Initial commit: full Phases 1-6 implementation

Post-repair hardware validation pipeline for Proxmox cluster hosts. Go orchestrator + in-image agent + mkosi live image + bundled dnsmasq PXE + SQLite + HTMX/SSE UI + notify registry + janitor + full docs.
2026-04-17 21:32:10 -04:00
commit 9bb4b09a04
98 changed files with 11960 additions and 0 deletions
@@ -0,0 +1,229 @@
+package store_test
+
+import (
+	"context"
+	"path/filepath"
+	"testing"
+
+	"vetting/internal/db"
+	"vetting/internal/model"
+	"vetting/internal/store"
+)
+
+func newDB(t *testing.T) *store.Runs {
+	t.Helper()
+	path := filepath.Join(t.TempDir(), "vetting.db")
+	conn, err := db.Open(path)
+	if err != nil {
+		t.Fatalf("open db: %v", err)
+	}
+	t.Cleanup(func() { _ = conn.Close() })
+	return &store.Runs{DB: conn}
+}
+
+// seedRun inserts a host + a run and returns (hostID, runID). Every
+// subsequent store test builds on this so run_id foreign keys resolve.
+func seedRun(t *testing.T, runs *store.Runs) (int64, int64) {
+	t.Helper()
+	hosts := &store.Hosts{DB: runs.DB}
+	hostID, err := hosts.Create(context.Background(), model.Host{
+		Name:             "t-host",
+		MAC:              "aa:bb:cc:dd:ee:ff",
+		WoLBroadcastIP:   "10.0.0.255",
+		WoLPort:          9,
+		ExpectedSpecYAML: "memory:\n  total_gib: 16\n",
+	})
+	if err != nil {
+		t.Fatalf("create host: %v", err)
+	}
+	runID, err := runs.Create(context.Background(), hostID, "deadbeef")
+	if err != nil {
+		t.Fatalf("create run: %v", err)
+	}
+	return hostID, runID
+}
+
+func TestArtifactsRoundtrip(t *testing.T) {
+	runs := newDB(t)
+	_, runID := seedRun(t, runs)
+	arts := &store.Artifacts{DB: runs.DB}
+
+	id, err := arts.Create(context.Background(), store.Artifact{
+		RunID:     runID,
+		Kind:      "inventory",
+		Path:      "/var/artifacts/run-1/inventory.json",
+		SHA256:    "abc123",
+		SizeBytes: 42,
+	})
+	if err != nil {
+		t.Fatalf("Create: %v", err)
+	}
+	if id == 0 {
+		t.Fatalf("expected non-zero id")
+	}
+
+	// Hold key on the same run — ListForRun should return both in
+	// insertion order and TileEnricher picks the hold_key row.
+	if _, err := arts.Create(context.Background(), store.Artifact{
+		RunID: runID, Kind: "hold_key", Path: "/var/artifacts/run-1/hold.key", SHA256: "def456", SizeBytes: 400,
+	}); err != nil {
+		t.Fatalf("Create hold_key: %v", err)
+	}
+
+	list, err := arts.ListForRun(context.Background(), runID)
+	if err != nil {
+		t.Fatalf("ListForRun: %v", err)
+	}
+	if len(list) != 2 {
+		t.Fatalf("ListForRun returned %d, want 2", len(list))
+	}
+	if list[0].Kind != "inventory" || list[1].Kind != "hold_key" {
+		t.Fatalf("unexpected order: %+v", list)
+	}
+	if list[1].Path != "/var/artifacts/run-1/hold.key" {
+		t.Fatalf("hold_key path lost: %q", list[1].Path)
+	}
+}
+
+func TestSpecDiffsReplaceForRun(t *testing.T) {
+	runs := newDB(t)
+	_, runID := seedRun(t, runs)
+	sd := &store.SpecDiffs{DB: runs.DB}
+	ctx := context.Background()
+
+	// First write: three diffs.
+	err := sd.ReplaceForRun(ctx, runID, []model.SpecDiff{
+		{RunID: runID, Field: "cpu.model", Expected: "Xeon", Actual: "EPYC", Severity: "critical"},
+		{RunID: runID, Field: "memory.total_gib", Expected: "16", Actual: "8", Severity: "critical"},
+		{RunID: runID, Field: "note", Expected: "", Actual: "dusty", Severity: "info"},
+	})
+	if err != nil {
+		t.Fatalf("ReplaceForRun: %v", err)
+	}
+
+	list, err := sd.ListForRun(ctx, runID)
+	if err != nil {
+		t.Fatalf("ListForRun: %v", err)
+	}
+	if len(list) != 3 {
+		t.Fatalf("got %d rows, want 3", len(list))
+	}
+
+	// Second write replaces, doesn't append — otherwise a re-run would
+	// double-count spec diffs and the tile badge would grow without bound.
