docs: add Getting Started section with prerequisites, quick start, and reference installation by andreliuNV · Pull Request #1227 · NVIDIA/infra-controller

andreliuNV · 2026-04-29T05:44:56Z

Summary

Adds the Getting Started section to the NICo documentation, covering the full deployment path from prerequisites through first host discovery.

Prerequisites (4 pages): Hardware, Network, Software, BMC/OOB — consolidated from site-reference-arch.md, site-setup.md, and networking_requirements.md
Building NICo Containers: repositioned in nav before Quick Start
Quick Start Guide: 7-step deployment walkthrough (build → K8s → configure → deploy → verify → OOB → first host) with site configuration pulled from helm-prereqs/README.md
Reference Installation: manual phase-by-phase installation from SETUP_PHASES.md, plus PKI architecture, PostgreSQL architecture, and troubleshooting
Provisioning (Day 0 Operations): new section with Ingesting Hosts, Host Validation, SKU Validation
helm-prereqs/README.md: slimmed to lightweight pointer to docs site
Old pages removed where fully superseded (site-reference-arch, site-setup, networking_requirements, expected_machine_update, bootstrap)
TLS/SPIFFE page moved from kubernetes/ to development/
CLI naming standardized: admin-cli (gRPC) vs carbide-cli (REST) with distinction table

Context

Part of NICo Docs Code Yellow (FORGE-8168). Follows the IA recommendation from CDEVS-2173 for the Getting Started section structure. Builds on the Overview section work in PR #1190.

Test plan

Preview with fern docs dev — verify all nav links resolve
Walk through Quick Start Guide steps 1-7 for completeness
Verify prerequisite pages cover all content from removed old pages
Verify Reference Installation troubleshooting section is complete
Check landing page persona table links resolve to new page paths
Verify helm-prereqs/README.md links point to published docs URLs

🤖 Generated with Claude Code

lachen-nv · 2026-04-30T02:53:27Z

PR conflict, can you resolve it?

github-actions · 2026-04-30T03:31:01Z

🌿 Preview your docs: https://nvidia-preview-pull-request-1227.docs.buildwithfern.com/infra-controller

andreliuNV · 2026-04-30T03:34:38Z

PR conflict, can you resolve it?

merge conflicts resolved, @shayan1995 since this touches your helm-prereqs README and SETUP-PHASES.md, please review.

shayan1995 · 2026-04-30T22:05:11Z

/ok to test 9ec63e8

benhuntley · 2026-04-30T23:16:41Z

Three issues which I noticed:

API Service (NICo Core) — ... Exposes a debug web UI on /admin for operators via HTTP with OIDC authentication.
Pedantically I think this should be via HTTPS.
PostgreSQL — stores all NICo system state in the forgedb database.
I believe the DB is called "forge_system_carbide" in the code (ex https://github.com/NVIDIA/infra-controller-core/blob/main/helm-prereqs/templates/postgresql.yaml#L49)
Route Server — manages BGP routing between DPUs and the Ethernet fabric via FRR (free range routing). Advertises service VIPs and tenant routes.
Is this accurate? I don't think there is a stand alone Route Server service that runs on the site controller?

andreliuNV · 2026-05-04T05:54:10Z

@benhuntley issues addressed across both #1190 and #1227

andreliuNV · 2026-05-04T05:56:03Z

/ok to test 19507de

andreliuNV · 2026-05-04T05:56:55Z

@Coco-Ben please review

Rewrites the entire Overview section of the NICo documentation and adds a new landing page, following the IA recommendation from CDEVS-2173. Overview section (5 pages): - What is NICo? — intro, "Why NICo exists" (sourced from VDR/Code Yellow), architecture overview with NICo Components and Prerequisite Components matching the architecture diagram, "Where NICo fits" stack diagram - Key Capabilities — hardware readiness, DPU lifecycle, multi-tenancy, trust/attestation, firmware control, deployment flexibility, GB200 rack-scale - Operational Principles — 5 foundational design principles - Day 0/1/2 Lifecycle — three operational phases - Scope and Boundaries — two-column tables showing NICo vs platform responsibilities (renamed from "What NICo Does Not Cover") Landing page (index.md): - Persona-based entry points: Deploy & Operate, Integrate, Evaluate - Quick links to HCL, release notes, FAQs, GitHub repos Other changes: - Replace "NCX Infra Controller" with "NVIDIA Infra Controller" in prose - Replace "GB200/GB300-class AI infrastructure" with "AI factory-scale infrastructure" - Fix /admin UI protocol: HTTP → HTTPS - Fix database name: forgedb → forge_system_carbide - Remove Route Server from NICo Components (not a NICo-deployed service) - Remove hand-wavey/marketing language across all overview pages - Remove duplicative content between overview pages - Add explicit URL slugs to prevent Fern slug mangling (what-is-ni-co) - Replace stale Introduction page (README.md) with new landing page Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> Signed-off-by: Andrew Liu <andreliu@nvidia.com> -e Signed-off-by: Peter Gambrill <pgambrill@nvidia.com>

