feat(inspur): source-PRI timezone timeline, PPIN serial, NVMe fault storage status

Event ingestion now uses the source syslog PRI and an explicit-offset timezone timeline instead of assuming host-local time. CPU PPIN is exported as the source-backed CPU serial, and an active NVMe fault SEL event promotes the matching drive's storage status. See ADL-056, ADL-057. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
2026-08-27 17:05:45 +03:00
parent 07c270cda2
commit e7b8a8badc
13 changed files with 671 additions and 48 deletions
@@ -0,0 +1,71 @@
+package inspur
+
+import (
+	"fmt"
+	"testing"
+	"time"
+
+	"git.mchus.pro/mchus/logpile/internal/models"
+)
+
+func TestApplyStorageStatusFromEventsMarksOnlyFaultyNVMeCritical(t *testing.T) {
+	hw := &models.HardwareConfig{}
+	for i := 1; i <= 7; i++ {
+		hw.Storage = append(hw.Storage, models.Storage{Slot: fmt.Sprintf("OB%02d", i), Type: "NVMe", Present: true})
+	}
+	ts := time.Date(2026, 7, 23, 23, 20, 32, 0, time.UTC)
+	fault := models.Event{
+		ID:          "0DFF0702",
+		Timestamp:   ts,
+		Source:      "BMC",
+		EventType:   "Assert",
+		Severity:    models.SeverityCritical,
+		Description: "NVME BP:Front0 NvmeIndex:6 drive fault: Status_Flags error.",
+	}
+
+	applyStorageStatusFromEvents(hw, []models.Event{fault}, []models.Event{fault}, true)
+
+	for i := range hw.Storage {
+		if hw.Storage[i].Slot == "OB07" {
+			if hw.Storage[i].Status != "Critical" || hw.Storage[i].ErrorDescription == "" {
+				t.Fatalf("expected OB07 critical with details, got %+v", hw.Storage[i])
+			}
+			if len(hw.Storage[i].StatusHistory) != 1 || hw.Storage[i].StatusChangedAt == nil {
+				t.Fatalf("expected one source-backed status transition, got %+v", hw.Storage[i])
+			}
+			continue
+		}
+		if hw.Storage[i].Status != "" {
+			t.Fatalf("unaffected %s must remain unknown/unset, got %q", hw.Storage[i].Slot, hw.Storage[i].Status)
+		}
+	}
+}
+
+func TestApplyStorageStatusFromEventsDeassertClearsCriticalToUnknown(t *testing.T) {
+	hw := &models.HardwareConfig{Storage: []models.Storage{{Slot: "OB07", Type: "NVMe", Present: true}}}
+	asserted := models.Event{
+		ID: "0DFF0702", Timestamp: time.Date(2026, 7, 23, 23, 20, 32, 0, time.UTC),
+		EventType: "Assert", Severity: models.SeverityCritical,
+		Description: "NVME BP:Front0 NvmeIndex:6 drive fault: Status_Flags error.",
+	}
+	deasserted := asserted
+	deasserted.Timestamp = asserted.Timestamp.Add(time.Hour)
+	deasserted.EventType = "Deassert"
+	deasserted.Severity = models.SeverityInfo
+
+	applyStorageStatusFromEvents(hw, []models.Event{asserted, deasserted}, nil, true)
+
+	if hw.Storage[0].Status != "Unknown" || hw.Storage[0].ErrorDescription != "" {
+		t.Fatalf("expected cleared drive fault to become Unknown, got %+v", hw.Storage[0])
+	}
+	if len(hw.Storage[0].StatusHistory) != 2 {
+		t.Fatalf("expected assert/deassert history, got %+v", hw.Storage[0].StatusHistory)
+	}
+}
+
+func TestStorageFaultIndexIgnoresNVMeInventoryChange(t *testing.T) {
+	event := models.Event{Description: "PCIE Device #NVME6 Device ID changed; Vendor ID changed"}
+	if _, ok := storageFaultIndex(event); ok {
+		t.Fatalf("inventory change must not be treated as an active drive fault")
+	}
+}