+	err = sd.ReplaceForRun(ctx, runID, []model.SpecDiff{
+		{RunID: runID, Field: "cpu.model", Expected: "Xeon", Actual: "Xeon Gold", Severity: "info"},
+	})
+	if err != nil {
+		t.Fatalf("second ReplaceForRun: %v", err)
+	}
+	list, err = sd.ListForRun(ctx, runID)
+	if err != nil {
+		t.Fatalf("ListForRun after replace: %v", err)
+	}
+	if len(list) != 1 {
+		t.Fatalf("expected 1 row after replace, got %d", len(list))
+	}
+	if list[0].Severity != "info" {
+		t.Fatalf("expected severity info, got %q", list[0].Severity)
+	}
+}
+
+func TestMeasurementsBatchAndList(t *testing.T) {
+	runs := newDB(t)
+	_, runID := seedRun(t, runs)
+	meas := &store.Measurements{DB: runs.DB}
+	ctx := context.Background()
+
+	err := meas.CreateBatch(ctx, []model.Measurement{
+		{RunID: runID, Kind: "thermal", Key: "cpu", Value: 52.5, Unit: "C"},
+		{RunID: runID, Kind: "iperf", Key: "throughput_mbps", Value: 940.1, Unit: "Mbps"},
+		{RunID: runID, Kind: "psu", Key: "in0", Value: 12.04, Unit: "V"},
+	})
+	if err != nil {
+		t.Fatalf("CreateBatch: %v", err)
+	}
+
+	// Zero-length batch must be a no-op, not an error.
+	if err := meas.CreateBatch(ctx, nil); err != nil {
+		t.Fatalf("empty CreateBatch: %v", err)
+	}
+
+	rows, err := meas.ListForRun(ctx, runID)
+	if err != nil {
+		t.Fatalf("ListForRun: %v", err)
+	}
+	if len(rows) != 3 {
+		t.Fatalf("got %d rows, want 3", len(rows))
+	}
+	foundIperf := false
+	for _, r := range rows {
+		if r.Kind == "iperf" && r.Key == "throughput_mbps" && r.Value > 900 {
+			foundIperf = true
+		}
+	}
+	if !foundIperf {
+		t.Fatalf("iperf row missing or wrong value: %+v", rows)
+	}
+}
+
+func TestRunsOverrideFlagsAndClearFailedStage(t *testing.T) {
+	runs := newDB(t)
+	_, runID := seedRun(t, runs)
+	ctx := context.Background()
+
+	if err := runs.SetFailedStage(ctx, runID, "Storage"); err != nil {
+		t.Fatalf("SetFailedStage: %v", err)
+	}
+	if err := runs.SetOverrideFlags(ctx, runID, `{"wipe":true}`); err != nil {
+		t.Fatalf("SetOverrideFlags: %v", err)
+	}
+	run, err := runs.Get(ctx, runID)
+	if err != nil {
+		t.Fatalf("Get: %v", err)
+	}
+	if run.OverrideFlagsJSON != `{"wipe":true}` {
+		t.Fatalf("OverrideFlagsJSON = %q, want {\"wipe\":true}", run.OverrideFlagsJSON)
+	}
+	if run.FailedStage != "Storage" {
+		t.Fatalf("FailedStage = %q, want Storage", run.FailedStage)
+	}
+	if err := runs.ClearFailedStage(ctx, runID); err != nil {
+		t.Fatalf("ClearFailedStage: %v", err)
+	}
+	run, err = runs.Get(ctx, runID)
+	if err != nil {
+		t.Fatalf("Get after clear: %v", err)
+	}
+	if run.FailedStage != "" {
+		t.Fatalf("FailedStage not cleared: %q", run.FailedStage)
+	}
+	// override_flags_json should persist across ClearFailedStage so the
+	// agent can still read it on its next heartbeat.
+	if run.OverrideFlagsJSON != `{"wipe":true}` {
+		t.Fatalf("OverrideFlagsJSON lost after ClearFailedStage: %q", run.OverrideFlagsJSON)
+	}
+}
+
+func TestRunsHoldAndFailedStage(t *testing.T) {
+	runs := newDB(t)
+	_, runID := seedRun(t, runs)
+	ctx := context.Background()
+
+	if err := runs.SetHoldIP(ctx, runID, "10.0.0.42"); err != nil {
+		t.Fatalf("SetHoldIP: %v", err)
+	}
+	if err := runs.SetFailedStage(ctx, runID, "SpecValidate"); err != nil {
+		t.Fatalf("SetFailedStage: %v", err)
+	}
+	run, err := runs.Get(ctx, runID)
+	if err != nil {
+		t.Fatalf("Get: %v", err)
+	}
+	if run.HoldIP != "10.0.0.42" {
+		t.Fatalf("HoldIP = %q, want 10.0.0.42", run.HoldIP)
+	}
+	if run.FailedStage != "SpecValidate" {
+		t.Fatalf("FailedStage = %q, want SpecValidate", run.FailedStage)
+	}
+}