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