sdk/python/tests/test_benchmark_collections.py

   1 import arvados
   2 import sys
   3
   4 import run_test_server
   5 import arvados_testutil as tutil
   6 import manifest_examples
   7 from performance.performance_profiler import profiled
   8
   9 class CollectionBenchmark(run_test_server.TestCaseWithServers,
  10                           tutil.ArvadosBaseTestCase,
  11                           manifest_examples.ManifestExamples):
  12     MAIN_SERVER = {}
  13     TEST_BLOCK_SIZE = 0
  14
  15     @classmethod
  16     def list_recursive(cls, coll, parent_name=None):
  17         if parent_name is None:
  18             current_name = coll.stream_name()
  19         else:
  20             current_name = '{}/{}'.format(parent_name, coll.name)
  21         try:
  22             for name in coll:
  23                 for item in cls.list_recursive(coll[name], current_name):
  24                     yield item
  25         except TypeError:
  26             yield current_name
  27
  28     @classmethod
  29     def setUpClass(cls):
  30         super(CollectionBenchmark, cls).setUpClass()
  31         run_test_server.authorize_with('active')
  32         cls.api_client = arvados.api('v1')
  33         cls.keep_client = arvados.KeepClient(api_client=cls.api_client,
  34                                              local_store=cls.local_store)
  35
  36     @profiled
  37     def profile_new_collection_from_manifest(self, manifest_text):
  38         return arvados.collection.Collection(manifest_text)
  39
  40     @profiled
  41     def profile_new_collection_from_server(self, uuid):
  42         return arvados.collection.Collection(uuid)
  43
  44     @profiled
  45     def profile_new_collection_copying_bytes_from_collection(self, src):
  46         dst = arvados.collection.Collection()
  47         with tutil.mock_keep_responses('x'*self.TEST_BLOCK_SIZE, 200):
  48             for name in self.list_recursive(src):
  49                 with src.open(name) as srcfile, dst.open(name, 'w') as dstfile:
  50                     dstfile.write(srcfile.read())
  51             dst.save_new()
  52
  53     @profiled
  54     def profile_new_collection_copying_files_from_collection(self, src):
  55         dst = arvados.collection.Collection()
  56         with tutil.mock_keep_responses('x'*self.TEST_BLOCK_SIZE, 200):
  57             for name in self.list_recursive(src):
  58                 dst.copy(name, name, src)
  59             dst.save_new()
  60
  61     @profiled
  62     def profile_collection_list_files(self, coll):
  63         return sum(1 for name in self.list_recursive(coll))
  64
  65     def test_medium_sized_manifest(self):
  66         """Exercise manifest-handling code.
  67
  68         Currently, this test puts undue emphasis on some code paths
  69         that don't reflect typical use because the contrived example
  70         manifest has some unusual characteristics:
  71
  72         * Block size is zero.
  73
  74         * Every block is identical, so block caching patterns are
  75           unrealistic.
  76
  77         * Every file begins and ends at a block boundary.
  78         """
  79         specs = {
  80             'streams': 100,
  81             'files_per_stream': 100,
  82             'blocks_per_file': 20,
  83             'bytes_per_block': self.TEST_BLOCK_SIZE,
  84         }
  85         my_manifest = self.make_manifest(**specs)
  86
  87         coll = self.profile_new_collection_from_manifest(my_manifest)
  88
  89         coll.save_new()
  90         self.profile_new_collection_from_server(coll.manifest_locator())
  91
  92         num_items = self.profile_collection_list_files(coll)
  93         self.assertEqual(num_items, specs['streams'] * specs['files_per_stream'])
  94
  95         self.profile_new_collection_copying_bytes_from_collection(coll)
  96
  97         self.profile_new_collection_copying_files_from_collection(coll)