sdk/python/tests/test_keep_client.py

   1 import hashlib
   2 import mock
   3 import os
   4 import pycurl
   5 import random
   6 import re
   7 import socket
   8 import threading
   9 import time
  10 import unittest
  11 import urlparse
  12
  13 import arvados
  14 import arvados.retry
  15 import arvados_testutil as tutil
  16 import keepstub
  17 import run_test_server
  18
  19 class KeepTestCase(run_test_server.TestCaseWithServers):
  20     MAIN_SERVER = {}
  21     KEEP_SERVER = {}
  22
  23     @classmethod
  24     def setUpClass(cls):
  25         super(KeepTestCase, cls).setUpClass()
  26         run_test_server.authorize_with("admin")
  27         cls.api_client = arvados.api('v1')
  28         cls.keep_client = arvados.KeepClient(api_client=cls.api_client,
  29                                              proxy='', local_store='')
  30
  31     def test_KeepBasicRWTest(self):
  32         self.assertEqual(0, self.keep_client.upload_counter.get())
  33         foo_locator = self.keep_client.put('foo')
  34         self.assertRegexpMatches(
  35             foo_locator,
  36             '^acbd18db4cc2f85cedef654fccc4a4d8\+3',
  37             'wrong md5 hash from Keep.put("foo"): ' + foo_locator)
  38
  39         # 6 bytes because uploaded 2 copies
  40         self.assertEqual(6, self.keep_client.upload_counter.get())
  41
  42         self.assertEqual(0, self.keep_client.download_counter.get())
  43         self.assertEqual(self.keep_client.get(foo_locator),
  44                          'foo',
  45                          'wrong content from Keep.get(md5("foo"))')
  46         self.assertEqual(3, self.keep_client.download_counter.get())
  47
  48     def test_KeepBinaryRWTest(self):
  49         blob_str = '\xff\xfe\xf7\x00\x01\x02'
  50         blob_locator = self.keep_client.put(blob_str)
  51         self.assertRegexpMatches(
  52             blob_locator,
  53             '^7fc7c53b45e53926ba52821140fef396\+6',
  54             ('wrong locator from Keep.put(<binarydata>):' + blob_locator))
  55         self.assertEqual(self.keep_client.get(blob_locator),
  56                          blob_str,
  57                          'wrong content from Keep.get(md5(<binarydata>))')
  58
  59     def test_KeepLongBinaryRWTest(self):
  60         blob_str = '\xff\xfe\xfd\xfc\x00\x01\x02\x03'
  61         for i in range(0,23):
  62             blob_str = blob_str + blob_str
  63         blob_locator = self.keep_client.put(blob_str)
  64         self.assertRegexpMatches(
  65             blob_locator,
  66             '^84d90fc0d8175dd5dcfab04b999bc956\+67108864',
  67             ('wrong locator from Keep.put(<binarydata>): ' + blob_locator))
  68         self.assertEqual(self.keep_client.get(blob_locator),
  69                          blob_str,
  70                          'wrong content from Keep.get(md5(<binarydata>))')
  71
  72     @unittest.skip("unreliable test - please fix and close #8752")
  73     def test_KeepSingleCopyRWTest(self):
  74         blob_str = '\xff\xfe\xfd\xfc\x00\x01\x02\x03'
  75         blob_locator = self.keep_client.put(blob_str, copies=1)
  76         self.assertRegexpMatches(
  77             blob_locator,
  78             '^c902006bc98a3eb4a3663b65ab4a6fab\+8',
  79             ('wrong locator from Keep.put(<binarydata>): ' + blob_locator))
  80         self.assertEqual(self.keep_client.get(blob_locator),
  81                          blob_str,
  82                          'wrong content from Keep.get(md5(<binarydata>))')
  83
  84     def test_KeepEmptyCollectionTest(self):
  85         blob_locator = self.keep_client.put('', copies=1)
  86         self.assertRegexpMatches(
  87             blob_locator,
  88             '^d41d8cd98f00b204e9800998ecf8427e\+0',
  89             ('wrong locator from Keep.put(""): ' + blob_locator))
  90
  91     def test_unicode_must_be_ascii(self):
  92         # If unicode type, must only consist of valid ASCII
  93         foo_locator = self.keep_client.put(u'foo')
  94         self.assertRegexpMatches(
  95             foo_locator,
  96             '^acbd18db4cc2f85cedef654fccc4a4d8\+3',
  97             'wrong md5 hash from Keep.put("foo"): ' + foo_locator)
  98
  99         with self.assertRaises(UnicodeEncodeError):
 100             # Error if it is not ASCII
 101             self.keep_client.put(u'\xe2')
 102
 103         with self.assertRaises(arvados.errors.ArgumentError):
 104             # Must be a string type
 105             self.keep_client.put({})
 106
 107     def test_KeepHeadTest(self):
 108         locator = self.keep_client.put('test_head')
 109         self.assertRegexpMatches(
 110             locator,
 111             '^b9a772c7049325feb7130fff1f8333e9\+9',
 112             'wrong md5 hash from Keep.put for "test_head": ' + locator)
 113         self.assertEqual(True, self.keep_client.head(locator))
 114         self.assertEqual(self.keep_client.get(locator),
 115                          'test_head',
 116                          'wrong content from Keep.get for "test_head"')
 117
 118 class KeepPermissionTestCase(run_test_server.TestCaseWithServers):
 119     MAIN_SERVER = {}
 120     KEEP_SERVER = {'blob_signing_key': 'abcdefghijk0123456789',
 121                    'enforce_permissions': True}
 122
 123     def test_KeepBasicRWTest(self):
 124         run_test_server.authorize_with('active')
 125         keep_client = arvados.KeepClient()
 126         foo_locator = keep_client.put('foo')
 127         self.assertRegexpMatches(
 128             foo_locator,
 129             r'^acbd18db4cc2f85cedef654fccc4a4d8\+3\+A[a-f0-9]+@[a-f0-9]+$',
 130             'invalid locator from Keep.put("foo"): ' + foo_locator)
 131         self.assertEqual(keep_client.get(foo_locator),
 132                          'foo',
 133                          'wrong content from Keep.get(md5("foo"))')
 134
 135         # GET with an unsigned locator => NotFound
 136         bar_locator = keep_client.put('bar')
 137         unsigned_bar_locator = "37b51d194a7513e45b56f6524f2d51f2+3"
 138         self.assertRegexpMatches(
 139             bar_locator,
 140             r'^37b51d194a7513e45b56f6524f2d51f2\+3\+A[a-f0-9]+@[a-f0-9]+$',
 141             'invalid locator from Keep.put("bar"): ' + bar_locator)
 142         self.assertRaises(arvados.errors.NotFoundError,
 143                           keep_client.get,
 144                           unsigned_bar_locator)
 145
 146         # GET from a different user => NotFound
 147         run_test_server.authorize_with('spectator')
 148         self.assertRaises(arvados.errors.NotFoundError,
 149                           arvados.Keep.get,
 150                           bar_locator)
 151
 152         # Unauthenticated GET for a signed locator => NotFound
 153         # Unauthenticated GET for an unsigned locator => NotFound
 154         keep_client.api_token = ''
 155         self.assertRaises(arvados.errors.NotFoundError,
 156                           keep_client.get,
 157                           bar_locator)
 158         self.assertRaises(arvados.errors.NotFoundError,
 159                           keep_client.get,
 160                           unsigned_bar_locator)
 161
 162
 163 # KeepOptionalPermission: starts Keep with --permission-key-file
 164 # but not --enforce-permissions (i.e. generate signatures on PUT
 165 # requests, but do not require them for GET requests)
 166 #
 167 # All of these requests should succeed when permissions are optional:
 168 # * authenticated request, signed locator
 169 # * authenticated request, unsigned locator
 170 # * unauthenticated request, signed locator
 171 # * unauthenticated request, unsigned locator
 172 class KeepOptionalPermission(run_test_server.TestCaseWithServers):
 173     MAIN_SERVER = {}
 174     KEEP_SERVER = {'blob_signing_key': 'abcdefghijk0123456789',
 175                    'enforce_permissions': False}
 176
 177     @classmethod
 178     def setUpClass(cls):
 179         super(KeepOptionalPermission, cls).setUpClass()
 180         run_test_server.authorize_with("admin")
 181         cls.api_client = arvados.api('v1')
 182
 183     def setUp(self):
 184         super(KeepOptionalPermission, self).setUp()
 185         self.keep_client = arvados.KeepClient(api_client=self.api_client,
 186                                               proxy='', local_store='')
 187
 188     def _put_foo_and_check(self):
 189         signed_locator = self.keep_client.put('foo')
 190         self.assertRegexpMatches(
 191             signed_locator,
 192             r'^acbd18db4cc2f85cedef654fccc4a4d8\+3\+A[a-f0-9]+@[a-f0-9]+$',
 193             'invalid locator from Keep.put("foo"): ' + signed_locator)
 194         return signed_locator
 195
 196     def test_KeepAuthenticatedSignedTest(self):
 197         signed_locator = self._put_foo_and_check()
 198         self.assertEqual(self.keep_client.get(signed_locator),
 199                          'foo',
 200                          'wrong content from Keep.get(md5("foo"))')
 201
 202     def test_KeepAuthenticatedUnsignedTest(self):
 203         signed_locator = self._put_foo_and_check()
 204         self.assertEqual(self.keep_client.get("acbd18db4cc2f85cedef654fccc4a4d8"),
 205                          'foo',
 206                          'wrong content from Keep.get(md5("foo"))')
 207
 208     def test_KeepUnauthenticatedSignedTest(self):
 209         # Check that signed GET requests work even when permissions
 210         # enforcement is off.
