17989: Reduces the manifest size limit so that it doesn't fail on extreme cases
[arvados.git] / tools / test-collection-create / test-collection-create.py
index 13754381e186016649fd4db281f9001b22e8a4c1..c8eae240d29b04cae64a1e8a51fc8d5955993826 100644 (file)
@@ -16,6 +16,8 @@ import arvados.collection
 logger = logging.getLogger('arvados.test_collection_create')
 logger.setLevel(logging.INFO)
 
+max_manifest_size = 127*1024*1024
+
 opts = argparse.ArgumentParser(add_help=False)
 opts.add_argument('--min-files', type=int, default=30000, help="""
 Minimum number of files on each directory. Default: 30000.
@@ -381,7 +383,7 @@ def create_substreams(depth, base_stream_name, max_filesize, data_loc, args, cur
     current_size += len(current_stream)
     streams = [current_stream]
 
-    if current_size >= (128 * 1024 * 1024):
+    if current_size >= max_manifest_size:
         logger.debug("Maximum manifest size reached -- finishing early at {}".format(base_stream_name))
     elif depth == 0:
         logger.debug("Finished stream {}".format(base_stream_name))
@@ -391,7 +393,7 @@ def create_substreams(depth, base_stream_name, max_filesize, data_loc, args, cur
             substreams = create_substreams(depth-1, stream_name, max_filesize,
                 data_loc, args, current_size)
             current_size += sum([len(x) for x in substreams])
-            if current_size >= (128 * 1024 * 1024):
+            if current_size >= max_manifest_size:
                 break
             streams.extend(substreams)
     return streams
@@ -421,20 +423,20 @@ def main(arguments=None):
         '.', max_filesize, data_loc, args)
     manifest = ''
     for s in streams:
-        if len(manifest)+len(s) > (1024*1024*128)-2:
+        if len(manifest)+len(s) > max_manifest_size:
             logger.info("Skipping stream {} to avoid making a manifest bigger than 128MiB".format(s.split(' ')[0]))
             break
         manifest += s + '\n'
     try:
-        coll = api.collections().create(body={
-            "ensure_unique_name": True,
-            "collection": {
-                "name": get_random_name(False),
+        coll_name = get_random_name(False)
+        coll = api.collections().create(
+            body={"collection": {
+                "name": coll_name,
                 "manifest_text": manifest
             },
         }).execute()
     except:
-        logger.info("ERROR trying manifest:\n'{}...'\nSize: {}".format(manifest[0:1024], len(manifest)))
+        logger.info("ERROR creating collection with name '{}' and manifest:\n'{}...'\nSize: {}".format(coll_name, manifest[0:1024], len(manifest)))
         raise
     logger.info("Created collection {} - manifest size: {}".format(coll["uuid"], len(manifest)))
     return 0