fix(exporter): drop dead bmc_event_summary field, fix status loss for inspur CPU/GPU
bmc_event_summary was a derived Assert/Deassert summary added in 4409594
alongside real fixes for GPU fault handling. It's not part of the Reanimator
hardware-ingest contract (event_logs is the only accepted log channel) and
was silently dropped on import — pure dead weight, removed.
Three related status bugs surfaced while auditing converted exports against
the contract, all specific to Inspur/onekeylog dumps:
- CPU status from RESTful CPU info was parsed but never assigned to
models.CPU, and was skipped entirely whenever asset.json already supplied
a CPU list (its own inventory has no status field) — CPUs always exported
as Unknown even when the source reported OK.
- PCIe device status (RESTful "status": 1) was parsed but never mapped onto
models.PCIeDevice, so RESTful-only devices always lost status.
- For GPUs specifically, asset.go emits two device records per physical GPU
(a generic pcie_devices entry enriched with real status, and a separate
gpus entry with a resolved model name but no status). dedupePCIe picks a
single winner by quality score, and a better model name outweighed having
a real status — the winner kept "Unknown" even when a losing duplicate had
the real value. dedupePCIe now backfills status onto the winner from a
losing duplicate when the winner's is Unknown.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
This commit is contained in:
Mikhail Chusavitin
co-authored by
Claude Sonnet 5
Claude Sonnet 5
parent
0867123a91
commit
42cfa3aa94
@@ -43,17 +43,16 @@ func ConvertToReanimator(result *models.AnalysisResult) (*ReanimatorExport, erro
|
||||
TargetHost: targetHost,
|
||||
CollectedAt: collectedAt,
|
||||
Hardware: ReanimatorHardware{
|
||||
Board: convertBoard(result.Hardware.BoardInfo),
|
||||
Firmware: dedupeFirmware(convertFirmware(result.Hardware.Firmware)),
|
||||
CPUs: dedupeCPUs(convertCPUsFromDevices(devices, collectedAt, result.Hardware.BoardInfo.SerialNumber, buildCPUMicrocodeBySocket(result.Hardware.Firmware))),
|
||||
Memory: dedupeMemory(convertMemoryFromDevices(devices, collectedAt)),
|
||||
Storage: dedupeStorage(convertStorageFromDevices(devices, collectedAt)),
|
||||
PCIeDevices: dedupePCIe(convertPCIeFromDevices(devices, collectedAt)),
|
||||
PowerSupplies: dedupePSUs(convertPSUsFromDevices(devices, collectedAt)),
|
||||
Sensors: convertSensors(result.Sensors),
|
||||
BMCEventSummary: buildBMCEventSummary(result.Events, collectedAt),
|
||||
EventLogs: convertEventLogs(result.Events, collectedAt),
|
||||
Licenses: dedupeLicenses(convertLicenses(result.Hardware.Licenses, collectedAt)),
|
||||
Board: convertBoard(result.Hardware.BoardInfo),
|
||||
Firmware: dedupeFirmware(convertFirmware(result.Hardware.Firmware)),
|
||||
CPUs: dedupeCPUs(convertCPUsFromDevices(devices, collectedAt, result.Hardware.BoardInfo.SerialNumber, buildCPUMicrocodeBySocket(result.Hardware.Firmware))),
|
||||
Memory: dedupeMemory(convertMemoryFromDevices(devices, collectedAt)),
|
||||
Storage: dedupeStorage(convertStorageFromDevices(devices, collectedAt)),
|
||||
PCIeDevices: dedupePCIe(convertPCIeFromDevices(devices, collectedAt)),
|
||||
PowerSupplies: dedupePSUs(convertPSUsFromDevices(devices, collectedAt)),
|
||||
Sensors: convertSensors(result.Sensors),
|
||||
EventLogs: convertEventLogs(result.Events, collectedAt),
|
||||
Licenses: dedupeLicenses(convertLicenses(result.Hardware.Licenses, collectedAt)),
|
||||
},
|
||||
}
|
||||
|
||||
@@ -2195,9 +2194,16 @@ func dedupePCIe(items []ReanimatorPCIe) []ReanimatorPCIe {
|
||||
order = append(order, key)
|
||||
continue
|
||||
}
|
||||
winner, loser := existing, curr
|
||||
if curr.score > existing.score {
|
||||
byKey[key] = curr
|
||||
winner, loser = curr, existing
|
||||
}
|
||||
// The winner is picked by overall quality score, but a losing duplicate
|
||||
// (e.g. a generic pcie_devices entry enriched with real status vs. a
|
||||
// gpus entry with a resolved model name but no status) may still carry
|
||||
// status data the winner lacks. Backfill it rather than dropping it.