 211         signed_locator = self._put_foo_and_check()
 212         self.keep_client.api_token = ''
 213         self.assertEqual(self.keep_client.get(signed_locator),
 214                          'foo',
 215                          'wrong content from Keep.get(md5("foo"))')
 216
 217     def test_KeepUnauthenticatedUnsignedTest(self):
 218         # Since --enforce-permissions is not in effect, GET requests
 219         # need not be authenticated.
 220         signed_locator = self._put_foo_and_check()
 221         self.keep_client.api_token = ''
 222         self.assertEqual(self.keep_client.get("acbd18db4cc2f85cedef654fccc4a4d8"),
 223                          'foo',
 224                          'wrong content from Keep.get(md5("foo"))')
 225
 226
 227 class KeepProxyTestCase(run_test_server.TestCaseWithServers):
 228     MAIN_SERVER = {}
 229     KEEP_SERVER = {}
 230     KEEP_PROXY_SERVER = {}
 231
 232     @classmethod
 233     def setUpClass(cls):
 234         super(KeepProxyTestCase, cls).setUpClass()
 235         run_test_server.authorize_with('active')
 236         cls.api_client = arvados.api('v1')
 237
 238     def tearDown(self):
 239         arvados.config.settings().pop('ARVADOS_EXTERNAL_CLIENT', None)
 240         super(KeepProxyTestCase, self).tearDown()
 241
 242     def test_KeepProxyTest1(self):
 243         # Will use ARVADOS_KEEP_SERVICES environment variable that
 244         # is set by setUpClass().
 245         keep_client = arvados.KeepClient(api_client=self.api_client,
 246                                          local_store='')
 247         baz_locator = keep_client.put('baz')
 248         self.assertRegexpMatches(
 249             baz_locator,
 250             '^73feffa4b7f6bb68e44cf984c85f6e88\+3',
 251             'wrong md5 hash from Keep.put("baz"): ' + baz_locator)
 252         self.assertEqual(keep_client.get(baz_locator),
 253                          'baz',
 254                          'wrong content from Keep.get(md5("baz"))')
 255         self.assertTrue(keep_client.using_proxy)
 256
 257     def test_KeepProxyTest2(self):
 258         # Don't instantiate the proxy directly, but set the X-External-Client
 259         # header.  The API server should direct us to the proxy.
 260         arvados.config.settings()['ARVADOS_EXTERNAL_CLIENT'] = 'true'
 261         keep_client = arvados.KeepClient(api_client=self.api_client,
 262                                          proxy='', local_store='')
 263         baz_locator = keep_client.put('baz2')
 264         self.assertRegexpMatches(
 265             baz_locator,
 266             '^91f372a266fe2bf2823cb8ec7fda31ce\+4',
 267             'wrong md5 hash from Keep.put("baz2"): ' + baz_locator)
 268         self.assertEqual(keep_client.get(baz_locator),
 269                          'baz2',
 270                          'wrong content from Keep.get(md5("baz2"))')
 271         self.assertTrue(keep_client.using_proxy)
 272
 273     def test_KeepProxyTestMultipleURIs(self):
 274         # Test using ARVADOS_KEEP_SERVICES env var overriding any
 275         # existing proxy setting and setting multiple proxies
 276         arvados.config.settings()['ARVADOS_KEEP_SERVICES'] = 'http://10.0.0.1 https://foo.example.org:1234/'
 277         keep_client = arvados.KeepClient(api_client=self.api_client,
 278                                          local_store='')
 279         uris = [x['_service_root'] for x in keep_client._keep_services]
 280         self.assertEqual(uris, ['http://10.0.0.1/',
 281                                 'https://foo.example.org:1234/'])
 282
 283     def test_KeepProxyTestInvalidURI(self):
 284         arvados.config.settings()['ARVADOS_KEEP_SERVICES'] = 'bad.uri.org'
 285         with self.assertRaises(arvados.errors.ArgumentError):
 286             keep_client = arvados.KeepClient(api_client=self.api_client,
 287                                              local_store='')
 288
 289
 290 class KeepClientServiceTestCase(unittest.TestCase, tutil.ApiClientMock):
 291     def get_service_roots(self, api_client):
 292         keep_client = arvados.KeepClient(api_client=api_client)
 293         services = keep_client.weighted_service_roots(arvados.KeepLocator('0'*32))
 294         return [urlparse.urlparse(url) for url in sorted(services)]
 295
 296     def test_ssl_flag_respected_in_roots(self):
 297         for ssl_flag in [False, True]:
 298             services = self.get_service_roots(self.mock_keep_services(
 299                 service_ssl_flag=ssl_flag))
 300             self.assertEqual(
 301                 ('https' if ssl_flag else 'http'), services[0].scheme)
 302
 303     def test_correct_ports_with_ipv6_addresses(self):
 304         service = self.get_service_roots(self.mock_keep_services(
 305             service_type='proxy', service_host='100::1', service_port=10, count=1))[0]
 306         self.assertEqual('100::1', service.hostname)
 307         self.assertEqual(10, service.port)
 308
 309     # test_*_timeout verify that KeepClient instructs pycurl to use
 310     # the appropriate connection and read timeouts. They don't care
 311     # whether pycurl actually exhibits the expected timeout behavior
 312     # -- those tests are in the KeepClientTimeout test class.
