Dropping multipart archive for StackGraph backup

cmd/elasticsearch/list_test.go

@@ -64,7 +64,6 @@ minio:
   secretKey: minioadmin
 stackgraph:
   bucket: stackgraph-bucket
-  multipartArchive: true
   restore:
     scaleDownLabelSelector: "app=stackgraph"
     loggingConfigConfigMap: logging-config
@@ -148,7 +147,6 @@ storage:
   secretKey: storageadmin
 stackgraph:
   bucket: stackgraph-bucket
-  multipartArchive: true
   restore:
     scaleDownLabelSelector: "app=stackgraph"
     loggingConfigConfigMap: logging-config

cmd/settings/list.go

@@ -27,7 +27,6 @@ import (
 )
 
 const (
-	isMultiPartArchive            = false
 	expectedListJobPodCount       = 1
 	expectedListJobContainerCount = 1
 	backupFileNameRegex           = `^sts-backup-.*\.sty$`
@@ -183,8 +182,7 @@ func getBackupListFromS3(appCtx *app.Context) ([]BackupFileInfo, error) {
 		return nil, fmt.Errorf("failed to list S3 objects: %w", err)
 	}
 
-	// Filter objects based on whether the archive is split or not
-	filteredObjects := s3client.FilterMultipartBackupObjects(result.Contents, isMultiPartArchive)
+	filteredObjects := s3client.FilterBackupObjects(result.Contents)
 
 	// Filter to only include direct children of the prefix that match the backup filename pattern,
 	// and strip the prefix from the key
@@ -238,7 +236,7 @@ func getBackupListFromLocalBucket(appCtx *app.Context) ([]BackupFileInfo, error)
 		return nil, fmt.Errorf("failed to list objects in local bucket: %w", err)
 	}
 
-	filteredObjects := s3client.FilterMultipartBackupObjects(result.Contents, isMultiPartArchive)
+	filteredObjects := s3client.FilterBackupObjects(result.Contents)
 
 	filteredObjects, err = s3client.FilterByPrefixAndRegex(filteredObjects, "", backupFileNameRegex)
 	if err != nil {

cmd/stackgraph/list.go

@@ -52,7 +52,6 @@ func runList(appCtx *app.Context) error {
 	// List objects in bucket
 	bucket := appCtx.Config.Stackgraph.Bucket
 	prefix := appCtx.Config.Stackgraph.S3Prefix
-	multipartArchive := appCtx.Config.Stackgraph.MultipartArchive
 
 	appCtx.Logger.Infof("Listing Stackgraph backups in bucket '%s'...", bucket)
 
@@ -66,8 +65,7 @@ func runList(appCtx *app.Context) error {
 		return fmt.Errorf("failed to list S3 objects: %w", err)
 	}
 
-	// Filter objects based on whether the archive is split or not
-	filteredObjects := s3client.FilterMultipartBackupObjects(result.Contents, multipartArchive)
+	filteredObjects := s3client.FilterBackupObjects(result.Contents)
 
 	// Filter to only include direct children of the prefix that match the backup filename pattern,
 	// and strip the prefix from the key

cmd/stackgraph/restore.go

@@ -169,7 +169,6 @@ func getLatestBackup(k8sClient *k8s.Client, namespace string, config *config.Con
 	// List objects in bucket
 	bucket := config.Stackgraph.Bucket
 	prefix := config.Stackgraph.S3Prefix
-	multipartArchive := config.Stackgraph.MultipartArchive
 
 	input := &s3.ListObjectsV2Input{
 		Bucket: aws.String(bucket),
@@ -181,8 +180,7 @@ func getLatestBackup(k8sClient *k8s.Client, namespace string, config *config.Con
 		return "", fmt.Errorf("failed to list S3 objects: %w", err)
 	}
 
-	// Filter objects based on whether the archive is split or not
-	filteredObjects := s3client.FilterMultipartBackupObjects(result.Contents, multipartArchive)
+	filteredObjects := s3client.FilterBackupObjects(result.Contents)
 
 	// Filter to only include direct children of the prefix that match the backup filename pattern,
 	// and strip the prefix from the key
@@ -278,7 +276,6 @@ func buildRestoreEnvVars(backupFile string, config *config.Config) []corev1.EnvV
 		{Name: "FORCE_DELETE", Value: purgeStackgraphDataFlag},
 		{Name: "BACKUP_STACKGRAPH_BUCKET_NAME", Value: config.Stackgraph.Bucket},
 		{Name: "BACKUP_STACKGRAPH_S3_PREFIX", Value: config.Stackgraph.S3Prefix},
-		{Name: "BACKUP_STACKGRAPH_MULTIPART_ARCHIVE", Value: strconv.FormatBool(config.Stackgraph.MultipartArchive)},
 		{Name: "MINIO_ENDPOINT", Value: fmt.Sprintf("%s:%d", storageService.Name, storageService.Port)},
 		{Name: "STACKSTATE_BASE_URL", Value: config.GetBaseURL()},
 		{Name: "RECEIVER_BASE_URL", Value: config.GetReceiverBaseURL()},

internal/clients/s3/filter.go

@@ -10,48 +10,22 @@ import (
 	s3types "github.com/aws/aws-sdk-go-v2/service/s3/types"
 )
 