|
||||
mergePCIeStatusInto(&winner.item, loser.item)
|
||||
byKey[key] = winner
|
||||
}
|
||||
result := make([]ReanimatorPCIe, 0, len(byKey))
|
||||
for _, key := range order {
|
||||
@@ -2206,6 +2212,23 @@ func dedupePCIe(items []ReanimatorPCIe) []ReanimatorPCIe {
|
||||
return result
|
||||
}
|
||||
|
||||
// mergePCIeStatusInto backfills dst's status fields from src when dst has none.
|
||||
func mergePCIeStatusInto(dst *ReanimatorPCIe, src ReanimatorPCIe) {
|
||||
// Status is already normalized to "Unknown" (never "") by the time items
|
||||
// reach dedup, so emptiness isn't the right check here.
|
||||
if !strings.EqualFold(strings.TrimSpace(dst.Status), "Unknown") {
|
||||
return
|
||||
}
|
||||
if srcStatus := strings.TrimSpace(src.Status); srcStatus == "" || strings.EqualFold(srcStatus, "Unknown") {
|
||||
return
|
||||
}
|
||||
dst.Status = src.Status
|
||||
dst.StatusCheckedAt = src.StatusCheckedAt
|
||||
dst.StatusChangedAt = src.StatusChangedAt
|
||||
dst.StatusHistory = src.StatusHistory
|
||||
dst.ErrorDescription = src.ErrorDescription
|
||||
}
|
||||
|
||||
func pcieDedupKey(item ReanimatorPCIe) string {
|
||||
slot := strings.ToLower(strings.TrimSpace(item.Slot))
|
||||
serial := strings.ToLower(strings.TrimSpace(item.SerialNumber))
|
||||
@@ -2686,75 +2709,3 @@ func inferTargetHost(targetHost, filename string) string {
|
||||
return ""
|
||||
}
|
||||
|
||||
// buildBMCEventSummary produces a summary table of Critical/Warning BMC events
|
||||
// with their resolution status derived from Assert/Deassert pairs.
|
||||
func buildBMCEventSummary(events []models.Event, collectedAt string) []ReanimatorBMCEventRow {
|
||||
type assertKey struct {
|
||||
id string
|
||||
desc string
|
||||
}
|
||||
type eventPair struct {
|
||||
assertEvent *models.Event
|
||||
deassertEvent *models.Event
|
||||
}
|
||||
|
||||
pairs := make(map[assertKey]*eventPair)
|
||||
order := make([]assertKey, 0)
|
||||
|
||||
for i := range events {
|
||||
e := &events[i]
|
||||
if e.Severity != models.SeverityCritical && e.Severity != models.SeverityWarning {
|
||||
continue
|
||||
}
|
||||
key := assertKey{id: e.ID, desc: e.Description}
|
||||
p, exists := pairs[key]
|
||||
if !exists {
|
||||
p = &eventPair{}
|
||||
pairs[key] = p
|
||||
order = append(order, key)
|
||||
}
|
||||
switch strings.ToLower(e.EventType) {
|
||||
case "deassert":
|
||||
if p.deassertEvent == nil || e.Timestamp.After(p.deassertEvent.Timestamp) {
|
||||
p.deassertEvent = e
|
||||
}
|
||||
default:
|
||||
if p.assertEvent == nil || e.Timestamp.Before(p.assertEvent.Timestamp) {
|
||||
p.assertEvent = e
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
rows := make([]ReanimatorBMCEventRow, 0, len(order))
|
||||
for _, key := range order {
|
||||
p := pairs[key]
|
||||
ref := p.assertEvent
|
||||
if ref == nil {
|
||||
ref = p.deassertEvent
|
||||
}
|
||||
if ref == nil {
|
||||
continue
|
||||
}
|
||||
|