 313
 314     def test_get_timeout(self):
 315         api_client = self.mock_keep_services(count=1)
 316         force_timeout = socket.timeout("timed out")
 317         with tutil.mock_keep_responses(force_timeout, 0) as mock:
 318             keep_client = arvados.KeepClient(api_client=api_client)
 319             with self.assertRaises(arvados.errors.KeepReadError):
 320                 keep_client.get('ffffffffffffffffffffffffffffffff')
 321             self.assertEqual(
 322                 mock.responses[0].getopt(pycurl.CONNECTTIMEOUT_MS),
 323                 int(arvados.KeepClient.DEFAULT_TIMEOUT[0]*1000))
 324             self.assertEqual(
 325                 mock.responses[0].getopt(pycurl.LOW_SPEED_TIME),
 326                 int(arvados.KeepClient.DEFAULT_TIMEOUT[1]))
 327             self.assertEqual(
 328                 mock.responses[0].getopt(pycurl.LOW_SPEED_LIMIT),
 329                 int(arvados.KeepClient.DEFAULT_TIMEOUT[2]))
 330
 331     def test_put_timeout(self):
 332         api_client = self.mock_keep_services(count=1)
 333         force_timeout = socket.timeout("timed out")
 334         with tutil.mock_keep_responses(force_timeout, 0) as mock:
 335             keep_client = arvados.KeepClient(api_client=api_client)
 336             with self.assertRaises(arvados.errors.KeepWriteError):
 337                 keep_client.put('foo')
 338             self.assertEqual(
 339                 mock.responses[0].getopt(pycurl.CONNECTTIMEOUT_MS),
 340                 int(arvados.KeepClient.DEFAULT_TIMEOUT[0]*1000))
 341             self.assertEqual(
 342                 mock.responses[0].getopt(pycurl.LOW_SPEED_TIME),
 343                 int(arvados.KeepClient.DEFAULT_TIMEOUT[1]))
 344             self.assertEqual(
 345                 mock.responses[0].getopt(pycurl.LOW_SPEED_LIMIT),
 346                 int(arvados.KeepClient.DEFAULT_TIMEOUT[2]))
 347
 348     def test_head_timeout(self):
 349         api_client = self.mock_keep_services(count=1)
 350         force_timeout = socket.timeout("timed out")
 351         with tutil.mock_keep_responses(force_timeout, 0) as mock:
 352             keep_client = arvados.KeepClient(api_client=api_client)
 353             with self.assertRaises(arvados.errors.KeepReadError):
 354                 keep_client.head('ffffffffffffffffffffffffffffffff')
 355             self.assertEqual(
 356                 mock.responses[0].getopt(pycurl.CONNECTTIMEOUT_MS),
 357                 int(arvados.KeepClient.DEFAULT_TIMEOUT[0]*1000))
 358             self.assertEqual(
 359                 mock.responses[0].getopt(pycurl.LOW_SPEED_TIME),
 360                 int(arvados.KeepClient.DEFAULT_TIMEOUT[1]))
 361             self.assertEqual(
 362                 mock.responses[0].getopt(pycurl.LOW_SPEED_LIMIT),
 363                 int(arvados.KeepClient.DEFAULT_TIMEOUT[2]))
 364
 365     def test_proxy_get_timeout(self):
 366         api_client = self.mock_keep_services(service_type='proxy', count=1)
 367         force_timeout = socket.timeout("timed out")
 368         with tutil.mock_keep_responses(force_timeout, 0) as mock:
 369             keep_client = arvados.KeepClient(api_client=api_client)
 370             with self.assertRaises(arvados.errors.KeepReadError):
 371                 keep_client.get('ffffffffffffffffffffffffffffffff')
 372             self.assertEqual(
 373                 mock.responses[0].getopt(pycurl.CONNECTTIMEOUT_MS),
 374                 int(arvados.KeepClient.DEFAULT_PROXY_TIMEOUT[0]*1000))
 375             self.assertEqual(
 376                 mock.responses[0].getopt(pycurl.LOW_SPEED_TIME),
 377                 int(arvados.KeepClient.DEFAULT_PROXY_TIMEOUT[1]))
 378             self.assertEqual(
 379                 mock.responses[0].getopt(pycurl.LOW_SPEED_LIMIT),
 380                 int(arvados.KeepClient.DEFAULT_PROXY_TIMEOUT[2]))
 381
 382     def test_proxy_head_timeout(self):
 383         api_client = self.mock_keep_services(service_type='proxy', count=1)
 384         force_timeout = socket.timeout("timed out")
 385         with tutil.mock_keep_responses(force_timeout, 0) as mock:
 386             keep_client = arvados.KeepClient(api_client=api_client)
 387             with self.assertRaises(arvados.errors.KeepReadError):
 388                 keep_client.head('ffffffffffffffffffffffffffffffff')
 389             self.assertEqual(
 390                 mock.responses[0].getopt(pycurl.CONNECTTIMEOUT_MS),
 391                 int(arvados.KeepClient.DEFAULT_PROXY_TIMEOUT[0]*1000))
 392             self.assertEqual(
 393                 mock.responses[0].getopt(pycurl.LOW_SPEED_TIME),
 394                 int(arvados.KeepClient.DEFAULT_PROXY_TIMEOUT[1]))
 395             self.assertEqual(
 396                 mock.responses[0].getopt(pycurl.LOW_SPEED_LIMIT),
 397                 int(arvados.KeepClient.DEFAULT_PROXY_TIMEOUT[2]))
 398
 399     def test_proxy_put_timeout(self):
 400         api_client = self.mock_keep_services(service_type='proxy', count=1)
 401         force_timeout = socket.timeout("timed out")
 402         with tutil.mock_keep_responses(force_timeout, 0) as mock:
 403             keep_client = arvados.KeepClient(api_client=api_client)
 404             with self.assertRaises(arvados.errors.KeepWriteError):
 405                 keep_client.put('foo')
 406             self.assertEqual(
 407                 mock.responses[0].getopt(pycurl.CONNECTTIMEOUT_MS),
 408                 int(arvados.KeepClient.DEFAULT_PROXY_TIMEOUT[0]*1000))
 409             self.assertEqual(
 410                 mock.responses[0].getopt(pycurl.LOW_SPEED_TIME),
 411                 int(arvados.KeepClient.DEFAULT_PROXY_TIMEOUT[1]))
 412             self.assertEqual(
 413                 mock.responses[0].getopt(pycurl.LOW_SPEED_LIMIT),
 414                 int(arvados.KeepClient.DEFAULT_PROXY_TIMEOUT[2]))
 415
 416     def check_no_services_error(self, verb, exc_class):
 417         api_client = mock.MagicMock(name='api_client')
 418         api_client.keep_services().accessible().execute.side_effect = (
 419             arvados.errors.ApiError)
 420         keep_client = arvados.KeepClient(api_client=api_client)
 421         with self.assertRaises(exc_class) as err_check:
 422             getattr(keep_client, verb)('d41d8cd98f00b204e9800998ecf8427e+0')
 423         self.assertEqual(0, len(err_check.exception.request_errors()))
 424
 425     def test_get_error_with_no_services(self):
 426         self.check_no_services_error('get', arvados.errors.KeepReadError)
 427
 428     def test_head_error_with_no_services(self):
 429         self.check_no_services_error('head', arvados.errors.KeepReadError)
 430
 431     def test_put_error_with_no_services(self):
 432         self.check_no_services_error('put', arvados.errors.KeepWriteError)
 433
 434     def check_errors_from_last_retry(self, verb, exc_class):
 435         api_client = self.mock_keep_services(count=2)
 436         req_mock = tutil.mock_keep_responses(
 437             "retry error reporting test", 500, 500, 403, 403)
 438         with req_mock, tutil.skip_sleep, \
 439                 self.assertRaises(exc_class) as err_check:
 440             keep_client = arvados.KeepClient(api_client=api_client)
 441             getattr(keep_client, verb)('d41d8cd98f00b204e9800998ecf8427e+0',
 442                                        num_retries=3)
 443         self.assertEqual([403, 403], [
