[state-device-plugin] infer feature enablement envars from kernel modules#2525
Merged
Conversation
cdesiniotis
reviewed
Jun 8, 2026
fc4e2ea to
3d4f1aa
Compare
00ea318 to
8ebad89
Compare
cdesiniotis
reviewed
Jun 9, 2026
9a10e36 to
ef03eb7
Compare
cdesiniotis
approved these changes
Jun 9, 2026
tariq1890
commented
Jun 10, 2026
| echo "GDS_ENABLED=true" >> feature-flags.env | ||
| fi | ||
| if [ -z "$MOFED_ENABLED" ]; then | ||
| echo "MOFED_ENABLED=true" >> feature-flags.env |
Contributor
Author
There was a problem hiding this comment.
@rahulait @cdesiniotis Note that I am pushing this change right now.
Suggested change
| echo "MOFED_ENABLED=true" >> feature-flags.env | |
| echo "MOFED_ENABLED=true" >> feature-flags.env |
ef03eb7 to
8dc251a
Compare
…ules Signed-off-by: Tariq Ibrahim <tibrahim@nvidia.com>
8dc251a to
18220f4
Compare
Contributor
Author
|
/cherry-pick release-26.3 |
Contributor
|
🤖 Backport PR created for |
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Problem
This PR is linked to NVIDIA/k8s-device-plugin#1837. After PR NVIDIA/k8s-device-plugin#1550 was merged to support
NVIDIADriveruse cases, we observed that default enablement of feature flags especiallyMOFED_ENABLEDwas leading to disruptive side effects. Users of GPU Operator v26.3.2 (which includes k8s-device-plugin v0.19.2) were observing that GPU workload containers were getting the entire list ofibverbsdev nodes injected into them. This led to failures in RDMA/NCCL-based workloads where the workloads were referencing the wrong NICs as opposed to only picking the appropriate the NICs suited for that workload. See here for an example - NVIDIA/k8s-device-plugin#1692.Solution
Instead of unconditionally enabling the feature flags, we enable them dynamically based on the kernel modules that are currently loaded on the node. This dynamic enablement of feature flags will ensure that
ClusterPolicy,NVIDIADriverand pre-installed driver use cases are supported without depending on a cluster-wide setting in the daemonset spec.