ci(docs): replace lychee with fern broken-links config

[pdmack]

TL;DR

Replace the lychee link-check step in Fern docs CI with fern's native `check.rules.broken-links: error` setting in `docs.yml`, and fix the 14 nav-tree link errors that would otherwise block CI.

Additional Details

Fern's checker validates links against the navigation tree, not just the filesystem. The lychee `--offline` check only caught missing files; fern also catches nav-context resolution errors -- links that exist on disk but resolve to the wrong section in the published docs.

The deprecated `--broken-links` CLI flag and `fern docs broken-links` subcommand both point to this config-based approach as the intended path.

The 14 nav-tree errors were all relative link resolution issues where Fern interprets `./target.md` relative to the page's slug rather than the filesystem directory. Fixes applied to quickstart.md, infrastructure-sizing.md, installation.md, cluster-management/index.md, and cluster-management/self-managed.md. Note: Fern slugifies "gRPC" as `g-rpc`, so links to grpc-load-testing.md use slug `g-rpc-load-testing`.

`fern check` passes with 0 errors after this change.

For the Reviewer

8 files changed: `fern/docs.yml` (warn → error), `fern-docs-ci.yml` (lychee step removed), and 6 `docs/user/` files with corrected nav-tree link paths.

For QA

`fern check` -- 0 errors, 13 warnings (unrelated redirect warnings, tracked in #351).

Issues

Fixes #330

Checklist

• I am familiar with the Contributing Guidelines.
• I have signed off my commits for Developer Certificate of Origin (DCO) compliance.
• New or existing tests cover these changes.
• The documentation is up to date with these changes.

Summary by CodeRabbit

• Documentation
  • Updated documentation links for cluster management, infrastructure sizing, installation, and quickstart guides.
  • Corrected references to local development and fake GPU operator documentation.
• Bug Fixes
  • Broken documentation links are now treated as errors during validation.
  • Improved link validation coverage across the documentation site.

[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: Enterprise

Run ID: d7dea900-c325-4e58-aa6a-e950948f65df

📥 Commits

Reviewing files that changed from the base of the PR and between b47f36a and dc783ff.

📒 Files selected for processing (7)

• .github/workflows/fern-docs-ci.yml
• docs/user/cluster-management/index.md
• docs/user/cluster-management/self-managed.md
• docs/user/infrastructure-sizing.md
• docs/user/installation.md
• docs/user/quickstart.md
• fern/docs.yml

💤 Files with no reviewable changes (1)

• .github/workflows/fern-docs-ci.yml

🚧 Files skipped from review as they are similar to previous changes (5)

• fern/docs.yml
• docs/user/cluster-management/self-managed.md
• docs/user/cluster-management/index.md
• docs/user/quickstart.md
• docs/user/installation.md

---

📝 Walkthrough

Walkthrough

Fern documentation CI now uses Fern’s broken-link validation instead of the Lychee workflow step. Internal documentation links were updated to use Fern-compatible parent-directory and extensionless paths.

Changes

Fern documentation link validation

Layer / File(s)	Summary
Fern check wiring .github/workflows/fern-docs-ci.yml, fern/docs.yml	Removes Lychee link scanning, retains fern check, and configures broken links as errors.
Documentation link path corrections docs/user/cluster-management/*, docs/user/infrastructure-sizing.md, docs/user/installation.md, docs/user/quickstart.md	Updates internal links to parent-directory paths and extensionless documentation targets.

Estimated code review effort: 2 (Simple) | ~10 minutes

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title uses Conventional Commits and accurately describes replacing Lychee with Fern broken-link validation in docs CI.
Linked Issues check	✅ Passed	The PR removes Lychee, adds Fern broken-link validation, and fixes the nav-tree link errors so fern check passes.
Out of Scope Changes check	✅ Passed	All changes are docs link-path fixes or CI/config updates directly needed for the broken-link migration.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

Warning

There were issues while running some tools. Please review the errors and either fix the tool's configuration or disable the tool if it's a critical failure.

🔧 Checkov (3.3.8)

fern/docs.yml

Traceback (most recent call last):
File "/usr/local/bin/checkov", line 2, in
from checkov.main import Checkov
ModuleNotFoundError: No module named 'checkov'

---

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

[github-actions]

🌿 Preview your docs: https://nvidia-preview-ci-fern-broken-links-config.docs.buildwithfern.com/nvcf

[sbaum1994]

Lets switch it to error instead of warn please

[pdmack]

Done -- switched to error in b47f36a along with fixes for the 14 nav-tree errors that would have blocked CI.

The errors were all relative link resolution issues where Fern interprets ./target.md relative to the page's slug rather than the filesystem directory. Fixes:

• quickstart.md, installation.md, infrastructure-sizing.md: changed ./target.md to ../target (no .md extension, one level up so Fern resolves from the version root)
• cluster-management/index.md, self-managed.md: same pattern for fake-gpu-operator
• infrastructure-sizing.md grpc-load-testing links: Fern slugifies gRPC as g-rpc, so the correct slug is g-rpc-load-testing not grpc-load-testing

fern check now passes with 0 errors.