 444                 getattr(error, 'status_code', None)
 445                 for error in err_check.exception.request_errors().itervalues()])
 446
 447     def test_get_error_reflects_last_retry(self):
 448         self.check_errors_from_last_retry('get', arvados.errors.KeepReadError)
 449
 450     def test_head_error_reflects_last_retry(self):
 451         self.check_errors_from_last_retry('head', arvados.errors.KeepReadError)
 452
 453     def test_put_error_reflects_last_retry(self):
 454         self.check_errors_from_last_retry('put', arvados.errors.KeepWriteError)
 455
 456     def test_put_error_does_not_include_successful_puts(self):
 457         data = 'partial failure test'
 458         data_loc = tutil.str_keep_locator(data)
 459         api_client = self.mock_keep_services(count=3)
 460         with tutil.mock_keep_responses(data_loc, 200, 500, 500) as req_mock, \
 461                 self.assertRaises(arvados.errors.KeepWriteError) as exc_check:
 462             keep_client = arvados.KeepClient(api_client=api_client)
 463             keep_client.put(data)
 464         self.assertEqual(2, len(exc_check.exception.request_errors()))
 465
 466     def test_proxy_put_with_no_writable_services(self):
 467         data = 'test with no writable services'
 468         data_loc = tutil.str_keep_locator(data)
 469         api_client = self.mock_keep_services(service_type='proxy', read_only=True, count=1)
 470         with tutil.mock_keep_responses(data_loc, 200, 500, 500) as req_mock, \
 471                 self.assertRaises(arvados.errors.KeepWriteError) as exc_check:
 472           keep_client = arvados.KeepClient(api_client=api_client)
 473           keep_client.put(data)
 474         self.assertEqual(True, ("no Keep services available" in str(exc_check.exception)))
 475         self.assertEqual(0, len(exc_check.exception.request_errors()))
 476
 477     def test_oddball_service_get(self):
 478         body = 'oddball service get'
 479         api_client = self.mock_keep_services(service_type='fancynewblobstore')
 480         with tutil.mock_keep_responses(body, 200):
 481             keep_client = arvados.KeepClient(api_client=api_client)
 482             actual = keep_client.get(tutil.str_keep_locator(body))
 483         self.assertEqual(body, actual)
 484
 485     def test_oddball_service_put(self):
 486         body = 'oddball service put'
 487         pdh = tutil.str_keep_locator(body)
 488         api_client = self.mock_keep_services(service_type='fancynewblobstore')
 489         with tutil.mock_keep_responses(pdh, 200):
 490             keep_client = arvados.KeepClient(api_client=api_client)
 491             actual = keep_client.put(body, copies=1)
 492         self.assertEqual(pdh, actual)
 493
 494     def test_oddball_service_writer_count(self):
 495         body = 'oddball service writer count'
 496         pdh = tutil.str_keep_locator(body)
 497         api_client = self.mock_keep_services(service_type='fancynewblobstore',
 498                                              count=4)
 499         headers = {'x-keep-replicas-stored': 3}
 500         with tutil.mock_keep_responses(pdh, 200, 418, 418, 418,
 501                                        **headers) as req_mock:
 502             keep_client = arvados.KeepClient(api_client=api_client)
 503             actual = keep_client.put(body, copies=2)
 504         self.assertEqual(pdh, actual)
 505         self.assertEqual(1, req_mock.call_count)
 506
 507
 508 @tutil.skip_sleep
 509 class KeepClientRendezvousTestCase(unittest.TestCase, tutil.ApiClientMock):
 510
 511     def setUp(self):
 512         # expected_order[i] is the probe order for
 513         # hash=md5(sprintf("%064x",i)) where there are 16 services
 514         # with uuid sprintf("anything-%015x",j) with j in 0..15. E.g.,
 515         # the first probe for the block consisting of 64 "0"
 516         # characters is the service whose uuid is
 517         # "zzzzz-bi6l4-000000000000003", so expected_order[0][0]=='3'.
 518         self.services = 16
 519         self.expected_order = [
 520             list('3eab2d5fc9681074'),
 521             list('097dba52e648f1c3'),
 522             list('c5b4e023f8a7d691'),
 523             list('9d81c02e76a3bf54'),
 524             ]
 525         self.blocks = [
 526             "{:064x}".format(x)
 527             for x in range(len(self.expected_order))]
 528         self.hashes = [
 529             hashlib.md5(self.blocks[x]).hexdigest()
 530             for x in range(len(self.expected_order))]
 531         self.api_client = self.mock_keep_services(count=self.services)
 532         self.keep_client = arvados.KeepClient(api_client=self.api_client)
 533
 534     def test_weighted_service_roots_against_reference_set(self):
 535         # Confirm weighted_service_roots() returns the correct order
 536         for i, hash in enumerate(self.hashes):
 537             roots = self.keep_client.weighted_service_roots(arvados.KeepLocator(hash))
 538             got_order = [
 539                 re.search(r'//\[?keep0x([0-9a-f]+)', root).group(1)
 540                 for root in roots]
 541             self.assertEqual(self.expected_order[i], got_order)
 542
 543     def test_get_probe_order_against_reference_set(self):
 544         self._test_probe_order_against_reference_set(
 545             lambda i: self.keep_client.get(self.hashes[i], num_retries=1))
 546
 547     def test_head_probe_order_against_reference_set(self):
 548         self._test_probe_order_against_reference_set(
 549             lambda i: self.keep_client.head(self.hashes[i], num_retries=1))
 550
 551     def test_put_probe_order_against_reference_set(self):
 552         # copies=1 prevents the test from being sensitive to races
 553         # between writer threads.
 554         self._test_probe_order_against_reference_set(
 555             lambda i: self.keep_client.put(self.blocks[i], num_retries=1, copies=1))
 556
 557     def _test_probe_order_against_reference_set(self, op):
 558         for i in range(len(self.blocks)):
 559             with tutil.mock_keep_responses('', *[500 for _ in range(self.services*2)]) as mock, \
 560                  self.assertRaises(arvados.errors.KeepRequestError):
 561                 op(i)
 562             got_order = [
 563                 re.search(r'//\[?keep0x([0-9a-f]+)', resp.getopt(pycurl.URL)).group(1)
 564                 for resp in mock.responses]
 565             self.assertEqual(self.expected_order[i]*2, got_order)
 566
 567     def test_put_probe_order_multiple_copies(self):
 568         for copies in range(2, 4):
 569             for i in range(len(self.blocks)):
 570                 with tutil.mock_keep_responses('', *[500 for _ in range(self.services*3)]) as mock, \
 571                      self.assertRaises(arvados.errors.KeepWriteError):
 572                     self.keep_client.put(self.blocks[i], num_retries=2, copies=copies)
 573                 got_order = [
 574                     re.search(r'//\[?keep0x([0-9a-f]+)', resp.getopt(pycurl.URL)).group(1)
 575                     for resp in mock.responses]
 576                 # With T threads racing to make requests, the position
 577                 # of a given server in the sequence of HTTP requests
 578                 # (got_order) cannot be more than T-1 positions
 579                 # earlier than that server's position in the reference
 580                 # probe sequence (expected_order).
