is enabled when users try to deploy a GPUCluster CR.
cdesiniotis
force-pushed
the
remove-mode-label
branch
from
July 31, 2026 00:49
32712e1 to
a5273e8
Compare
…tence This commit removes the 'nvidia.com/gpu-operator.resource-allocation.mode' label. The node-labeling-controller no longer adds this label to nodes or uses it when determining what software stack to deploy on a node. In addition, this commit makes it so a ClusterPolicy CR and GPUCluster CR cannot co-exist in the same cluster. Only one software stack (device-plugin or dra) will be deployed in the cluster. Signed-off-by: Christopher Desiniotis <cdesiniotis@nvidia.com>
cdesiniotis
force-pushed
the
remove-mode-label
branch
from
July 31, 2026 00:55
a5273e8 to
9654928
Compare
tariq1890
approved these changes
Jul 31, 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
4 participants
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
This commit removes the 'nvidia.com/gpu-operator.resource-allocation.mode' label. The node-labeling-controller no longer adds this label to nodes or uses it when determining what software stack to deploy on a node. In addition, this commit makes it so a ClusterPolicy CR and GPUCluster CR cannot co-exist in the same cluster. Only one software stack (device-plugin or dra) will be deployed in the cluster.
In the future, we may remove this limitation and allow both stacks to co-exist in the same cluster once we decide on a mechanism for users to choose which stack to deploy per-node.
Checklist
make lint)make validate-generated-assets)make validate-modules)Testing
Manually tested on a single GPU node. Tested several scenarios:
--set clusterPolicy.deployCR=false --set gpuCluster.deployCR=true)