Report blob counts from snapshot verify --deep --json (closes #314)

Deep verify set verified only on success and never set missing, so a
failed --deep --json run reported zero for both. It now counts each blob
whose contents pass as it goes, and the existence check runs through
every blob, counting the missing and wrong-size ones as shallow verify
does, instead of stopping at the first; it fails if it found any.

Deep verify still stops at the first damaged blob and reads no blob
contents while any blob is missing or the wrong size, so verified is 0
there where shallow mode counts the present blobs.

Each missing or wrong-size blob is logged as a warning with its hash;
the error now gives counts instead of naming one blob.

Model: opus-5-5

internal/vaultik/deep_verify_test.go

@@ -1,15 +1,20 @@
 package vaultik_test
 
 import (
+	"bytes"
 	"context"
+	"crypto/rand"
 	"io"
 	"os"
 	"path/filepath"
+	"slices"
 	"testing"
 
 	"github.com/spf13/afero"
 	"github.com/stretchr/testify/require"
+	"sneak.berlin/go/vaultik/internal/config"
 	"sneak.berlin/go/vaultik/internal/log"
+	"sneak.berlin/go/vaultik/internal/storage"
 	"sneak.berlin/go/vaultik/internal/ui"
 	"sneak.berlin/go/vaultik/internal/vaultik"
 )
@@ -77,6 +82,116 @@ func TestDeepVerifyAcceptsHealthyAndRejectsCorruptBlob(t *testing.T) {
 		"deep verify should fail on a corrupted blob")
 }
 