 581                 #
 582                 # Loop invariant: we have accounted for +pos+ expected
 583                 # probes, either by seeing them in +got_order+ or by
 584                 # putting them in +pending+ in the hope of seeing them
 585                 # later. As long as +len(pending)<T+, we haven't
 586                 # started a request too early.
 587                 pending = []
 588                 for pos, expected in enumerate(self.expected_order[i]*3):
 589                     got = got_order[pos-len(pending)]
 590                     while got in pending:
 591                         del pending[pending.index(got)]
 592                         got = got_order[pos-len(pending)]
 593                     if got != expected:
 594                         pending.append(expected)
 595                         self.assertLess(
 596                             len(pending), copies,
 597                             "pending={}, with copies={}, got {}, expected {}".format(
 598                                 pending, copies, repr(got_order), repr(self.expected_order[i]*3)))
 599
 600     def test_probe_waste_adding_one_server(self):
 601         hashes = [
 602             hashlib.md5("{:064x}".format(x)).hexdigest() for x in range(100)]
 603         initial_services = 12
 604         self.api_client = self.mock_keep_services(count=initial_services)
 605         self.keep_client = arvados.KeepClient(api_client=self.api_client)
 606         probes_before = [
 607             self.keep_client.weighted_service_roots(arvados.KeepLocator(hash)) for hash in hashes]
 608         for added_services in range(1, 12):
 609             api_client = self.mock_keep_services(count=initial_services+added_services)
 610             keep_client = arvados.KeepClient(api_client=api_client)
 611             total_penalty = 0
 612             for hash_index in range(len(hashes)):
 613                 probe_after = keep_client.weighted_service_roots(
 614                     arvados.KeepLocator(hashes[hash_index]))
 615                 penalty = probe_after.index(probes_before[hash_index][0])
 616                 self.assertLessEqual(penalty, added_services)
 617                 total_penalty += penalty
 618             # Average penalty per block should not exceed
 619             # N(added)/N(orig) by more than 20%, and should get closer
 620             # to the ideal as we add data points.
 621             expect_penalty = (
 622                 added_services *
 623                 len(hashes) / initial_services)
 624             max_penalty = (
 625                 expect_penalty *
 626                 (120 - added_services)/100)
 627             min_penalty = (
 628                 expect_penalty * 8/10)
 629             self.assertTrue(
 630                 min_penalty <= total_penalty <= max_penalty,
 631                 "With {}+{} services, {} blocks, penalty {} but expected {}..{}".format(
 632                     initial_services,
 633                     added_services,
 634                     len(hashes),
 635                     total_penalty,
 636                     min_penalty,
 637                     max_penalty))
 638
 639     def check_64_zeros_error_order(self, verb, exc_class):
 640         data = '0' * 64
 641         if verb == 'get':
 642             data = tutil.str_keep_locator(data)
 643         # Arbitrary port number:
 644         aport = random.randint(1024,65535)
 645         api_client = self.mock_keep_services(service_port=aport, count=self.services)
 646         keep_client = arvados.KeepClient(api_client=api_client)
 647         with mock.patch('pycurl.Curl') as curl_mock, \
 648              self.assertRaises(exc_class) as err_check:
 649             curl_mock.return_value.side_effect = socket.timeout
 650             getattr(keep_client, verb)(data)
 651         urls = [urlparse.urlparse(url)
 652                 for url in err_check.exception.request_errors()]
 653         self.assertEqual([('keep0x' + c, aport) for c in '3eab2d5fc9681074'],
 654                          [(url.hostname, url.port) for url in urls])
 655
 656     def test_get_error_shows_probe_order(self):
 657         self.check_64_zeros_error_order('get', arvados.errors.KeepReadError)
 658
 659     def test_put_error_shows_probe_order(self):
 660         self.check_64_zeros_error_order('put', arvados.errors.KeepWriteError)
 661
 662
 663 class KeepClientTimeout(unittest.TestCase, tutil.ApiClientMock):
 664     # BANDWIDTH_LOW_LIM must be less than len(DATA) so we can transfer
 665     # 1s worth of data and then trigger bandwidth errors before running
 666     # out of data.
 667     DATA = 'x'*2**11
 668     BANDWIDTH_LOW_LIM = 1024
 669     TIMEOUT_TIME = 1.0
 670
 671     class assertTakesBetween(unittest.TestCase):
 672         def __init__(self, tmin, tmax):
 673             self.tmin = tmin
 674             self.tmax = tmax
 675
 676         def __enter__(self):
 677             self.t0 = time.time()
 678
 679         def __exit__(self, *args, **kwargs):
 680             # Round times to milliseconds, like CURL. Otherwise, we
 681             # fail when CURL reaches a 1s timeout at 0.9998s.
 682             delta = round(time.time() - self.t0, 3)
 683             self.assertGreaterEqual(delta, self.tmin)
 684             self.assertLessEqual(delta, self.tmax)
 685
 686     class assertTakesGreater(unittest.TestCase):
 687         def __init__(self, tmin):
 688             self.tmin = tmin
 689
 690         def __enter__(self):
 691             self.t0 = time.time()
 692
 693         def __exit__(self, *args, **kwargs):
 694             delta = round(time.time() - self.t0, 3)
 695             self.assertGreaterEqual(delta, self.tmin)
 696
 697     def setUp(self):
 698         sock = socket.socket()
 699         sock.bind(('0.0.0.0', 0))
 700         self.port = sock.getsockname()[1]
 701         sock.close()
 702         self.server = keepstub.Server(('0.0.0.0', self.port), keepstub.Handler)
 703         self.thread = threading.Thread(target=self.server.serve_forever)
 704         self.thread.daemon = True # Exit thread if main proc exits
 705         self.thread.start()
 706         self.api_client = self.mock_keep_services(
 707             count=1,
 708             service_host='localhost',
 709             service_port=self.port,
 710         )
 711
 712     def tearDown(self):
 713         self.server.shutdown()
 714
 715     def keepClient(self, timeouts=(0.1, TIMEOUT_TIME, BANDWIDTH_LOW_LIM)):
 716         return arvados.KeepClient(
 717             api_client=self.api_client,
 718             timeout=timeouts)
 719
 720     def test_timeout_slow_connect(self):
 721         # Can't simulate TCP delays with our own socket. Leave our
 722         # stub server running uselessly, and try to connect to an
 723         # unroutable IP address instead.
