looksLikeGPU now falls back to resolving VendorId/DeviceId through the
pci.ids database when the BMC leaves Name/Model/Manufacturer/ClassCode
empty, so GPUs identifiable only by raw PCI IDs (e.g. NVIDIA H100 SXM5
0x10de/0x2330) are no longer misclassified as generic PCIe devices.
The replay pipeline's "backed by canonical NIC" dedup used to trust a
PCIeDevice's Links.NetworkDeviceFunctions reference at face value and
drop the device, assuming a NetworkAdapters record existed elsewhere.
On BMCs that expose resource IDs with characters (parentheses) that
404 on fetch, that canonical NIC never gets captured, so the device
carrying its actual hardware identity vanished from the export
entirely. hasResolvableLinkedMember now verifies the linked resource
is actually present in the snapshot before treating it as authoritative.
Also normalize PartNumber through normalizeRedfishIdentityField in the
GPU/PCIe parsers so a BMC-supplied literal "null" string doesn't leak
into exports verbatim.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Add PCI Vendor ID / Device ID to hardware.storage[] per the updated Reanimator
ingest contract, mirroring the existing pcie_devices[] fields. Populated for
Redfish-collected NVMe drives (live + TSR replay, via linked PCIeFunctions)
and for Inspur (from asset.json's own PcieInfo[], joined by PcieSlot). Also
fixes canonicalDevicesForExport dropping the fields when converting Storage
into the canonical HardwareDevice list used by the actual export path.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
parseGPUWithSupplementalDocs did not read PCIeInterface from the device
doc, only from function docs. xFusion GPU PCIeCard entries carry link
width/speed in PCIeInterface (LanesInUse/Maxlanes/PCIeType/MaxPCIeType)
so GPU link width was always empty for xFusion servers.
Also apply the xFusion OEM function-level fallback for GPU function docs,
consistent with the NIC and PCIeDevice paths.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
xFusion iBMC exposes PCIe link width in two non-standard ways:
- PCIeInterface uses "Maxlanes" (lowercase 'l') instead of "MaxLanes"
- PCIeFunction docs carry width/speed in Oem.xFusion.LinkWidth ("X8"),
Oem.xFusion.LinkWidthAbility, Oem.xFusion.LinkSpeed, and
Oem.xFusion.LinkSpeedAbility rather than the standard CurrentLinkWidth int
Add redfishEnrichFromOEMxFusionPCIeLink and parseXFusionLinkWidth helpers,
apply them as fallbacks in NIC and PCIeDevice enrichment paths.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Remove power-on and power-off functionality from the Redfish collector;
keep host power-state detection and show a warning in the UI when the
host is powered off before collection starts.
Add a "Пропустить зависшие" (skip hung) button that lets the user abort
stuck Redfish collection phases without losing already-collected data.
Introduces a two-level context model in Collect(): the outer job context
covers the full lifecycle including replay; an inner collectCtx covers
snapshot, prefetch, and plan-B phases only. Closing the skipCh cancels
collectCtx immediately — aborts all in-flight HTTP requests and exits
plan-B loops — then replay runs on whatever rawTree was collected.
Signal path: UI → POST /api/collect/{id}/skip → JobManager.SkipJob()
→ close(skipCh) → goroutine in Collect() → cancelCollect().
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
BMC readiness after power-on (waitForStablePoweredOnHost):
- After initial 1m stabilization, poll BMC inventory readiness before collecting
- Ready if MemorySummary.TotalSystemMemoryGiB > 0 OR PCIeDevices.Members non-empty
- On failure: wait +60s, retry; on second failure: wait +120s, retry; then warn and proceed
- Configurable via LOGPILE_REDFISH_BMC_READY_WAITS (default: 60s,120s)
Empty critical collection plan-B retry (EnableEmptyCriticalCollectionRetry):
- Hardware inventory collections that returned Members=[] are now re-probed in plan-B
- Covers PCIeDevices, NetworkAdapters, Processors, Drives, Storage, EthernetInterfaces
- Enabled by default in generic profile (applies to all vendors)
Ghost NIC dedup fix (enrichNICsFromNetworkInterfaces):
- NetworkInterface entries (e.g. Id=2) that don't match existing NIC slots are now
resolved via Links.NetworkAdapter cross-reference to the real Chassis NIC
- Prevents duplicate ghost entries (slot=2 "Network Device View") from appearing
alongside real NICs (slot="RISER 5 slot 1 (7)") with the same MAC addresses
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Implement the full architectural plan: unified ingest.Service entry point
for archive and Redfish payloads, modular redfishprofile package with
composable profiles (generic, ami-family, msi, supermicro, dell,
hgx-topology), score-based profile matching with fallback expansion mode,
and profile-driven acquisition/analysis plans.
