chore: remove experimental k8s/ deployment manifests#2107
Conversation
The k8s/ directory contained an experimental DinD-based Kubernetes deployment sample. Nothing in src/, tests, CI, or docs depends on it. Remove it to reduce maintenance surface.
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Path: .coderabbit.yaml Review profile: CHILL Plan: Pro Plus Run ID: 📒 Files selected for processing (2)
💤 Files with no reviewable changes (2)
📝 WalkthroughWalkthroughRemoved Kubernetes-related artifacts and automation: deleted Changes
Estimated code review effort🎯 2 (Simple) | ⏱️ ~10 minutes Poem
🚥 Pre-merge checks | ✅ 3✅ Passed checks (3 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Comment |
The test validates the now-deleted k8s/nemoclaw-k8s.yaml.
Both exist solely to validate the now-deleted k8s/nemoclaw-k8s.yaml.
|
After #2045 lands, the rebase of this branch will need to also delete |
Keep k8s/README.md and k8s/nemoclaw-k8s.yaml from this branch (the docker socket proxy changes). Accept all other deletions from #2107 (installer-hash-check workflow, check-installer-hash.sh, AGENTS.md k8s row, .test.js file).
The test asserted against k8s/nemoclaw-k8s.yaml which was removed in #2107. The remaining change is the rootless DinD daemon.json path fallback in preflight.ts.
The k8s/ directory and its associated hash-check script/workflow were removed on main (#2107). This PR's versions of check-installer-hash.sh and installer-hash-check.yaml are retained because they now serve the Ollama installer hash verification. Removed the NemoClaw k8s installer registry entry that referenced the deleted k8s/nemoclaw-k8s.yaml. Signed-off-by: Aaron Erickson <aerickson@nvidia.com>
## Summary After removing the experimental k8s manifests in #2107, this PR retains the one non-k8s change from the original docker-socket-proxy work: `readDockerDefaultCgroupnsMode()` in `preflight.ts` now checks both the standard `/etc/docker/daemon.json` and the rootless DinD path `/home/rootless/.config/docker/daemon.json` when reading the default cgroupns mode. Previously only `/etc/docker/daemon.json` was checked, so rootless DinD setups would always report `"unknown"` for cgroupns mode. ## Changes - **`src/lib/preflight.ts`**: `readDockerDefaultCgroupnsMode()` iterates over both daemon.json paths, returning the first valid mode found. ## Test plan - [x] `npm run typecheck:cli` passes - [x] `npm test` passes <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Bug Fixes** * Improved detection of Docker configuration settings by checking multiple standard configuration locations, making the system more resilient to different Docker installation types and setups. The tool now gracefully falls back through multiple paths instead of failing immediately if a configuration location is unavailable. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
|
@ericksoa Now that NemoClaw has remove k8s support, I wonder if that means you officially don't want people to run NemoClaw on Kubernetes cluster? |
|
@ryanzhang-oss not at all! Just means we are going to be handling that on the OpenShell layer, so it helps with any agent or system that operates within the sandbox. More to come there! |
Summary
Remove the
k8s/directory containing the experimental DinD-based Kubernetes deployment sample. Nothing insrc/, tests, CI, or docs depends on it.What's removed
k8s/nemoclaw-k8s.yaml— sample manifest (privileged DinD pod with nested k3s)k8s/README.md— usage instructionsCLAUDE.md/AGENTS.mdWhy
Signed-off-by: Aaron Erickson aerickson@nvidia.com
Summary by CodeRabbit