backport: minimal SMC GB300 support (#4121) to release/v2.1 - #4645
backport: minimal SMC GB300 support (#4121) to release/v2.1#4645huaweic-nv wants to merge 4 commits into
Conversation
This PR adds minimal support for OpenBMC-based Supermicro GB300 systems and bumps libredfish to `v0.46.3` for the required platform support which brings in the following changes: - feat: minimal SMC GB300 support by @krish-nvidia in NVIDIA/libredfish#111 - fix: pick the correct manager/system on multi-system, multi-chassis hosts by @shevchenko-evgeny in NVIDIA/libredfish#108 Validated on a real SMC GB300. ## Related issues #1909 ## Type of Change - [x] **Add** - New feature or capability - [ ] **Change** - Changes in existing functionality - [ ] **Fix** - Bug fixes - [ ] **Remove** - Removed features or deprecated functionality - [ ] **Internal** - Internal changes (refactoring, tests, docs, etc.) ## Breaking Changes - [ ] **This PR contains breaking changes** ## Testing - [ ] Unit tests added/updated - [ ] Integration tests added/updated - [x] Manual testing performed - [ ] No testing required (docs, internal refactor, etc.) ## Additional Notes --------- Signed-off-by: Krish Dandiwala <kdandiwala@nvidia.com>
|
Auto-sync is disabled for draft pull requests in this repository. Workflows must be run manually. Contributors can view more details about this message here. |
|
Important Review skippedAuto reviews are disabled on base/target branches other than the default branch. Please check the settings in the CodeRabbit UI or the ⚙️ Run configurationConfiguration used: Path: .coderabbit.yaml Review profile: CHILL Plan: Enterprise Run ID: You can disable this status message by setting the Use the checkbox below for a quick retry:
Comment |
🔐 TruffleHog Secret Scan✅ No secrets or credentials found! Your code has been scanned for 700+ types of secrets and credentials. All clear! 🎉 🕐 Last updated: 2026-08-06 02:56:46 UTC | Commit: bb7ff72 |
forge-scout-network.sh ran its readiness checks once, immediately, and exited 0 when the interface was absent, had no carrier, or had no address. On multi-NIC hosts the kernel probes PCI devices in domain order, so the boot NIC can appear tens of seconds into boot -- on a GB300 tray the BlueField-3 is domain 0016 and is probed after four ConnectX-8s, ~24s in, while the loader invokes the script at ~9s. The filter therefore silently no-opped on exactly the hosts it exists to serve, leaving every NIC running DHCP: competing default routes, and an unreachable provisioning server. Poll for the interface to appear, gain carrier and lease an address (SCOUT_NETWORK_WAIT_SECONDS, default 60), and exit non-zero on timeout so the caller can report and retry. Move the loader's invocation inside the download retry loop, guarded so it runs once on success.
4 participants
Backport of SMC GB300 support onto
release/v2.1to build a v2.1 image for onboarding the GB300 trays.Commits
bb7ff72f) — minimal SMC GB300 support: libredfish v0.46.2→v0.46.3 (OpenBMC vendor routing, libredfish FORGE-5259: change default boot url to use golan driver for hosts wit… #111) + bmc-explorer GB300 handling. Cherry-picked frommain(556bdd89).4354b769— include ARM host boot artifacts in BFB image (fork41a957228): packagesscout.efifor aarch64 so Grace/ARM GB300 can HTTP-boot scout.11f862a1— prefer boot NIC during Scout startup (fork8d7ae4c80): addsforge-scout-network.shboot-NIC selection.32d4ae24— wait for boot NIC before filtering Scout DHCP (fork241af6089).Commits 2–4 are cherry-picked from
martinraumann/smcgb300/v2.0-controller-backport(never upstreamed to main; only #4121 was). They apply cleanly onto v2.1.Intentionally NOT included (already in v2.1)
02c0f6ffd"use DHCP link address for prediction promotion" + test4403a8256— redundant: v2.1crates/api-core/src/dhcp/discover.rsalready selectslink_addressover the relay/giaddr (address_to_use_for_dhcp = link_address.as_ref().unwrap_or(&relay_address)).#4145setup-status correlation,#3454host-boot-interface convergence — already onrelease/v2.1.cc @martinraumann — please sanity-check the v2.0→v2.1 adaptation of commits 2–4 (they applied clean, but they're your patches).