 724         self.api_client = self.mock_keep_services(
 725             count=1,
 726             service_host='240.0.0.0',
 727         )
 728         with self.assertTakesBetween(0.1, 0.5):
 729             with self.assertRaises(arvados.errors.KeepWriteError):
 730                 self.keepClient().put(self.DATA, copies=1, num_retries=0)
 731
 732     def test_low_bandwidth_no_delays_success(self):
 733         self.server.setbandwidth(2*self.BANDWIDTH_LOW_LIM)
 734         kc = self.keepClient()
 735         loc = kc.put(self.DATA, copies=1, num_retries=0)
 736         self.assertEqual(self.DATA, kc.get(loc, num_retries=0))
 737
 738     def test_too_low_bandwidth_no_delays_failure(self):
 739         # Check that lessening bandwidth corresponds to failing
 740         kc = self.keepClient()
 741         loc = kc.put(self.DATA, copies=1, num_retries=0)
 742         self.server.setbandwidth(0.5*self.BANDWIDTH_LOW_LIM)
 743         with self.assertTakesGreater(self.TIMEOUT_TIME):
 744             with self.assertRaises(arvados.errors.KeepReadError) as e:
 745                 kc.get(loc, num_retries=0)
 746         with self.assertTakesGreater(self.TIMEOUT_TIME):
 747             with self.assertRaises(arvados.errors.KeepWriteError):
 748                 kc.put(self.DATA, copies=1, num_retries=0)
 749
 750     def test_low_bandwidth_with_server_response_delay_failure(self):
 751         kc = self.keepClient()
 752         loc = kc.put(self.DATA, copies=1, num_retries=0)
 753         self.server.setbandwidth(self.BANDWIDTH_LOW_LIM)
 754         self.server.setdelays(response=self.TIMEOUT_TIME)
 755         with self.assertTakesGreater(self.TIMEOUT_TIME):
 756             with self.assertRaises(arvados.errors.KeepReadError) as e:
 757                 kc.get(loc, num_retries=0)
 758         with self.assertTakesGreater(self.TIMEOUT_TIME):
 759             with self.assertRaises(arvados.errors.KeepWriteError):
 760                 kc.put(self.DATA, copies=1, num_retries=0)
 761         with self.assertTakesGreater(self.TIMEOUT_TIME):
 762             with self.assertRaises(arvados.errors.KeepReadError) as e:
 763                 kc.head(loc, num_retries=0)
 764
 765     def test_low_bandwidth_with_server_mid_delay_failure(self):
 766         kc = self.keepClient()
 767         loc = kc.put(self.DATA, copies=1, num_retries=0)
 768         self.server.setbandwidth(self.BANDWIDTH_LOW_LIM)
 769         self.server.setdelays(mid_write=self.TIMEOUT_TIME, mid_read=self.TIMEOUT_TIME)
 770         with self.assertTakesGreater(self.TIMEOUT_TIME):
 771             with self.assertRaises(arvados.errors.KeepReadError) as e:
 772                 kc.get(loc, num_retries=0)
 773         with self.assertTakesGreater(self.TIMEOUT_TIME):
 774             with self.assertRaises(arvados.errors.KeepWriteError):
 775                 kc.put(self.DATA, copies=1, num_retries=0)
 776
 777     def test_timeout_slow_request(self):
 778         loc = self.keepClient().put(self.DATA, copies=1, num_retries=0)
 779         self.server.setdelays(request=.2)
 780         self._test_connect_timeout_under_200ms(loc)
 781         self.server.setdelays(request=2)
 782         self._test_response_timeout_under_2s(loc)
 783
 784     def test_timeout_slow_response(self):
 785         loc = self.keepClient().put(self.DATA, copies=1, num_retries=0)
 786         self.server.setdelays(response=.2)
 787         self._test_connect_timeout_under_200ms(loc)
 788         self.server.setdelays(response=2)
 789         self._test_response_timeout_under_2s(loc)
 790
 791     def test_timeout_slow_response_body(self):
 792         loc = self.keepClient().put(self.DATA, copies=1, num_retries=0)
 793         self.server.setdelays(response_body=.2)
 794         self._test_connect_timeout_under_200ms(loc)
 795         self.server.setdelays(response_body=2)
 796         self._test_response_timeout_under_2s(loc)
 797
 798     def _test_connect_timeout_under_200ms(self, loc):
 799         # Allow 100ms to connect, then 1s for response. Everything
 800         # should work, and everything should take at least 200ms to
 801         # return.
 802         kc = self.keepClient(timeouts=(.1, 1))
 803         with self.assertTakesBetween(.2, .3):
 804             kc.put(self.DATA, copies=1, num_retries=0)
 805         with self.assertTakesBetween(.2, .3):
 806             self.assertEqual(self.DATA, kc.get(loc, num_retries=0))
 807
 808     def _test_response_timeout_under_2s(self, loc):
 809         # Allow 10s to connect, then 1s for response. Nothing should
 810         # work, and everything should take at least 1s to return.
 811         kc = self.keepClient(timeouts=(10, 1))
 812         with self.assertTakesBetween(1, 9):
 813             with self.assertRaises(arvados.errors.KeepReadError):
 814                 kc.get(loc, num_retries=0)
 815         with self.assertTakesBetween(1, 9):
 816             with self.assertRaises(arvados.errors.KeepWriteError):
 817                 kc.put(self.DATA, copies=1, num_retries=0)
 818
 819
 820 class KeepClientGatewayTestCase(unittest.TestCase, tutil.ApiClientMock):
 821     def mock_disks_and_gateways(self, disks=3, gateways=1):
 822         self.gateways = [{
 823                 'uuid': 'zzzzz-bi6l4-gateway{:08d}'.format(i),
 824                 'owner_uuid': 'zzzzz-tpzed-000000000000000',
 825                 'service_host': 'gatewayhost{}'.format(i),
 826                 'service_port': 12345,
 827                 'service_ssl_flag': True,
 828                 'service_type': 'gateway:test',
 829         } for i in range(gateways)]
 830         self.gateway_roots = [
 831             "https://{service_host}:{service_port}/".format(**gw)
 832             for gw in self.gateways]
 833         self.api_client = self.mock_keep_services(
 834             count=disks, additional_services=self.gateways)
 835         self.keepClient = arvados.KeepClient(api_client=self.api_client)
 836
 837     @mock.patch('pycurl.Curl')
 838     def test_get_with_gateway_hint_first(self, MockCurl):
 839         MockCurl.return_value = tutil.FakeCurl.make(
 840             code=200, body='foo', headers={'Content-Length': 3})
 841         self.mock_disks_and_gateways()
 842         locator = 'acbd18db4cc2f85cedef654fccc4a4d8+3+K@' + self.gateways[0]['uuid']
 843         self.assertEqual('foo', self.keepClient.get(locator))
 844         self.assertEqual(self.gateway_roots[0]+locator,
 845                          MockCurl.return_value.getopt(pycurl.URL))
 846         self.assertEqual(True, self.keepClient.head(locator))
 847
 848     @mock.patch('pycurl.Curl')
 849     def test_get_with_gateway_hints_in_order(self, MockCurl):
 850         gateways = 4
 851         disks = 3
 852         mocks = [
 853             tutil.FakeCurl.make(code=404, body='')
 854             for _ in range(gateways+disks)
 855         ]
 856         MockCurl.side_effect = tutil.queue_with(mocks)
 857         self.mock_disks_and_gateways(gateways=gateways, disks=disks)
 858         locator = '+'.join(['acbd18db4cc2f85cedef654fccc4a4d8+3'] +
 859                            ['K@'+gw['uuid'] for gw in self.gateways])
 860         with self.assertRaises(arvados.errors.NotFoundError):
 861             self.keepClient.get(locator)
 862         # Gateways are tried first, in the order given.
 863         for i, root in enumerate(self.gateway_roots):
 864             self.assertEqual(root+locator,
 865                              mocks[i].getopt(pycurl.URL))
 866         # Disk services are tried next.
 867         for i in range(gateways, gateways+disks):
 868             self.assertRegexpMatches(
 869                 mocks[i].getopt(pycurl.URL),
 870                 r'keep0x')
 871
 872     @mock.patch('pycurl.Curl')
 873     def test_head_with_gateway_hints_in_order(self, MockCurl):
 874         gateways = 4
 875         disks = 3
 876         mocks = [
 877             tutil.FakeCurl.make(code=404, body='')
 878             for _ in range(gateways+disks)
 879         ]
 880         MockCurl.side_effect = tutil.queue_with(mocks)
 881         self.mock_disks_and_gateways(gateways=gateways, disks=disks)
 882         locator = '+'.join(['acbd18db4cc2f85cedef654fccc4a4d8+3'] +
 883                            ['K@'+gw['uuid'] for gw in self.gateways])
 884         with self.assertRaises(arvados.errors.NotFoundError):
 885             self.keepClient.head(locator)
 886         # Gateways are tried first, in the order given.