Vendor-specific logic moved out of common executors and into profile hooks.
GPU chassis lookup strategies and known storage recovery collections
(IntelVROC/HA-RAID/MRVL) now live in ResolvedAnalysisPlan, populated by
profiles at analysis time. Replay helpers read from the plan; no hardcoded
path lists remain in generic code.
Also splits redfish_replay.go into domain modules (gpu, storage, inventory,
fru, profiles) and adds full fixture/matcher/directive test coverage
including Dell, AMI, unknown-vendor fallback, and deterministic ordering.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
On Supermicro HGX systems (SYS-A21GE-NBRT) ~35 sub-chassis (GPU, NVSwitch,
PCIeRetimer, ERoT/IRoT, BMC, FPGA) all carry ChassisType=Module/Component/Zone
and expose empty /Drives collections. shouldAdaptiveNVMeProbe returned true for
all of them, triggering 35 × 384 = 13 440 HTTP requests → ~22 min wasted per
collection (more than half of total 35 min collection time).
Fix: chassisTypeCanHaveNVMe returns false for Module, Component, Zone. The
candidate selection loop in collectRawRedfishTree now checks the parent chassis
doc before adding a /Drives path to the probe list. Enclosure (NVMe backplane),
RackMount, and unknown types are unaffected.
Tests:
- TestChassisTypeCanHaveNVMe: table-driven, covers excluded and storage-capable types
- TestNVMePostProbeSkipsNonStorageChassis: topology integration, GPU chassis +
backplane with empty /Drives → exactly 1 candidate selected (backplane only)
Docs:
- ADL-018 in bible-local/10-decisions.md
- Candidate-selection test matrix in bible-local/09-testing.md
- SYS-A21GE-NBRT baseline row in docs/test_server_collection_memory.md
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Three bugs, all related to GPU dedup in the Redfish replay pipeline:
1. collectGPUsFromProcessors (redfish_replay.go): GPU-type Processor entries
(Systems/HGX_Baseboard_0/Processors/GPU_SXM_N) were not deduplicated against
existing PCIeDevice GPUs on Supermicro HGX. The chassis-ID lookup keyed on
processor Id ("GPU_SXM_1") but the chassis is named "HGX_GPU_SXM_1" — lookup
returned nothing, serial stayed empty, UUID was unseen → 8 duplicate GPU rows.
Fix: read SerialNumber directly from the Processor doc first; chassis lookup
is now a fallback override (as it was designed for MSI).
2. looksLikeGPU (redfish.go): NVSwitch PCIe devices (Model="NVSwitch",
Manufacturer="NVIDIA") were classified as GPUs because "nvidia" matched the
GPU hint list. Fix: early return false when Model contains "nvswitch".
3. gpuDocDedupKey (redfish.go): commit 9df29b1 changed the dedup key to prefer
slot|model before path, which collapsed two distinct GPUs with identical model
names in GraphicsControllers into one entry. Fix: only serial and BDF are used
as cross-path stable dedup keys; fall back to Redfish path when neither is
present. This also restores TestReplayCollectGPUs_DedupUsesRedfishPathBeforeHeuristics
which had been broken on main since 9df29b1.
Added tests:
- TestCollectGPUsFromProcessors_SupermicroHGX: Processor GPU dedup when
chassis-ID naming convention does not match processor Id
- TestReplayCollectGPUs_DedupCrossChassisSerial: same GPU via two Chassis
PCIeDevice trees with matching serials → collapsed to one
- TestLooksLikeGPU_NVSwitchExcluded: NVSwitch is not a GPU
Added rule to bible-local/09-testing.md: dedup/filter/classify functions must
cover true-positive, true-negative, and the vendor counter-case axes.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
mchus