Fix Local NIM onboarding on RTX 3090 / WSL#236
Closed
benwgarton wants to merge 2 commits into
Closed
Conversation
…ocal-nim # Conflicts: # bin/lib/onboard.js
Collaborator
|
Hey @benwgarton, appreciate you sorting out the local NIM onboarding on RTX 3090 / WSL — that's a setup a lot of people run into issues with. Just a quick ask: there have been a good number of changes to main since this PR (new CI, features, etc.), and a rebase would help us review this with confidence. Could you update against the latest main whenever you get a chance? Looking forward to checking it out! |
mafueee
pushed a commit
to mafueee/NemoClaw
that referenced
this pull request
Mar 28, 2026
* Updated readme * Updated readme * Updated readme * Updated readme * Updated readme
13 tasks
ericksoa
added a commit
that referenced
this pull request
Apr 20, 2026
<!-- markdownlint-disable MD041 --> ## Summary Local NIM onboarding fails at the image pull step because `docker pull nvcr.io/nim/...` requires NGC registry authentication. This adds an NGC API key prompt during onboard that runs `docker login nvcr.io --password-stdin` before pulling the NIM image. The key is masked during input and handled securely via stdin. ## Related Issue Based on the investigation in PR #236. ## Changes - `src/lib/nim.ts`: Add `isNgcLoggedIn()` to check if Docker is already authenticated with nvcr.io, and `dockerLoginNgc()` to login securely via `--password-stdin`. - `src/lib/onboard.ts`: Prompt for NGC API key before NIM image pull when not already logged in. Masked input, one retry on failure. - `test/onboard-selection.test.ts`: Mock `isNgcLoggedIn` in NIM-local selection test. ## Type of Change - [x] Code change (feature, bug fix, or refactor) - [ ] Code change with doc updates - [ ] Doc only (prose changes, no code sample modifications) - [ ] Doc only (includes code sample changes) ## Verification - [x] `npx prek run --all-files` passes - [x] `npm test` passes - [x] Tests added or updated for new or changed behavior - [x] No secrets, API keys, or credentials committed - [ ] Docs updated for user-facing behavior changes - [ ] `make docs` builds without warnings (doc changes only) - [ ] Doc pages follow the [style guide](https://github.com/NVIDIA/NemoClaw/blob/main/docs/CONTRIBUTING.md) (doc changes only) - [ ] New doc pages include SPDX header and frontmatter (new pages only) ## AI Disclosure - [x] AI-assisted — tool: Claude Code --- <!-- DCO sign-off required by CI. Run: git config user.name && git config user.email --> Signed-off-by: zyang-dev <267119621+zyang-dev@users.noreply.github.com> <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **New Features** * Setup wizard enforces NGC Docker authentication for NIM model setup: interactive mode prompts for an NGC API key (one retry); non-interactive mode prints login instructions and exits. * **Bug Fixes / Reliability** * Improved detection and login handling for NGC Docker credentials so image pulls proceed only after successful authentication and failures are reported. * **Tests** * Added unit tests for NGC auth detection and updated onboarding tests to cover authenticated flows. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Signed-off-by: zyang-dev <267119621+zyang-dev@users.noreply.github.com> Co-authored-by: Aaron Erickson 🦞 <aerickson@nvidia.com>
Contributor
|
Thanks for digging into the RTX 3090 / WSL Local NIM onboarding path — the specific gaps you hit here are real. The files this PR targets ( If the NIM onboarding issue persists on the current codebase, we'd welcome a resubmit targeting the TypeScript equivalents in |
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
5 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.
Summary
This PR fixes a Local NIM onboarding path that fails on a common consumer-GPU setup: a single RTX 3090 running under WSL.
What this changes
Why
During validation on an RTX 3090, the original requested model (
vidia/llama-3.3-nemotron-super-49b-v1) had no runnable profile on this hardware. Switching to the smaller Local NIM image (meta/llama-3.1-8b-instruct) worked, but only after:
Without these changes, onboarding either falls back incorrectly or fails even though a working Local NIM configuration exists for this host class.
Scope
This PR keeps the fix narrow and only changes the Local NIM onboarding/runtime path.
Validation
Validated locally with: