sdk/python/tests/test_keep_client.py

   1 import mock
   2 import os
   3 import socket
   4 import unittest
   5
   6 import arvados
   7 import arvados.retry
   8 import arvados_testutil as tutil
   9 import run_test_server
  10
  11 class KeepTestCase(run_test_server.TestCaseWithServers):
  12     MAIN_SERVER = {}
  13     KEEP_SERVER = {}
  14
  15     @classmethod
  16     def setUpClass(cls):
  17         super(KeepTestCase, cls).setUpClass()
  18         run_test_server.authorize_with("admin")
  19         cls.api_client = arvados.api('v1')
  20         cls.keep_client = arvados.KeepClient(api_client=cls.api_client,
  21                                              proxy='', local_store='')
  22
  23     def test_KeepBasicRWTest(self):
  24         foo_locator = self.keep_client.put('foo')
  25         self.assertRegexpMatches(
  26             foo_locator,
  27             '^acbd18db4cc2f85cedef654fccc4a4d8\+3',
  28             'wrong md5 hash from Keep.put("foo"): ' + foo_locator)
  29         self.assertEqual(self.keep_client.get(foo_locator),
  30                          'foo',
  31                          'wrong content from Keep.get(md5("foo"))')
  32
  33     def test_KeepBinaryRWTest(self):
  34         blob_str = '\xff\xfe\xf7\x00\x01\x02'
  35         blob_locator = self.keep_client.put(blob_str)
  36         self.assertRegexpMatches(
  37             blob_locator,
  38             '^7fc7c53b45e53926ba52821140fef396\+6',
  39             ('wrong locator from Keep.put(<binarydata>):' + blob_locator))
  40         self.assertEqual(self.keep_client.get(blob_locator),
  41                          blob_str,
  42                          'wrong content from Keep.get(md5(<binarydata>))')
  43
  44     def test_KeepLongBinaryRWTest(self):
  45         blob_str = '\xff\xfe\xfd\xfc\x00\x01\x02\x03'
  46         for i in range(0,23):
  47             blob_str = blob_str + blob_str
  48         blob_locator = self.keep_client.put(blob_str)
  49         self.assertRegexpMatches(
  50             blob_locator,
  51             '^84d90fc0d8175dd5dcfab04b999bc956\+67108864',
  52             ('wrong locator from Keep.put(<binarydata>): ' + blob_locator))
  53         self.assertEqual(self.keep_client.get(blob_locator),
  54                          blob_str,
  55                          'wrong content from Keep.get(md5(<binarydata>))')
  56
  57     def test_KeepSingleCopyRWTest(self):
  58         blob_str = '\xff\xfe\xfd\xfc\x00\x01\x02\x03'
  59         blob_locator = self.keep_client.put(blob_str, copies=1)
  60         self.assertRegexpMatches(
  61             blob_locator,
  62             '^c902006bc98a3eb4a3663b65ab4a6fab\+8',
  63             ('wrong locator from Keep.put(<binarydata>): ' + blob_locator))
  64         self.assertEqual(self.keep_client.get(blob_locator),
  65                          blob_str,
  66                          'wrong content from Keep.get(md5(<binarydata>))')
  67
  68
  69 class KeepPermissionTestCase(run_test_server.TestCaseWithServers):
  70     MAIN_SERVER = {}
  71     KEEP_SERVER = {'blob_signing_key': 'abcdefghijk0123456789',
  72                    'enforce_permissions': True}
  73
  74     def test_KeepBasicRWTest(self):
  75         run_test_server.authorize_with('active')
  76         keep_client = arvados.KeepClient()
  77         foo_locator = keep_client.put('foo')
  78         self.assertRegexpMatches(
  79             foo_locator,
  80             r'^acbd18db4cc2f85cedef654fccc4a4d8\+3\+A[a-f0-9]+@[a-f0-9]+$',
  81             'invalid locator from Keep.put("foo"): ' + foo_locator)
  82         self.assertEqual(keep_client.get(foo_locator),
  83                          'foo',
  84                          'wrong content from Keep.get(md5("foo"))')
  85
  86         # GET with an unsigned locator => NotFound
  87         bar_locator = keep_client.put('bar')
  88         unsigned_bar_locator = "37b51d194a7513e45b56f6524f2d51f2+3"
  89         self.assertRegexpMatches(
  90             bar_locator,
  91             r'^37b51d194a7513e45b56f6524f2d51f2\+3\+A[a-f0-9]+@[a-f0-9]+$',
  92             'invalid locator from Keep.put("bar"): ' + bar_locator)
  93         self.assertRaises(arvados.errors.NotFoundError,
  94                           keep_client.get,
  95                           unsigned_bar_locator)
  96
  97         # GET from a different user => NotFound
  98         run_test_server.authorize_with('spectator')
  99         self.assertRaises(arvados.errors.NotFoundError,
 100                           arvados.Keep.get,
 101                           bar_locator)
 102
 103         # Unauthenticated GET for a signed locator => NotFound
 104         # Unauthenticated GET for an unsigned locator => NotFound
 105         keep_client.api_token = ''
 106         self.assertRaises(arvados.errors.NotFoundError,
 107                           keep_client.get,
 108                           bar_locator)
 109         self.assertRaises(arvados.errors.NotFoundError,
 110                           keep_client.get,
 111                           unsigned_bar_locator)
 112
 113
 114 # KeepOptionalPermission: starts Keep with --permission-key-file
 115 # but not --enforce-permissions (i.e. generate signatures on PUT
 116 # requests, but do not require them for GET requests)
 117 #
 118 # All of these requests should succeed when permissions are optional:
 119 # * authenticated request, signed locator
 120 # * authenticated request, unsigned locator
 121 # * unauthenticated request, signed locator
 122 # * unauthenticated request, unsigned locator
 123 class KeepOptionalPermission(run_test_server.TestCaseWithServers):
 124     MAIN_SERVER = {}
 125     KEEP_SERVER = {'blob_signing_key': 'abcdefghijk0123456789',
 126                    'enforce_permissions': False}
 127
 128     @classmethod
 129     def setUpClass(cls):
 130         super(KeepOptionalPermission, cls).setUpClass()
 131         run_test_server.authorize_with("admin")
 132         cls.api_client = arvados.api('v1')
 133
 134     def setUp(self):
 135         super(KeepOptionalPermission, self).setUp()
 136         self.keep_client = arvados.KeepClient(api_client=self.api_client,
 137                                               proxy='', local_store='')
 138
 139     def _put_foo_and_check(self):
 140         signed_locator = self.keep_client.put('foo')
 141         self.assertRegexpMatches(
 142             signed_locator,
 143             r'^acbd18db4cc2f85cedef654fccc4a4d8\+3\+A[a-f0-9]+@[a-f0-9]+$',
 144             'invalid locator from Keep.put("foo"): ' + signed_locator)
 145         return signed_locator
 146
 147     def test_KeepAuthenticatedSignedTest(self):
 148         signed_locator = self._put_foo_and_check()
 149         self.assertEqual(self.keep_client.get(signed_locator),
 150                          'foo',
 151                          'wrong content from Keep.get(md5("foo"))')
 152
 153     def test_KeepAuthenticatedUnsignedTest(self):
 154         signed_locator = self._put_foo_and_check()
 155         self.assertEqual(self.keep_client.get("acbd18db4cc2f85cedef654fccc4a4d8"),
 156                          'foo',
 157                          'wrong content from Keep.get(md5("foo"))')
 158
 159     def test_KeepUnauthenticatedSignedTest(self):
 160         # Check that signed GET requests work even when permissions
 161         # enforcement is off.
 162         signed_locator = self._put_foo_and_check()
 163         self.keep_client.api_token = ''
 164         self.assertEqual(self.keep_client.get(signed_locator),
 165                          'foo',
 166                          'wrong content from Keep.get(md5("foo"))')
 167
 168     def test_KeepUnauthenticatedUnsignedTest(self):
 169         # Since --enforce-permissions is not in effect, GET requests
 170         # need not be authenticated.