+// TestDeepVerifyJSONCountsBlobs checks the verified and missing counts
+// that deep verify reports under --json: on a healthy snapshot, with a
+// damaged blob, and with missing blobs. Deep verify stops at the first
+// damaged blob, so the blobs read before it are the ones verified; when
+// blobs are missing it reads no blob contents, but counts every missing
+// blob.
+func TestDeepVerifyJSONCountsBlobs(t *testing.T) {
+	log.Initialize(log.Config{})
+	t.Parallel()
+
+	fs := afero.NewOsFs()
+	tempDir := t.TempDir()
+
+	dataDir := filepath.Join(tempDir, "source")
+	storeDir := filepath.Join(tempDir, "remote")
+	dbPath := filepath.Join(tempDir, "index.sqlite")
+
+	chunkSize := int64(32 * 1024)
+	maxBlobSize := int64(128 * 1024)
+
+	// Random data does not compress or deduplicate, so it fills several
+	// blobs.
+	data := make([]byte, maxBlobSize*4)
+	_, err := rand.Read(data)
+	require.NoError(t, err)
+	require.NoError(t, fs.MkdirAll(dataDir, 0o755))
+	require.NoError(t, afero.WriteFile(fs,
+		filepath.Join(dataDir, "data.bin"), data, 0o644))
+
+	ctx := context.Background()
+
+	cfg, storer, snapshotID := runFileStorageBackup(
+		ctx, t, fs, dataDir, storeDir, dbPath, chunkSize, maxBlobSize)
+
+	healthy, err := deepVerifyJSON(ctx, t, cfg, storer, snapshotID)
+	require.NoError(t, err)
+	require.Equal(t, "ok", healthy.Status)
+	require.GreaterOrEqual(t, healthy.BlobCount, 3)
+	require.Equal(t, healthy.BlobCount, healthy.Verified)
+	require.Zero(t, healthy.Missing)
+
+	// Deep verify reads blobs in hash order, which is the sorted order of
+	// their storage keys.
+	keys, err := storer.List(ctx, "blobs/")
+	require.NoError(t, err)
+	slices.Sort(keys)
+	require.Len(t, keys, healthy.BlobCount)
+
+	// Damage the last blob read, keeping its length, so every blob before
+	// it is read and passes first.
+	lastPath := filepath.Join(storeDir, keys[len(keys)-1])
+	blob, err := afero.ReadFile(fs, lastPath)
+	require.NoError(t, err)
+
+	blob[len(blob)/2] ^= 0xff
+	require.NoError(t, afero.WriteFile(fs, lastPath, blob, 0o600))
+
+	got, err := deepVerifyJSON(ctx, t, cfg, storer, snapshotID)
+	require.Error(t, err)
+	require.Equal(t, "failed", got.Status)
+	require.Equal(t, healthy.BlobCount-1, got.Verified)
+	require.Zero(t, got.Missing)
+
+	// Delete the first blob and the damaged last one: both must be
+	// counted as missing, not only the first one found.
+	var deletedSize int64
+
+	for _, key := range []string{keys[0], keys[len(keys)-1]} {
+		info, statErr := storer.Stat(ctx, key)
+		require.NoError(t, statErr)
+
+		deletedSize += info.Size
+
+		require.NoError(t, storer.Delete(ctx, key))
+	}
+
+	got, err = deepVerifyJSON(ctx, t, cfg, storer, snapshotID)
+	require.Error(t, err)
+	require.Equal(t, "failed", got.Status)
+	require.Equal(t, 2, got.Missing)
+	require.Equal(t, deletedSize, got.MissingSize)
+	require.Zero(t, got.Verified)
+}
+
+// deepVerifyJSON runs deep verify with --json on snapshotID and returns
+// the decoded result and the error RunDeepVerify returned.
+func deepVerifyJSON(
+	ctx context.Context, t *testing.T,
+	cfg *config.Config, storer storage.Storer, snapshotID string,
+) (vaultik.VerifyResult, error) {
+	t.Helper()
+
+	var out bytes.Buffer
+
+	v := &vaultik.Vaultik{
+		Config:  cfg,
+		Storage: storer,
+		Fs:      afero.NewOsFs(),
+		Stdout:  &out,
+		Stderr:  io.Discard,
+		UI:      ui.NewWithColor(io.Discard, false),
+	}
+	v.SetContext(ctx)
+
+	err := v.RunDeepVerify(snapshotID,
+		&vaultik.VerifyOptions{Deep: true, JSON: true})
+
+	return decodeVerifyResult(t, out.Bytes()), err
+}
+
 // corruptOneBlob flips a middle byte of the first blob file found under
 // blobsDir, leaving the file length unchanged.
 func corruptOneBlob(t *testing.T, fs afero.Fs, blobsDir string) {

internal/vaultik/verify.go

@@ -32,7 +32,9 @@ var (
 	errManifestExtraBlob   = errors.New("manifest contains blob not in database")
 	errManifestMissingBlob = errors.New(
 		"manifest omits blob present in database")
-	errBlobSizeMismatch = errors.New("blob size mismatch")
+	errBlobSizeMismatch  = errors.New("blob size mismatch")
+	errBlobsNotInStorage = errors.New(
+		"blobs missing from storage or stored with the wrong size")
 )
 
 // verifyStatusFailed is the JSON status value for a failed verification.
@@ -135,7 +137,6 @@ func (v *Vaultik) RunDeepVerify(snapshotID string, opts *VerifyOptions) error {
 	}
 
 	result.Status = "ok"
-	result.Verified = len(dbBlobs)
 
 	if opts.JSON {
 		return v.outputVerifyJSON(result)
@@ -290,7 +291,7 @@ func (v *Vaultik) runVerificationSteps(
 		v.stdoutf("Checking blob existence in remote storage...\n")
 	}
 
-	err = v.verifyBlobExistenceFromDB(dbBlobs)
+	err = v.verifyBlobExistenceFromDB(dbBlobs, result)
 	if err != nil {
 		return v.deepVerifyFailure(result, opts, err.Error(), err)
 	}
@@ -301,7 +302,8 @@ func (v *Vaultik) runVerificationSteps(
 			len(dbBlobs), ubytes(totalSize))
 	}
 
