sdk/python/tests/test_benchmark_collections.py

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