…d reference installation Adds the Getting Started section covering the full deployment path from prerequisites through first host discovery, following the IA recommendation from CDEVS-2173. Prerequisites (4 pages): - Hardware — site controller specs, compute system requirements, DPU details, BIOS/UEFI settings, link to HCL for supported GPUs/systems - Network — IP sizing formulas, address pools, ASN/VNI allocations, underlay BGP, EVPN overlay (with route server explanation), route-targets, switch config, site controller topology, physical cabling - Software — validated versions for all dependencies with "if you already have / if deploying reference" decision paths, installation order - BMC/OOB Setup — DHCP relay, BMC credentials, expected machines manifest, Redfish requirements Quick Start Guide (7 steps): 1. Build NICo Containers (links to existing build guide) 2. Prepare K8s Cluster (requirements, tools) 3. Configure Site (full checklist from helm-prereqs README) 4. Run setup.sh (phase table, what gets deployed) 5. Verify Site Controller (pod checks, LoadBalancer, Keycloak, carbide-cli) 6. Connect OOB Network (DHCP relay verification) 7. Discover First Host (credentials, manifest, TPM approval) Reference Installation: - Manual phase-by-phase installation (from SETUP_PHASES.md) - PKI architecture (3-layer cert chain) - PostgreSQL architecture (Zalando operator, credential flow) - Full troubleshooting guide Provisioning (Day 0 Operations): - Ingesting Hosts (consolidated with expected machines management) - Host Validation, SKU Validation (moved from Operations) Other changes: - helm-prereqs/README.md slimmed to config reference + pointers to docs - helm-prereqs/SETUP_PHASES.md removed (content in reference installation) - Old pages removed: site-reference-arch, site-setup, networking_requirements, expected_machine_update, kubernetes/bootstrap - TLS/SPIFFE page moved from kubernetes/ to development/ - CLI naming standardized: admin-cli (gRPC) vs carbide-cli (REST) - NTP clarified: not a NICo service, provided via DHCP option 42 - carbide-ntp removed from VIP table (chart doesn't exist) - Database name corrected: forgedb → forge_system_carbide - Landing page persona table updated to reference new pages Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> Signed-off-by: Andrew Liu <andreliu@nvidia.com> -e Signed-off-by: Peter Gambrill <pgambrill@nvidia.com>

-e Signed-off-by: Peter Gambrill <pgambrill@nvidia.com>

…d reference installation (NVIDIA#1227) ## Summary Adds the Getting Started section to the NICo documentation, covering the full deployment path from prerequisites through first host discovery. - **Prerequisites** (4 pages): Hardware, Network, Software, BMC/OOB — consolidated from site-reference-arch.md, site-setup.md, and networking_requirements.md - **Building NICo Containers**: repositioned in nav before Quick Start - **Quick Start Guide**: 7-step deployment walkthrough (build → K8s → configure → deploy → verify → OOB → first host) with site configuration pulled from helm-prereqs/README.md - **Reference Installation**: manual phase-by-phase installation from SETUP_PHASES.md, plus PKI architecture, PostgreSQL architecture, and troubleshooting - **Provisioning (Day 0 Operations)**: new section with Ingesting Hosts, Host Validation, SKU Validation - **helm-prereqs/README.md**: slimmed to lightweight pointer to docs site - Old pages removed where fully superseded (site-reference-arch, site-setup, networking_requirements, expected_machine_update, bootstrap) - TLS/SPIFFE page moved from kubernetes/ to development/ - CLI naming standardized: admin-cli (gRPC) vs carbide-cli (REST) with distinction table ## Context Part of NICo Docs Code Yellow (FORGE-8168). Follows the IA recommendation from CDEVS-2173 for the Getting Started section structure. Builds on the Overview section work in PR NVIDIA#1190. ## Test plan - [ ] Preview with `fern docs dev` — verify all nav links resolve - [ ] Walk through Quick Start Guide steps 1-7 for completeness - [ ] Verify prerequisite pages cover all content from removed old pages - [ ] Verify Reference Installation troubleshooting section is complete - [ ] Check landing page persona table links resolve to new page paths - [ ] Verify helm-prereqs/README.md links point to published docs URLs 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Signed-off-by: Peter Gambrill <pgambrill@nvidia.com> Signed-off-by: Andrew Liu <andreliu@nvidia.com> Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com> Co-authored-by: Peter Gambrill <pgambrill@nvidia.com>