-	err = v.performDeepVerificationFromDB(dbBlobs, tdb.db.Conn(), opts, identities)
+	err = v.performDeepVerificationFromDB(
+		dbBlobs, tdb.db.Conn(), opts, result, identities)
 	if err != nil {
 		return v.deepVerifyFailure(result, opts, err.Error(), err)
 	}
@@ -653,8 +655,13 @@ func (v *Vaultik) verifyManifestAgainstDatabase(
 	return nil
 }
 
-// verifyBlobExistenceFromDB checks that all blobs from database exist in S3
+// verifyBlobExistenceFromDB checks that every blob from the database is
-func (v *Vaultik) verifyBlobExistenceFromDB(blobs []snapshot.BlobInfo) error {
+// in storage with the size the database records. It checks them all,
+// counting the missing and wrong-size blobs in result, and fails if it
+// found any.
+func (v *Vaultik) verifyBlobExistenceFromDB(
+	blobs []snapshot.BlobInfo, result *VerifyResult,
+) error {
 	log.Info("Verifying blob existence in S3", "blob_count", len(blobs))
 
 	for i, blob := range blobs {
@@ -667,17 +674,21 @@ func (v *Vaultik) verifyBlobExistenceFromDB(blobs []snapshot.BlobInfo) error {
 		// Construct blob path
 		blobPath := fmt.Sprintf("blobs/%s/%s/%s", blob.Hash[:2], blob.Hash[2:4], blob.Hash)
 
-		// Check blob exists
 		stat, err := v.Storage.Stat(v.ctx, blobPath)
-		if err != nil {
+		switch {
-			return fmt.Errorf("blob %s missing from storage: %w", blob.Hash, err)
+		case err != nil:
-		}
+			log.Warn("Blob missing from storage",
+				"hash", blob.Hash, "error", err)
 
-		// Verify size matches
+			result.Missing++
-		if stat.Size != blob.CompressedSize {
+			result.MissingSize += blob.CompressedSize
-			return fmt.Errorf(
+		case stat.Size != blob.CompressedSize:
-				"%w: blob %s: S3 has %d bytes, database has %d bytes",
+			log.Warn("Blob has the wrong size in storage",
-				errBlobSizeMismatch, blob.Hash, stat.Size, blob.CompressedSize)
+				"hash", blob.Hash,
+				"storage_size", stat.Size,
+				"database_size", blob.CompressedSize)
+
+			result.Mismatched++
 		}
 
 		// Progress update every 100 blobs
@@ -691,16 +702,22 @@ func (v *Vaultik) verifyBlobExistenceFromDB(blobs []snapshot.BlobInfo) error {
 		}
 	}
 
+	if result.Missing > 0 || result.Mismatched > 0 {
+		return fmt.Errorf("%w: %d missing, %d with the wrong size",
+			errBlobsNotInStorage, result.Missing, result.Mismatched)
+	}
+
 	log.Info("✓ All blobs exist in storage")
 
 	return nil
 }
 
 // performDeepVerificationFromDB downloads and verifies the content of
-// each blob using the database as source.
+// each blob using the database as source, stopping at the first blob that
+// fails. Each blob that passes is counted in result.Verified.
 func (v *Vaultik) performDeepVerificationFromDB(
 	blobs []snapshot.BlobInfo, db *sql.DB, opts *VerifyOptions,
-	identities []age.Identity,
+	result *VerifyResult, identities []age.Identity,
 ) error {
 	// Calculate total bytes for ETA
 	var totalBytesExpected int64
@@ -723,6 +740,7 @@ func (v *Vaultik) performDeepVerificationFromDB(
 			return fmt.Errorf("blob %s verification failed: %w", blobInfo.Hash, err)
 		}
 
+		result.Verified++
 		bytesProcessed += blobInfo.CompressedSize
 		elapsed := time.Since(startTime)
 		remaining := len(blobs) - (i + 1)