||||
status := "Active"
|
||||
resolvedAt := ""
|
||||
if p.deassertEvent != nil {
|
||||
status = "Resolved"
|
||||
resolvedAt = formatEventLogTime(p.deassertEvent.Timestamp, collectedAt)
|
||||
}
|
||||
|
||||
rows = append(rows, ReanimatorBMCEventRow{
|
||||
Severity: normalizeEventLogSeverity(ref.Severity),
|
||||
Component: strings.ToUpper(strings.TrimSpace(ref.SensorType)),
|
||||
MessageID: strings.TrimSpace(ref.ID),
|
||||
Timestamp: formatEventLogTime(ref.Timestamp, collectedAt),
|
||||
Description: strings.TrimSpace(ref.Description),
|
||||
Status: status,
|
||||
ResolvedAt: resolvedAt,
|
||||
})
|
||||
}
|
||||
if len(rows) == 0 {
|
||||
return nil
|
||||
}
|
||||
return rows
|
||||
}
|
||||
|
||||
@@ -12,29 +12,17 @@ type ReanimatorExport struct {
|
||||
|
||||
// ReanimatorHardware contains all hardware components
|
||||
type ReanimatorHardware struct {
|
||||
Board ReanimatorBoard `json:"board"`
|
||||
Firmware []ReanimatorFirmware `json:"firmware,omitempty"`
|
||||
CPUs []ReanimatorCPU `json:"cpus,omitempty"`
|
||||
Memory []ReanimatorMemory `json:"memory,omitempty"`
|
||||
Storage []ReanimatorStorage `json:"storage,omitempty"`
|
||||
PCIeDevices []ReanimatorPCIe `json:"pcie_devices,omitempty"`
|
||||
PowerSupplies []ReanimatorPSU `json:"power_supplies,omitempty"`
|
||||
Sensors *ReanimatorSensors `json:"sensors,omitempty"`
|
||||
BMCEventSummary []ReanimatorBMCEventRow `json:"bmc_event_summary,omitempty"`
|
||||
EventLogs []ReanimatorEventLog `json:"event_logs,omitempty"`
|
||||
PlatformConfig map[string]any `json:"platform_config,omitempty"`
|
||||
Licenses []ReanimatorLicense `json:"licenses,omitempty"`
|
||||
}
|
||||
|
||||
// ReanimatorBMCEventRow is one row in the BMC critical/warning event summary table.
|
||||
type ReanimatorBMCEventRow struct {
|
||||
Severity string `json:"severity"`
|
||||
Component string `json:"component"`
|
||||
MessageID string `json:"message_id"`
|
||||
Timestamp string `json:"timestamp"`
|
||||
Description string `json:"description"`
|
||||
Status string `json:"status"`
|
||||
ResolvedAt string `json:"resolved_at,omitempty"`
|
||||
Board ReanimatorBoard `json:"board"`
|
||||
Firmware []ReanimatorFirmware `json:"firmware,omitempty"`
|
||||
CPUs []ReanimatorCPU `json:"cpus,omitempty"`
|
||||
Memory []ReanimatorMemory `json:"memory,omitempty"`
|
||||
Storage []ReanimatorStorage `json:"storage,omitempty"`
|
||||
PCIeDevices []ReanimatorPCIe `json:"pcie_devices,omitempty"`
|
||||
PowerSupplies []ReanimatorPSU `json:"power_supplies,omitempty"`
|
||||
Sensors *ReanimatorSensors `json:"sensors,omitempty"`
|
||||
EventLogs []ReanimatorEventLog `json:"event_logs,omitempty"`
|
||||
PlatformConfig map[string]any `json:"platform_config,omitempty"`
|
||||
Licenses []ReanimatorLicense `json:"licenses,omitempty"`
|
||||
}
|
||||
|
||||
// ReanimatorBoard represents motherboard/server information
|
||||
|
||||
Reference in New Issue
Block a user