andreliuNV requested a review from a team as a code owner April 29, 2026 05:44

andreliuNV force-pushed the docs-getting-started branch 2 times, most recently from bfeb251 to cff6bad Compare April 29, 2026 15:27

andreliuNV requested review from a team and shayan1995 as code owners April 29, 2026 15:27

andreliuNV force-pushed the docs-getting-started branch 2 times, most recently from 8db8e6f to 3664db9 Compare April 29, 2026 20:19

andreliuNV requested a review from Coco-Ben as a code owner April 29, 2026 20:19

andreliuNV force-pushed the docs-getting-started branch from 3664db9 to 64bc25b Compare April 29, 2026 20:23

andreliuNV requested a review from a team as a code owner April 29, 2026 20:23

andreliuNV force-pushed the docs-getting-started branch from 64bc25b to 7b32e7c Compare April 30, 2026 03:30

andreliuNV force-pushed the docs-getting-started branch from 7b32e7c to 4d41ce9 Compare April 30, 2026 03:33

vigviswa mentioned this pull request Apr 30, 2026

docs: end-to-end installation guide and prerequisite gap fixes #512

Closed

8 tasks

ajf approved these changes Apr 30, 2026

View reviewed changes

andreliuNV mentioned this pull request May 4, 2026

docs: update architecture diagram with missing elements from legacy diagram #1192

Open

andreliuNV force-pushed the docs-getting-started branch 2 times, most recently from 5a7ef10 to 3ae2584 Compare May 4, 2026 05:49

andreliuNV mentioned this pull request May 4, 2026

docs: rewrite Overview section and add landing page #1190

Merged

5 tasks

andreliuNV force-pushed the docs-getting-started branch from 3ae2584 to 19507de Compare May 4, 2026 05:55

Coco-Ben reviewed May 5, 2026

View reviewed changes

Comment thread docs/getting-started/quick-start.md

Coco-Ben approved these changes May 5, 2026

View reviewed changes

andreliuNV and others added 3 commits May 5, 2026 16:15

docs: fix getting started review issues

448c395

-e Signed-off-by: Peter Gambrill <pgambrill@nvidia.com>

Copyedit new overview/quickstart pages

02354fd

-e Signed-off-by: Peter Gambrill <pgambrill@nvidia.com>

Coco-Ben force-pushed the docs-getting-started branch from 80a32c6 to 02354fd Compare May 5, 2026 23:17

Coco-Ben enabled auto-merge (squash) May 5, 2026 23:18

shayan1995 approved these changes May 5, 2026

View reviewed changes

Coco-Ben merged commit 2ce1e99 into NVIDIA:main May 6, 2026
43 checks passed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

docs: add Getting Started section with prerequisites, quick start, and reference installation#1227

docs: add Getting Started section with prerequisites, quick start, and reference installation#1227
Coco-Ben merged 4 commits into
NVIDIA:mainfrom
andreliuNV:docs-getting-started

andreliuNV commented Apr 29, 2026

Uh oh!

lachen-nv commented Apr 30, 2026

Uh oh!

github-actions Bot commented Apr 30, 2026

Uh oh!

andreliuNV commented Apr 30, 2026

Uh oh!

shayan1995 commented Apr 30, 2026

Uh oh!

benhuntley commented Apr 30, 2026 •

edited

Loading

Uh oh!

andreliuNV commented May 4, 2026

Uh oh!

andreliuNV commented May 4, 2026

Uh oh!

andreliuNV commented May 4, 2026

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

Conversation

andreliuNV commented Apr 29, 2026

Summary

Context

Test plan

Uh oh!

lachen-nv commented Apr 30, 2026

Uh oh!

github-actions Bot commented Apr 30, 2026

Uh oh!

andreliuNV commented Apr 30, 2026

Uh oh!

shayan1995 commented Apr 30, 2026

Uh oh!

benhuntley commented Apr 30, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

andreliuNV commented May 4, 2026

Uh oh!

andreliuNV commented May 4, 2026

Uh oh!

andreliuNV commented May 4, 2026

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

benhuntley commented Apr 30, 2026 •

edited

Loading