fix(inspur): correctly handle PCIe Assert/Deassert GPU fault events

Three related fixes for IDL event processing:

1. idl.go: include EventType in dedup key so Deassert events are no
   longer silently dropped as duplicates of their Assert counterparts.

2. gpu_status.go: treat Deassert events as clearing all GPU faults —
   previously the code re-applied the same faulty GPU set from the
   description, leaving GPUs stuck in Critical even after alarm cleared.

3. reanimator_models/converter: add bmc_event_summary section to the
   Reanimator export — a deduplicated Critical/Warning event table with
   Active/Resolved status derived from Assert/Deassert pairs.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

internal/exporter/reanimator_converter.go

@@ -49,9 +49,10 @@ func ConvertToReanimator(result *models.AnalysisResult) (*ReanimatorExport, erro
 			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),
+			PowerSupplies:   dedupePSUs(convertPSUsFromDevices(devices, collectedAt)),
+			Sensors:         convertSensors(result.Sensors),
+			BMCEventSummary: buildBMCEventSummary(result.Events, collectedAt),
+			EventLogs:       convertEventLogs(result.Events, collectedAt),
 		},
 	}
 
@@ -2457,3 +2458,76 @@ 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
+}

internal/exporter/reanimator_models.go

@@ -12,16 +12,28 @@ 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"`
-	EventLogs      []ReanimatorEventLog `json:"event_logs,omitempty"`
-	PlatformConfig map[string]any       `json:"platform_config,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"`
+	BMCEventSummary  []ReanimatorBMCEventRow    `json:"bmc_event_summary,omitempty"`
+	EventLogs        []ReanimatorEventLog       `json:"event_logs,omitempty"`
+	PlatformConfig   map[string]any             `json:"platform_config,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"`
 }
 
 // ReanimatorBoard represents motherboard/server information

internal/parser/vendors/inspur/gpu_status.go

@@ -56,10 +56,12 @@ func applyGPUStatusFromEvents(hw *models.HardwareConfig, events []models.Event)
 	}
 
 	for _, e := range relevantEvents {
+		// Deassert means the alarm was cleared: all GPUs return to OK.
+		isDeassert := strings.EqualFold(strings.TrimSpace(e.EventType), "Deassert")
 		faultySet := extractFaultyGPUSet(e.Description)
 		for idx, gpu := range gpuByIndex {
 			newStatus := "OK"
-			if faultySet[idx] {
+			if !isDeassert && faultySet[idx] {
 				newStatus = "Critical"
 				lastCriticalDetails[idx] = strings.TrimSpace(e.Description)
 			}

internal/parser/vendors/inspur/hgx_gpu_status_test.go

@@ -155,6 +155,40 @@ func TestApplyGPUStatusFromEvents_UsesLatestEventAsCurrentStatusAndKeepsHistory(
 	}
 }
 
+func TestApplyGPUStatusFromEvents_DeassertClearsAllGPUs(t *testing.T) {
+	hw := &models.HardwareConfig{
+		GPUs: []models.GPU{
+			{Slot: "#GPU1"},
+			{Slot: "#GPU3"},
+			{Slot: "#GPU5"},
+			{Slot: "#GPU6"},
+		},
+	}
+
+	events := []models.Event{
+		{
+			ID:          "17FFB002",
+			EventType:   "Assert",
+			Timestamp:   time.Date(2026, 5, 27, 13, 6, 56, 0, time.FixedZone("UTC+8", 8*3600)),
+			Description: "PCIe Present mismatch BIOS Scan, BIOS miss F_GPU1 F_GPU3 F_GPU5 F_GPU6",
+		},
+		{
+			ID:          "17FFB002",
+			EventType:   "Deassert",
+			Timestamp:   time.Date(2026, 5, 27, 13, 15, 56, 0, time.FixedZone("UTC+8", 8*3600)),
+			Description: "PCIe Present mismatch BIOS Scan, BIOS miss F_GPU1 F_GPU3 F_GPU5 F_GPU6",
+		},
+	}
+
+	applyGPUStatusFromEvents(hw, events)
+
+	for _, gpu := range hw.GPUs {
+		if gpu.Status != "OK" {
+			t.Fatalf("expected %s to recover to OK after Deassert, got %q", gpu.Slot, gpu.Status)
+		}
+	}
+}
+
 func TestParseIDLLog_ParsesStructuredJSONLine(t *testing.T) {
 	content := []byte(`{ "MESSAGE": "|2026-01-12T23:05:18+08:00|PCIE|Assert|Critical|17FFB002|PCIe Present mismatch BIOS miss F_GPU6 - Assert|" }`)
 

internal/parser/vendors/inspur/idl.go

@@ -48,7 +48,7 @@ func ParseIDLLog(content []byte) []models.Event {
 		description = cleanDescription(description)
 
 		// Create unique key for deduplication
-		eventKey := eventID + "|" + description
+		eventKey := eventID + "|" + eventType + "|" + description
 		if seenEvents[eventKey] {
 			continue
 		}