 887         for i, root in enumerate(self.gateway_roots):
 888             self.assertEqual(root+locator,
 889                              mocks[i].getopt(pycurl.URL))
 890         # Disk services are tried next.
 891         for i in range(gateways, gateways+disks):
 892             self.assertRegexpMatches(
 893                 mocks[i].getopt(pycurl.URL),
 894                 r'keep0x')
 895
 896     @mock.patch('pycurl.Curl')
 897     def test_get_with_remote_proxy_hint(self, MockCurl):
 898         MockCurl.return_value = tutil.FakeCurl.make(
 899             code=200, body='foo', headers={'Content-Length': 3})
 900         self.mock_disks_and_gateways()
 901         locator = 'acbd18db4cc2f85cedef654fccc4a4d8+3+K@xyzzy'
 902         self.assertEqual('foo', self.keepClient.get(locator))
 903         self.assertEqual('https://keep.xyzzy.arvadosapi.com/'+locator,
 904                          MockCurl.return_value.getopt(pycurl.URL))
 905
 906     @mock.patch('pycurl.Curl')
 907     def test_head_with_remote_proxy_hint(self, MockCurl):
 908         MockCurl.return_value = tutil.FakeCurl.make(
 909             code=200, body='foo', headers={'Content-Length': 3})
 910         self.mock_disks_and_gateways()
 911         locator = 'acbd18db4cc2f85cedef654fccc4a4d8+3+K@xyzzy'
 912         self.assertEqual(True, self.keepClient.head(locator))
 913         self.assertEqual('https://keep.xyzzy.arvadosapi.com/'+locator,
 914                          MockCurl.return_value.getopt(pycurl.URL))
 915
 916
 917 class KeepClientRetryTestMixin(object):
 918     # Testing with a local Keep store won't exercise the retry behavior.
 919     # Instead, our strategy is:
 920     # * Create a client with one proxy specified (pointed at a black
 921     #   hole), so there's no need to instantiate an API client, and
 922     #   all HTTP requests come from one place.
 923     # * Mock httplib's request method to provide simulated responses.
 924     # This lets us test the retry logic extensively without relying on any
 925     # supporting servers, and prevents side effects in case something hiccups.
 926     # To use this mixin, define DEFAULT_EXPECT, DEFAULT_EXCEPTION, and
 927     # run_method().
 928     #
 929     # Test classes must define TEST_PATCHER to a method that mocks
 930     # out appropriate methods in the client.
 931
 932     PROXY_ADDR = 'http://[%s]:65535/' % (tutil.TEST_HOST,)
 933     TEST_DATA = 'testdata'
 934     TEST_LOCATOR = 'ef654c40ab4f1747fc699915d4f70902+8'
 935
 936     def setUp(self):
 937         self.client_kwargs = {'proxy': self.PROXY_ADDR, 'local_store': ''}
 938
 939     def new_client(self, **caller_kwargs):
 940         kwargs = self.client_kwargs.copy()
 941         kwargs.update(caller_kwargs)
 942         return arvados.KeepClient(**kwargs)
 943
 944     def run_method(self, *args, **kwargs):
 945         raise NotImplementedError("test subclasses must define run_method")
 946
 947     def check_success(self, expected=None, *args, **kwargs):
 948         if expected is None:
 949             expected = self.DEFAULT_EXPECT
 950         self.assertEqual(expected, self.run_method(*args, **kwargs))
 951
 952     def check_exception(self, error_class=None, *args, **kwargs):
 953         if error_class is None:
 954             error_class = self.DEFAULT_EXCEPTION
 955         self.assertRaises(error_class, self.run_method, *args, **kwargs)
 956
 957     def test_immediate_success(self):
 958         with self.TEST_PATCHER(self.DEFAULT_EXPECT, 200):
 959             self.check_success()
 960
 961     def test_retry_then_success(self):
 962         with self.TEST_PATCHER(self.DEFAULT_EXPECT, 500, 200):
 963             self.check_success(num_retries=3)
 964
 965     def test_exception_then_success(self):
 966         with self.TEST_PATCHER(self.DEFAULT_EXPECT, Exception('mock err'), 200):
 967             self.check_success(num_retries=3)
 968
 969     def test_no_default_retry(self):
 970         with self.TEST_PATCHER(self.DEFAULT_EXPECT, 500, 200):
 971             self.check_exception()
 972
 973     def test_no_retry_after_permanent_error(self):
 974         with self.TEST_PATCHER(self.DEFAULT_EXPECT, 403, 200):
 975             self.check_exception(num_retries=3)
 976
 977     def test_error_after_retries_exhausted(self):
 978         with self.TEST_PATCHER(self.DEFAULT_EXPECT, 500, 500, 200):
 979             self.check_exception(num_retries=1)
 980
 981     def test_num_retries_instance_fallback(self):
 982         self.client_kwargs['num_retries'] = 3
 983         with self.TEST_PATCHER(self.DEFAULT_EXPECT, 500, 200):
 984             self.check_success()
 985
 986
 987 @tutil.skip_sleep
 988 class KeepClientRetryGetTestCase(KeepClientRetryTestMixin, unittest.TestCase):
 989     DEFAULT_EXPECT = KeepClientRetryTestMixin.TEST_DATA
 990     DEFAULT_EXCEPTION = arvados.errors.KeepReadError
 991     HINTED_LOCATOR = KeepClientRetryTestMixin.TEST_LOCATOR + '+K@xyzzy'
 992     TEST_PATCHER = staticmethod(tutil.mock_keep_responses)
 993
 994     def run_method(self, locator=KeepClientRetryTestMixin.TEST_LOCATOR,
 995                    *args, **kwargs):
 996         return self.new_client().get(locator, *args, **kwargs)
 997
 998     def test_specific_exception_when_not_found(self):
 999         with tutil.mock_keep_responses(self.DEFAULT_EXPECT, 404, 200):
1000             self.check_exception(arvados.errors.NotFoundError, num_retries=3)
1001
1002     def test_general_exception_with_mixed_errors(self):
1003         # get should raise a NotFoundError if no server returns the block,
1004         # and a high threshold of servers report that it's not found.
1005         # This test rigs up 50/50 disagreement between two servers, and
1006         # checks that it does not become a NotFoundError.
1007         client = self.new_client()
1008         with tutil.mock_keep_responses(self.DEFAULT_EXPECT, 404, 500):
1009             with self.assertRaises(arvados.errors.KeepReadError) as exc_check:
1010                 client.get(self.HINTED_LOCATOR)
1011             self.assertNotIsInstance(
1012                 exc_check.exception, arvados.errors.NotFoundError,
1013                 "mixed errors raised NotFoundError")
1014
1015     def test_hint_server_can_succeed_without_retries(self):
1016         with tutil.mock_keep_responses(self.DEFAULT_EXPECT, 404, 200, 500):
1017             self.check_success(locator=self.HINTED_LOCATOR)
1018
1019     def test_try_next_server_after_timeout(self):
1020         with tutil.mock_keep_responses(
1021                 (socket.timeout("timed out"), 200),
1022                 (self.DEFAULT_EXPECT, 200)):
1023             self.check_success(locator=self.HINTED_LOCATOR)
1024
1025     def test_retry_data_with_wrong_checksum(self):
1026         with tutil.mock_keep_responses(
1027                 ('baddata', 200),
1028                 (self.DEFAULT_EXPECT, 200)):
1029             self.check_success(locator=self.HINTED_LOCATOR)
1030
1031 @tutil.skip_sleep
1032 class KeepClientRetryHeadTestCase(KeepClientRetryTestMixin, unittest.TestCase):
1033     DEFAULT_EXPECT = True
1034     DEFAULT_EXCEPTION = arvados.errors.KeepReadError
1035     HINTED_LOCATOR = KeepClientRetryTestMixin.TEST_LOCATOR + '+K@xyzzy'
1036     TEST_PATCHER = staticmethod(tutil.mock_keep_responses)
1037
1038     def run_method(self, locator=KeepClientRetryTestMixin.TEST_LOCATOR,
1039                    *args, **kwargs):
1040         return self.new_client().head(locator, *args, **kwargs)
1041
1042     def test_specific_exception_when_not_found(self):
1043         with tutil.mock_keep_responses(self.DEFAULT_EXPECT, 404, 200):
1044             self.check_exception(arvados.errors.NotFoundError, num_retries=3)
1045
1046     def test_general_exception_with_mixed_errors(self):
1047         # head should raise a NotFoundError if no server returns the block,
1048         # and a high threshold of servers report that it's not found.
