sdk/python/tests/test_benchmark_collections.py

   1 # Copyright (C) The Arvados Authors. All rights reserved.
   2 #
   3 # SPDX-License-Identifier: Apache-2.0
   4
   5 import arvados
   6 import sys
   7
   8 from . import run_test_server
   9 from . import arvados_testutil as tutil
  10 from . import manifest_examples
  11 from .performance.performance_profiler import profiled
  12
  13 class CollectionBenchmark(run_test_server.TestCaseWithServers,
  14                           tutil.ArvadosBaseTestCase,
  15                           manifest_examples.ManifestExamples):
  16     MAIN_SERVER = {}
  17     TEST_BLOCK_SIZE = 0
  18
  19     @classmethod
  20     def list_recursive(cls, coll, parent_name=None):
  21         if parent_name is None:
  22             current_name = coll.stream_name()
  23         else:
  24             current_name = '{}/{}'.format(parent_name, coll.name)
  25         try:
  26             for name in coll:
  27                 for item in cls.list_recursive(coll[name], current_name):
  28                     yield item
  29         except TypeError:
  30             yield current_name
  31
  32     @classmethod
  33     def setUpClass(cls):
  34         super(CollectionBenchmark, cls).setUpClass()
  35         run_test_server.authorize_with('active')
  36         cls.api_client = arvados.api('v1')
  37         cls.keep_client = arvados.KeepClient(api_client=cls.api_client,
  38                                              local_store=cls.local_store)
  39
  40     @profiled
  41     def profile_new_collection_from_manifest(self, manifest_text):
  42         return arvados.collection.Collection(manifest_text)
  43
  44     @profiled
  45     def profile_new_collection_from_server(self, uuid):
  46         return arvados.collection.Collection(uuid)
  47
  48     @profiled
  49     def profile_new_collection_copying_bytes_from_collection(self, src):
  50         dst = arvados.collection.Collection()
  51         with tutil.mock_keep_responses('x'*self.TEST_BLOCK_SIZE, 200):
  52             for name in self.list_recursive(src):
  53                 with src.open(name, 'rb') as srcfile, dst.open(name, 'wb') as dstfile:
  54                     dstfile.write(srcfile.read())
  55             dst.save_new()
  56
  57     @profiled
  58     def profile_new_collection_copying_files_from_collection(self, src):
  59         dst = arvados.collection.Collection()
  60         with tutil.mock_keep_responses('x'*self.TEST_BLOCK_SIZE, 200):
  61             for name in self.list_recursive(src):
  62                 dst.copy(name, name, src)
  63             dst.save_new()
  64
  65     @profiled
  66     def profile_collection_list_files(self, coll):
  67         return sum(1 for name in self.list_recursive(coll))
  68
  69     def test_medium_sized_manifest(self):
  70         """Exercise manifest-handling code.
  71
  72         Currently, this test puts undue emphasis on some code paths
  73         that don't reflect typical use because the contrived example
  74         manifest has some unusual characteristics:
  75
  76         * Block size is zero.
  77
  78         * Every block is identical, so block caching patterns are
  79           unrealistic.
  80
  81         * Every file begins and ends at a block boundary.
  82         """
  83         specs = {
  84             'streams': 100,
  85             'files_per_stream': 100,
  86             'blocks_per_file': 20,
  87             'bytes_per_block': self.TEST_BLOCK_SIZE,
  88         }
  89         my_manifest = self.make_manifest(**specs)
  90
  91         coll = self.profile_new_collection_from_manifest(my_manifest)
  92
  93         coll.save_new()
  94         self.profile_new_collection_from_server(coll.manifest_locator())
  95
  96         num_items = self.profile_collection_list_files(coll)
  97         self.assertEqual(num_items, specs['streams'] * specs['files_per_stream'])
  98
  99         self.profile_new_collection_copying_bytes_from_collection(coll)
 100
 101         self.profile_new_collection_copying_files_from_collection(coll)