Skip to content

fix: third-party RDMA unload env var misdescribed as list - #219

Open
andrewwhitecdw wants to merge 1 commit into
NVIDIA:mainfrom
andrewwhitecdw:bugfix/config-schema-third-party-rdma-unload-env-var
Open

fix: third-party RDMA unload env var misdescribed as list#219
andrewwhitecdw wants to merge 1 commit into
NVIDIA:mainfrom
andrewwhitecdw:bugfix/config-schema-third-party-rdma-unload-env-var

Conversation

@andrewwhitecdw

@andrewwhitecdw andrewwhitecdw commented Aug 17, 2026

Copy link
Copy Markdown

This PR addresses the following issue in skills/k8s-network-engineer/references/config-schema.md: third-party RDMA unload env var misdescribed as list, and synchronizes the related skill references identified by Greptile.

Changes

  • skills/k8s-network-engineer/references/config-schema.md: third-party RDMA unload env var misdescribed as list.
  • skills/k8s-launch-kit-config/SKILL.md: clarify that the env var is a boolean flag.
  • skills/k8s-launch-kit-config/references/config-reference.md: clarify that the env var is set to "true".
  • skills/k8s-launch-kit-discover/references/discovery-internals.md: clarify that discovered modules are saved for visibility/warnings only.
  • skills/k8s-launch-kit-troubleshoot/references/common-failures.md: clarify that the env var is a boolean flag.

Details

--- a/skills/k8s-network-engineer/references/config-schema.md
+++ b/skills/k8s-network-engineer/references/config-schema.md
@@ -1,3 +1,5 @@
-When `unloadThirdPartyRDMAModules` is true and dependent modules are discovered,
-the generated NicClusterPolicy includes `UNLOAD_THIRD_PARTY_RDMA_MODULES` env var
-(space-separated module names) in the ofedDriver section.
+When `unloadThirdPartyRDMAModules` is true, the generated NicClusterPolicy
+sets the `UNLOAD_THIRD_PARTY_RDMA_MODULES` environment variable to `"true"`
+(a boolean flag) in the ofedDriver section. Dependent modules discovered on each
+group are saved as `thirdPartyRDMAModules` for visibility and warnings, but are
+not passed as module names to that environment variable.

Tests

Documentation-only fix. make test could not be run in this environment because Go is not installed, but the change is confined to Markdown reference files.

Contributor guidelines

Per this repo's CONTRIBUTING.md:

  • All commits are signed off (Signed-off-by trailer, DCO).
  • Branch contains a single squashed commit.

@greptile-apps

greptile-apps Bot commented Aug 17, 2026

Copy link
Copy Markdown

Greptile Summary

The PR corrects and synchronizes agent-facing guidance for third-party RDMA module unloading.

  • Documents UNLOAD_THIRD_PARTY_RDMA_MODULES as a boolean flag set to "true".
  • Clarifies that discovered module names are retained in thirdPartyRDMAModules for visibility and warnings rather than passed through the environment variable.
  • Aligns configuration, discovery, troubleshooting, and network-engineering references.

Confidence Score: 5/5

The PR appears safe to merge.

No blocking failure remains; the previously reported synchronization issue is resolved by the aligned descriptions across the related skill references.

Important Files Changed

Filename Overview
skills/k8s-launch-kit-config/SKILL.md Correctly distinguishes the boolean manifest flag from the discovered module list.
skills/k8s-launch-kit-config/references/config-reference.md Clarifies that enabling the option sets the OFED container environment variable to "true".
skills/k8s-launch-kit-discover/references/discovery-internals.md Correctly limits discovered module names to visibility and warning purposes.
skills/k8s-launch-kit-troubleshoot/references/common-failures.md Updates redeployment guidance to describe the generated boolean environment flag accurately.
skills/k8s-network-engineer/references/config-schema.md Corrects the original schema description and aligns it with the related skill references.

Reviews (3): Last reviewed commit: "fix: third-party RDMA unload env var mis..." | Re-trigger Greptile

Comment on lines +35 to +39
When `unloadThirdPartyRDMAModules` is true, the generated NicClusterPolicy
sets the `UNLOAD_THIRD_PARTY_RDMA_MODULES` environment variable to `"true"`
(a boolean flag) in the ofedDriver section. Dependent modules discovered on each
group are saved as `thirdPartyRDMAModules` for visibility and warnings, but are
not passed as module names to that environment variable.