 171         signed_locator = self._put_foo_and_check()
 172         self.keep_client.api_token = ''
 173         self.assertEqual(self.keep_client.get("acbd18db4cc2f85cedef654fccc4a4d8"),
 174                          'foo',
 175                          'wrong content from Keep.get(md5("foo"))')
 176
 177
 178 class KeepProxyTestCase(run_test_server.TestCaseWithServers):
 179     MAIN_SERVER = {}
 180     KEEP_SERVER = {}
 181     KEEP_PROXY_SERVER = {'auth': 'admin'}
 182
 183     @classmethod
 184     def setUpClass(cls):
 185         super(KeepProxyTestCase, cls).setUpClass()
 186         cls.api_client = arvados.api('v1')
 187
 188     def tearDown(self):
 189         arvados.config.settings().pop('ARVADOS_EXTERNAL_CLIENT', None)
 190         super(KeepProxyTestCase, self).tearDown()
 191
 192     def test_KeepProxyTest1(self):
 193         # Will use ARVADOS_KEEP_PROXY environment variable that is set by
 194         # setUpClass().
 195         keep_client = arvados.KeepClient(api_client=self.api_client,
 196                                          local_store='')
 197         baz_locator = keep_client.put('baz')
 198         self.assertRegexpMatches(
 199             baz_locator,
 200             '^73feffa4b7f6bb68e44cf984c85f6e88\+3',
 201             'wrong md5 hash from Keep.put("baz"): ' + baz_locator)
 202         self.assertEqual(keep_client.get(baz_locator),
 203                          'baz',
 204                          'wrong content from Keep.get(md5("baz"))')
 205         self.assertTrue(keep_client.using_proxy)
 206
 207     def test_KeepProxyTest2(self):
 208         # Don't instantiate the proxy directly, but set the X-External-Client
 209         # header.  The API server should direct us to the proxy.
 210         arvados.config.settings()['ARVADOS_EXTERNAL_CLIENT'] = 'true'
 211         keep_client = arvados.KeepClient(api_client=self.api_client,
 212                                          proxy='', local_store='')
 213         baz_locator = keep_client.put('baz2')
 214         self.assertRegexpMatches(
 215             baz_locator,
 216             '^91f372a266fe2bf2823cb8ec7fda31ce\+4',
 217             'wrong md5 hash from Keep.put("baz2"): ' + baz_locator)
 218         self.assertEqual(keep_client.get(baz_locator),
 219                          'baz2',
 220                          'wrong content from Keep.get(md5("baz2"))')
 221         self.assertTrue(keep_client.using_proxy)
 222
 223
 224 class KeepClientRetryTestMixin(object):
 225     # Testing with a local Keep store won't exercise the retry behavior.
 226     # Instead, our strategy is:
 227     # * Create a client with one proxy specified (pointed at a black
 228     #   hole), so there's no need to instantiate an API client, and
 229     #   all HTTP requests come from one place.
 230     # * Mock httplib's request method to provide simulated responses.
 231     # This lets us test the retry logic extensively without relying on any
 232     # supporting servers, and prevents side effects in case something hiccups.
 233     # To use this mixin, define DEFAULT_EXPECT, DEFAULT_EXCEPTION, and
 234     # run_method().
 235     PROXY_ADDR = 'http://[%s]:65535/' % (tutil.TEST_HOST,)
 236     TEST_DATA = 'testdata'
 237     TEST_LOCATOR = 'ef654c40ab4f1747fc699915d4f70902+8'
 238
 239     def setUp(self):
 240         self.client_kwargs = {'proxy': self.PROXY_ADDR, 'local_store': ''}
 241
 242     def new_client(self, **caller_kwargs):
 243         kwargs = self.client_kwargs.copy()
 244         kwargs.update(caller_kwargs)
 245         return arvados.KeepClient(**kwargs)
 246
 247     def run_method(self, *args, **kwargs):
 248         raise NotImplementedError("test subclasses must define run_method")
 249
 250     def check_success(self, expected=None, *args, **kwargs):
 251         if expected is None:
 252             expected = self.DEFAULT_EXPECT
 253         self.assertEqual(expected, self.run_method(*args, **kwargs))
 254
 255     def check_exception(self, error_class=None, *args, **kwargs):
 256         if error_class is None:
 257             error_class = self.DEFAULT_EXCEPTION
 258         self.assertRaises(error_class, self.run_method, *args, **kwargs)