1049         # This test rigs up 50/50 disagreement between two servers, and
1050         # checks that it does not become a NotFoundError.
1051         client = self.new_client()
1052         with tutil.mock_keep_responses(self.DEFAULT_EXPECT, 404, 500):
1053             with self.assertRaises(arvados.errors.KeepReadError) as exc_check:
1054                 client.head(self.HINTED_LOCATOR)
1055             self.assertNotIsInstance(
1056                 exc_check.exception, arvados.errors.NotFoundError,
1057                 "mixed errors raised NotFoundError")
1058
1059     def test_hint_server_can_succeed_without_retries(self):
1060         with tutil.mock_keep_responses(self.DEFAULT_EXPECT, 404, 200, 500):
1061             self.check_success(locator=self.HINTED_LOCATOR)
1062
1063     def test_try_next_server_after_timeout(self):
1064         with tutil.mock_keep_responses(
1065                 (socket.timeout("timed out"), 200),
1066                 (self.DEFAULT_EXPECT, 200)):
1067             self.check_success(locator=self.HINTED_LOCATOR)
1068
1069 @tutil.skip_sleep
1070 class KeepClientRetryPutTestCase(KeepClientRetryTestMixin, unittest.TestCase):
1071     DEFAULT_EXPECT = KeepClientRetryTestMixin.TEST_LOCATOR
1072     DEFAULT_EXCEPTION = arvados.errors.KeepWriteError
1073     TEST_PATCHER = staticmethod(tutil.mock_keep_responses)
1074
1075     def run_method(self, data=KeepClientRetryTestMixin.TEST_DATA,
1076                    copies=1, *args, **kwargs):
1077         return self.new_client().put(data, copies, *args, **kwargs)
1078
1079     def test_do_not_send_multiple_copies_to_same_server(self):
1080         with tutil.mock_keep_responses(self.DEFAULT_EXPECT, 200):
1081             self.check_exception(copies=2, num_retries=3)
1082
1083
1084 class AvoidOverreplication(unittest.TestCase, tutil.ApiClientMock):
1085
1086     class FakeKeepService(object):
1087         def __init__(self, delay, will_succeed=False, will_raise=None, replicas=1):
1088             self.delay = delay
1089             self.will_succeed = will_succeed
1090             self.will_raise = will_raise
1091             self._result = {}
1092             self._result['headers'] = {}
1093             self._result['headers']['x-keep-replicas-stored'] = str(replicas)
1094             self._result['body'] = 'foobar'
1095
1096         def put(self, data_hash, data, timeout):
1097             time.sleep(self.delay)
1098             if self.will_raise is not None:
1099                 raise self.will_raise
1100             return self.will_succeed
1101
1102         def last_result(self):
1103             if self.will_succeed:
1104                 return self._result
1105
1106         def finished(self):
1107             return False
1108
1109     def setUp(self):
1110         self.copies = 3
1111         self.pool = arvados.KeepClient.KeepWriterThreadPool(
1112             data = 'foo',
1113             data_hash = 'acbd18db4cc2f85cedef654fccc4a4d8+3',
1114             max_service_replicas = self.copies,
1115             copies = self.copies
1116         )
1117
1118     def test_only_write_enough_on_success(self):
1119         for i in range(10):
1120             ks = self.FakeKeepService(delay=i/10.0, will_succeed=True)
1121             self.pool.add_task(ks, None)
1122         self.pool.join()
1123         self.assertEqual(self.pool.done(), self.copies)
1124
1125     def test_only_write_enough_on_partial_success(self):
1126         for i in range(5):
1127             ks = self.FakeKeepService(delay=i/10.0, will_succeed=False)
1128             self.pool.add_task(ks, None)
1129             ks = self.FakeKeepService(delay=i/10.0, will_succeed=True)
1130             self.pool.add_task(ks, None)
1131         self.pool.join()
1132         self.assertEqual(self.pool.done(), self.copies)
1133
1134     def test_only_write_enough_when_some_crash(self):
1135         for i in range(5):
1136             ks = self.FakeKeepService(delay=i/10.0, will_raise=Exception())
1137             self.pool.add_task(ks, None)
1138             ks = self.FakeKeepService(delay=i/10.0, will_succeed=True)
1139             self.pool.add_task(ks, None)
1140         self.pool.join()
1141         self.assertEqual(self.pool.done(), self.copies)
1142
1143     def test_fail_when_too_many_crash(self):
1144         for i in range(self.copies+1):
1145             ks = self.FakeKeepService(delay=i/10.0, will_raise=Exception())
1146             self.pool.add_task(ks, None)
1147         for i in range(self.copies-1):
1148             ks = self.FakeKeepService(delay=i/10.0, will_succeed=True)
1149             self.pool.add_task(ks, None)
1150         self.pool.join()
1151         self.assertEqual(self.pool.done(), self.copies-1)
1152
1153
1154 @tutil.skip_sleep
1155 class RetryNeedsMultipleServices(unittest.TestCase, tutil.ApiClientMock):
1156     # Test put()s that need two distinct servers to succeed, possibly
1157     # requiring multiple passes through the retry loop.
1158
1159     def setUp(self):
1160         self.api_client = self.mock_keep_services(count=2)
1161         self.keep_client = arvados.KeepClient(api_client=self.api_client)
1162
1163     def test_success_after_exception(self):
1164         with tutil.mock_keep_responses(
1165                 'acbd18db4cc2f85cedef654fccc4a4d8+3',
1166                 Exception('mock err'), 200, 200) as req_mock:
1167             self.keep_client.put('foo', num_retries=1, copies=2)
1168         self.assertEqual(3, req_mock.call_count)
1169
1170     def test_success_after_retryable_error(self):
1171         with tutil.mock_keep_responses(
1172                 'acbd18db4cc2f85cedef654fccc4a4d8+3',
1173                 500, 200, 200) as req_mock:
1174             self.keep_client.put('foo', num_retries=1, copies=2)
1175         self.assertEqual(3, req_mock.call_count)
1176
1177     def test_fail_after_final_error(self):
1178         # First retry loop gets a 200 (can't achieve replication by
1179         # storing again on that server) and a 400 (can't retry that
1180         # server at all), so we shouldn't try a third request.
1181         with tutil.mock_keep_responses(
1182                 'acbd18db4cc2f85cedef654fccc4a4d8+3',
1183                 200, 400, 200) as req_mock:
1184             with self.assertRaises(arvados.errors.KeepWriteError):
1185                 self.keep_client.put('foo', num_retries=1, copies=2)
1186         self.assertEqual(2, req_mock.call_count)