-const (
-	multipartArchiveSuffixLength = 2
-)
-
 // Object represents a simplified S3 object with key metadata
 type Object struct {
 	Key          string
 	LastModified time.Time
 	Size         int64
 }
 
-// FilterMultipartBackupObjects filters S3 objects based on whether the archive is split or not
-// If it is not multipartArchive, it filters out multipart archives (files ending with .digits)
-// Otherwise, it groups multipart archives by base name and sums their sizes
-func FilterMultipartBackupObjects(objects []s3types.Object, multipartArchive bool) []Object {
-	if !multipartArchive {
-		return filterNonMultipart(objects)
-	}
-	return aggregateMultipart(objects)
-}
-
-// filterNonMultipart filters out multipart archives (files ending with .digits)
-func filterNonMultipart(objects []s3types.Object) []Object {
+// FilterBackupObjects filters out backup part files ending with .digits.
+func FilterBackupObjects(objects []s3types.Object) []Object {
 	var filteredObjects []Object
 
 	for _, obj := range objects {
 		key := aws.ToString(obj.Key)
 
-		// Skip if it ends with .digits (multipart archive)
-		if strings.Contains(key, ".") {
-			parts := strings.Split(key, ".")
-			lastPart := parts[len(parts)-1]
-			isDigits := true
-			for _, c := range lastPart {
-				if c < '0' || c > '9' {
-					isDigits = false
-					break
-				}
-			}
-			if isDigits && len(lastPart) > 0 {
-				continue
-			}
+		if hasNumericFileSuffix(key) {
+			continue
 		}
 
 		filteredObjects = append(filteredObjects, Object{
@@ -64,83 +38,25 @@ func filterNonMultipart(objects []s3types.Object) []Object {
 	return filteredObjects
 }
 
-// aggregateMultipart groups multipart archives by base name and sums their sizes
-func aggregateMultipart(objects []s3types.Object) []Object {
-	// Map to group objects by base name
-	archiveMap := make(map[string]*Object)
-
-	for _, obj := range objects {
-		key := aws.ToString(obj.Key)
-
-		// Check if this is a multipart file (ends with .NN where NN are digits)
-		baseName, isMultipart := getBaseName(key)
-		if !isMultipart {
-			// Not a multipart file, include as-is
-			archiveMap[key] = &Object{
-				Key:          key,
-				LastModified: aws.ToTime(obj.LastModified),
-				Size:         aws.ToInt64(obj.Size),
-			}
-			continue
-		}
-
-		// Group multipart files by base name
-		if existing, exists := archiveMap[baseName]; exists {
-			// Add size to existing entry
-			existing.Size += aws.ToInt64(obj.Size)
-			// Keep the most recent LastModified time
-			if aws.ToTime(obj.LastModified).After(existing.LastModified) {
-				existing.LastModified = aws.ToTime(obj.LastModified)
-			}
-		} else {
-			// Create new entry
-			archiveMap[baseName] = &Object{
-				Key:          baseName,
-				LastModified: aws.ToTime(obj.LastModified),
-				Size:         aws.ToInt64(obj.Size),
-			}
-		}
-	}
-
-	// Convert map to slice
-	var filteredObjects []Object
-	for _, obj := range archiveMap {
-		filteredObjects = append(filteredObjects, *obj)
-	}
-
-	return filteredObjects
-}
-
-// getBaseName extracts the base name from a multipart archive filename
-// Returns (baseName, isMultipart)
-// Example: "backup.graph.00" -> ("backup.graph", true)
-//
-//	"backup.graph" -> ("backup.graph", false)
-func getBaseName(key string) (string, bool) {
+func hasNumericFileSuffix(key string) bool {
 	if !strings.Contains(key, ".") {
-		return key, false
+		return false
 	}
 
 	parts := strings.Split(key, ".")
 	lastPart := parts[len(parts)-1]
 
-	// Check if last part is all digits (2 digits for part numbers like .00, .01, etc.)
-	if len(lastPart) == multipartArchiveSuffixLength {
-		isDigits := true
-		for _, c := range lastPart {
-			if c < '0' || c > '9' {
-				isDigits = false
-				break
-			}
-		}
-		if isDigits {
-			// Remove the .NN suffix to get base name
-			baseName := strings.Join(parts[:len(parts)-1], ".")
-			return baseName, true
+	if len(lastPart) == 0 {
+		return false
+	}
+
+	for _, c := range lastPart {
+		if c < '0' || c > '9' {
+			return false
 		}
 	}
 
-	return key, false
+	return true
 }
 
 // FilterByPrefixAndRegex filters objects to only include direct children of the given prefix