 259
 260     def test_immediate_success(self):
 261         with tutil.mock_responses(self.DEFAULT_EXPECT, 200):
 262             self.check_success()
 263
 264     def test_retry_then_success(self):
 265         with tutil.mock_responses(self.DEFAULT_EXPECT, 500, 200):
 266             self.check_success(num_retries=3)
 267
 268     def test_no_default_retry(self):
 269         with tutil.mock_responses(self.DEFAULT_EXPECT, 500, 200):
 270             self.check_exception()
 271
 272     def test_no_retry_after_permanent_error(self):
 273         with tutil.mock_responses(self.DEFAULT_EXPECT, 403, 200):
 274             self.check_exception(num_retries=3)
 275
 276     def test_error_after_retries_exhausted(self):
 277         with tutil.mock_responses(self.DEFAULT_EXPECT, 500, 500, 200):
 278             self.check_exception(num_retries=1)
 279
 280     def test_num_retries_instance_fallback(self):
 281         self.client_kwargs['num_retries'] = 3
 282         with tutil.mock_responses(self.DEFAULT_EXPECT, 500, 200):
 283             self.check_success()
 284
 285
 286 @tutil.skip_sleep
 287 class KeepClientRetryGetTestCase(KeepClientRetryTestMixin, unittest.TestCase):
 288     DEFAULT_EXPECT = KeepClientRetryTestMixin.TEST_DATA
 289     DEFAULT_EXCEPTION = arvados.errors.KeepReadError
 290     HINTED_LOCATOR = KeepClientRetryTestMixin.TEST_LOCATOR + '+K@xyzzy'
 291
 292     def run_method(self, locator=KeepClientRetryTestMixin.TEST_LOCATOR,
 293                    *args, **kwargs):
 294         return self.new_client().get(locator, *args, **kwargs)
 295
 296     def test_specific_exception_when_not_found(self):
 297         with tutil.mock_responses(self.DEFAULT_EXPECT, 404, 200):
 298             self.check_exception(arvados.errors.NotFoundError, num_retries=3)
 299
 300     def test_general_exception_with_mixed_errors(self):
 301         # get should raise a NotFoundError if no server returns the block,
 302         # and a high threshold of servers report that it's not found.
 303         # This test rigs up 50/50 disagreement between two servers, and
 304         # checks that it does not become a NotFoundError.
 305         client = self.new_client()
 306         with tutil.mock_responses(self.DEFAULT_EXPECT, 404, 500):
 307             with self.assertRaises(arvados.errors.KeepReadError) as exc_check:
 308                 client.get(self.HINTED_LOCATOR)
 309             self.assertNotIsInstance(
 310                 exc_check.exception, arvados.errors.NotFoundError,
 311                 "mixed errors raised NotFoundError")
 312
 313     def test_hint_server_can_succeed_without_retries(self):
 314         with tutil.mock_responses(self.DEFAULT_EXPECT, 404, 200, 500):
 315             self.check_success(locator=self.HINTED_LOCATOR)
 316
 317     def test_try_next_server_after_timeout(self):
 318         side_effects = [
 319             socket.timeout("timed out"),
 320             (tutil.fake_httplib2_response(200), self.DEFAULT_EXPECT)]
 321         with mock.patch('httplib2.Http.request',
 322                         side_effect=iter(side_effects)):
 323             self.check_success(locator=self.HINTED_LOCATOR)
 324
 325     def test_retry_data_with_wrong_checksum(self):
 326         side_effects = ((tutil.fake_httplib2_response(200), s)
 327                         for s in ['baddata', self.TEST_DATA])
 328         with mock.patch('httplib2.Http.request', side_effect=side_effects):
 329             self.check_success(locator=self.HINTED_LOCATOR)
 330
 331
 332 @tutil.skip_sleep
 333 class KeepClientRetryPutTestCase(KeepClientRetryTestMixin, unittest.TestCase):
 334     DEFAULT_EXPECT = KeepClientRetryTestMixin.TEST_LOCATOR
 335     DEFAULT_EXCEPTION = arvados.errors.KeepWriteError
 336
 337     def run_method(self, data=KeepClientRetryTestMixin.TEST_DATA,
 338                    copies=1, *args, **kwargs):
 339         return self.new_client().put(data, copies, *args, **kwargs)
 340
 341     def test_do_not_send_multiple_copies_to_same_server(self):
 342         with tutil.mock_responses(self.DEFAULT_EXPECT, 200):
 343             self.check_exception(copies=2, num_retries=3)