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P2 Synchronize the related skill references

This correction now conflicts with skills/k8s-launch-kit-config/SKILL.md, which says discovered modules populate this environment variable, and skills/k8s-launch-kit-discover/references/discovery-internals.md, which says they configure the OFED secondary-module list. Agents consulting these complementary references receive contradictory configuration and troubleshooting guidance, so update those references alongside this correction.

Knowledge Base Used: Agent Skills for k8s-launch-kit

Note: If this suggestion doesn't match your team's coding style, reply to this and let me know. I'll remember it for next time!

Copy link
Copy Markdown
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Addressed — config-schema.md now describes UNLOAD_THIRD_PARTY_RDMA_MODULES as a boolean flag and notes that discovered modules are saved as thirdPartyRDMAModules for visibility, consistent with k8s-launch-kit-config/SKILL.md and k8s-launch-kit-discover/references/discovery-internals.md.

@andrewwhitecdw

Copy link
Copy Markdown
Author

@greptile-apps thanks for the catch. I synchronized the related references in:\n- skills/k8s-launch-kit-config/SKILL.md\n- skills/k8s-launch-kit-config/references/config-reference.md\n- skills/k8s-launch-kit-discover/references/discovery-internals.md\n- skills/k8s-launch-kit-troubleshoot/references/common-failures.md\n\nAll now state that UNLOAD_THIRD_PARTY_RDMA_MODULES is set to "true" as a boolean flag, while discovered modules are saved as thirdPartyRDMAModules for visibility/warnings only.

@greptile-apps

greptile-apps Bot commented Aug 17, 2026

Copy link
Copy Markdown

All five files look correct and consistent. The changes across all references accurately describe:

  1. UNLOAD_THIRD_PARTY_RDMA_MODULES is set to "true" as a boolean flag (not a space-separated module list)
  2. Discovered modules are saved as thirdPartyRDMAModules for visibility/warnings only, not passed to that env var

The synchronization is thorough — every previously incorrect description has been updated. Looks good to merge.

Tip: You can customize Greptile's behavior for this repo with .greptile/rules.md and .greptile/config.json.

Correct the description of unloadThirdPartyRDMAModules in the network-engineer config-schema reference and synchronize the related skill references (k8s-launch-kit-config, k8s-launch-kit-config/config-reference, k8s-launch-kit-discover, k8s-launch-kit-troubleshoot) so they all state that UNLOAD_THIRD_PARTY_RDMA_MODULES is set to "true" as a boolean flag, while discovered modules are saved as thirdPartyRDMAModules for visibility and warnings only.

Signed-off-by: andrewwhitecdw <andrewwhitecdw@users.noreply.github.com>
@andrewwhitecdw
andrewwhitecdw force-pushed the bugfix/config-schema-third-party-rdma-unload-env-var branch from 24e0ca5 to a13da90 Compare August 17, 2026 21:59
@andrewwhitecdw

Copy link
Copy Markdown
Author

@greptile-apps thanks for the review. The cross-file synchronization is already in place: skills/k8s-launch-kit-config/SKILL.md, skills/k8s-launch-kit-config/references/config-reference.md, skills/k8s-launch-kit-discover/references/discovery-internals.md, and skills/k8s-launch-kit-troubleshoot/references/common-failures.md all now describe UNLOAD_THIRD_PARTY_RDMA_MODULES as a boolean flag set to "true", with discovered modules saved as thirdPartyRDMAModules for visibility/warnings only. I squashed the branch to a single commit (a13da90) per the repo contribution guidelines.

@greptile-apps

greptile-apps Bot commented Aug 17, 2026

Copy link
Copy Markdown

All five files look correct and consistent. The changes accurately describe UNLOAD_THIRD_PARTY_RDMA_MODULES as a boolean flag set to "true", and correctly clarify that thirdPartyRDMAModules holds discovered module names for visibility/warnings only — not passed to that env var. The squash to a single commit is clean. LGTM.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant