fix(builder): key GPU interface node selector by interface, not memory size

[neyy91]

What

The GPU interface node-selector requirement in cluster/kube/builder/workload.go was built with gpu.MemorySize in the label key instead of gpu.Interface:

if gpu.Interface != "" {
    Key: fmt.Sprintf("%s.interface.%s", key, gpu.MemorySize) // -> gpu.Interface
}

Why

The inventory operator publishes the node label as <gpu>.interface.<interface> (operator/inventory/node-discovery.go, via info.Interface). Because the builder keyed the selector off the memory size instead, a workload requesting a specific GPU interface (e.g. pcie) produced a node-affinity requirement that can never match a real node label — so the pod fails to schedule.

Testing

Added Test_nodeSelectorsFromResources_GPUInterface: a GPU with memory_size=80Gi, interface=pcie must yield a selector keyed ...interface.pcie, and must NOT yield ...interface.80Gi. Fails before the fix

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

Run ID: 5efad696-6370-4394-b98a-5086760f69c1

📥 Commits

Reviewing files that changed from the base of the PR and between 0e0fd8a and 9046d42.

📒 Files selected for processing (2)

• cluster/kube/builder/workload.go
• cluster/kube/builder/workload_gpu_selector_test.go

---

Walkthrough

A one-line bug fix in cluster/kube/builder/workload.go corrects the GPU node selector key for the interface capability, substituting gpu.Interface for the previously used gpu.MemorySize. A new test in workload_gpu_selector_test.go verifies the corrected selector key is generated and the wrong key is absent.

Changes

GPU Interface Node Selector Fix

Layer / File(s)	Summary
GPU interface selector fix and regression test cluster/kube/builder/workload.go, cluster/kube/builder/workload_gpu_selector_test.go	nodeSelectorsFromResources now embeds gpu.Interface (not gpu.MemorySize) in the interface selector key; Test_nodeSelectorsFromResources_GPUInterface asserts the correct key is present and the memory-size-keyed interface key is absent.

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~3 minutes

Poem

A bunny once hopped through the selector maze,
Where memory-size wore an interface's face.
One line set it right, swapped the key with a hop,
Now interface selects what the GPU should top.
🐇 No more wrong keys — the scheduler's at peace!

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 50.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title accurately describes the main fix: correcting GPU interface node selector to use interface value instead of memory size.
Description check	✅ Passed	The description clearly explains the bug, its impact on pod scheduling, and provides context about why the fix was needed.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

Warning

Tools execution failed with the following error:

Failed to run tools: 13 INTERNAL: Received RST_STREAM with code 2 (Internal server error)

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}