Fixing gsutil upload/download. Divide file list into chunks.

Author: Alexandr Bogomyakov

medusa/backup.py

@@ -31,7 +31,8 @@
 from medusa.index import add_backup_start_to_index, add_backup_finish_to_index, set_latest_backup_in_index
 from medusa.monitoring import Monitoring
 from medusa.storage.s3_storage import is_aws_s3
-from medusa.storage import Storage, format_bytes_str, ManifestObject
+from medusa.storage.google_storage import GSUTIL_MAX_FILES_PER_CHUNK
+from medusa.storage import Storage, format_bytes_str, ManifestObject, divide_chunks
 
 
 class NodeBackupCache(object):
@@ -322,7 +323,11 @@ def backup_snapshots(storage, manifest, node_backup, node_backup_cache, snapshot
 
             manifest_objects = list()
             if len(needs_backup) > 0:
-                manifest_objects = storage.storage_driver.upload_blobs(needs_backup, dst_path)
+                # If there is a plenty of files to upload it should be
+                # splitted to batches due to 'gsutil cp' which
+                # can't handle too much source files via STDIN.
+                for src_batch in divide_chunks(needs_backup, GSUTIL_MAX_FILES_PER_CHUNK):
+                    manifest_objects += storage.storage_driver.upload_blobs(src_batch, dst_path)
 
             # Reintroducing already backed up objects in the manifest in differential
             for obj in already_backed_up:

medusa/download.py

@@ -18,7 +18,8 @@
 import pathlib
 import sys
 
-from medusa.storage import Storage
+from medusa.storage import Storage, divide_chunks
+from medusa.storage.google_storage import GSUTIL_MAX_FILES_PER_CHUNK
 
 
 def download_data(storageconfig, backup, fqtns_to_restore, destination):
@@ -43,8 +44,9 @@ def download_data(storageconfig, backup, fqtns_to_restore, destination):
             subfolder.mkdir(parents=False)
 
         if len(srcs) > 0 and (len(fqtns_to_restore) == 0 or fqtn in fqtns_to_restore):
-            logging.info('Downloading backup data')
-            storage.storage_driver.download_blobs(srcs, dst)
+            logging.debug('Downloading  %s files to %s', len(srcs), dst)
+            for src_batch in divide_chunks(srcs, GSUTIL_MAX_FILES_PER_CHUNK):
+                storage.storage_driver.download_blobs(src_batch, dst)
         elif len(srcs) == 0 and (len(fqtns_to_restore) == 0 or fqtn in fqtns_to_restore):
             logging.debug('There is nothing to download for {}'.format(fqtn))
         else:

medusa/storage/__init__.py

@@ -44,6 +44,11 @@
 INDEX_BLOB_WITH_TIMESTAMP_PATTERN = re.compile('.*(started|finished)_(.*)_([0-9]+).timestamp$')
 
 
+def divide_chunks(values, step):
+    for i in range(0, len(values), step):
+        yield values[i:i + step]
+
+
 def format_bytes_str(value):
     for unit_shift, unit in enumerate(['B', 'KB', 'MB', 'GB', 'TB']):
         if value >> (unit_shift * 10) < 1024:

medusa/storage/google_storage.py

@@ -30,6 +30,9 @@
 from medusa.storage.google_cloud_storage.gsutil import GSUtil
 
 
+GSUTIL_MAX_FILES_PER_CHUNK = 64
+
+
 class GoogleStorage(AbstractStorage):
 
     def connect_storage(self):