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         foo_locator = self.keep_client.put('foo')
  33         self.assertRegexpMatches(
  34             foo_locator,
  35             '^acbd18db4cc2f85cedef654fccc4a4d8\+3',
  36             'wrong md5 hash from Keep.put("foo"): ' + foo_locator)
  37         self.assertEqual(self.keep_client.get(foo_locator),
  38                          'foo',
  39                          'wrong content from Keep.get(md5("foo"))')
  40
  41     def test_KeepBinaryRWTest(self):
  42         blob_str = '\xff\xfe\xf7\x00\x01\x02'
  43         blob_locator = self.keep_client.put(blob_str)
  44         self.assertRegexpMatches(
  45             blob_locator,
  46             '^7fc7c53b45e53926ba52821140fef396\+6',
  47             ('wrong locator from Keep.put(<binarydata>):' + blob_locator))
  48         self.assertEqual(self.keep_client.get(blob_locator),
  49                          blob_str,
  50                          'wrong content from Keep.get(md5(<binarydata>))')
  51
  52     def test_KeepLongBinaryRWTest(self):
  53         blob_str = '\xff\xfe\xfd\xfc\x00\x01\x02\x03'
  54         for i in range(0,23):
  55             blob_str = blob_str + blob_str
  56         blob_locator = self.keep_client.put(blob_str)
  57         self.assertRegexpMatches(
  58             blob_locator,
  59             '^84d90fc0d8175dd5dcfab04b999bc956\+67108864',
  60             ('wrong locator from Keep.put(<binarydata>): ' + blob_locator))
  61         self.assertEqual(self.keep_client.get(blob_locator),
  62                          blob_str,
  63                          'wrong content from Keep.get(md5(<binarydata>))')
  64
  65     def test_KeepSingleCopyRWTest(self):
  66         blob_str = '\xff\xfe\xfd\xfc\x00\x01\x02\x03'
  67         blob_locator = self.keep_client.put(blob_str, copies=1)
  68         self.assertRegexpMatches(
  69             blob_locator,
  70             '^c902006bc98a3eb4a3663b65ab4a6fab\+8',
  71             ('wrong locator from Keep.put(<binarydata>): ' + blob_locator))
  72         self.assertEqual(self.keep_client.get(blob_locator),
  73                          blob_str,
  74                          'wrong content from Keep.get(md5(<binarydata>))')
  75
  76     def test_KeepEmptyCollectionTest(self):
  77         blob_locator = self.keep_client.put('', copies=1)
  78         self.assertRegexpMatches(
  79             blob_locator,
  80             '^d41d8cd98f00b204e9800998ecf8427e\+0',
  81             ('wrong locator from Keep.put(""): ' + blob_locator))
  82
  83     def test_unicode_must_be_ascii(self):
  84         # If unicode type, must only consist of valid ASCII
  85         foo_locator = self.keep_client.put(u'foo')
  86         self.assertRegexpMatches(
  87             foo_locator,
  88             '^acbd18db4cc2f85cedef654fccc4a4d8\+3',
  89             'wrong md5 hash from Keep.put("foo"): ' + foo_locator)
  90
  91         with self.assertRaises(UnicodeEncodeError):
  92             # Error if it is not ASCII
  93             self.keep_client.put(u'\xe2')
  94
  95         with self.assertRaises(arvados.errors.ArgumentError):
  96             # Must be a string type
  97             self.keep_client.put({})
  98
  99 class KeepPermissionTestCase(run_test_server.TestCaseWithServers):
 100     MAIN_SERVER = {}
 101     KEEP_SERVER = {'blob_signing_key': 'abcdefghijk0123456789',
 102                    'enforce_permissions': True}
 103
 104     def test_KeepBasicRWTest(self):
 105         run_test_server.authorize_with('active')
 106         keep_client = arvados.KeepClient()
 107         foo_locator = keep_client.put('foo')
 108         self.assertRegexpMatches(
 109             foo_locator,
 110             r'^acbd18db4cc2f85cedef654fccc4a4d8\+3\+A[a-f0-9]+@[a-f0-9]+$',
 111             'invalid locator from Keep.put("foo"): ' + foo_locator)
 112         self.assertEqual(keep_client.get(foo_locator),
 113                          'foo',
 114                          'wrong content from Keep.get(md5("foo"))')
 115
 116         # GET with an unsigned locator => NotFound
 117         bar_locator = keep_client.put('bar')
 118         unsigned_bar_locator = "37b51d194a7513e45b56f6524f2d51f2+3"
 119         self.assertRegexpMatches(
 120             bar_locator,
 121             r'^37b51d194a7513e45b56f6524f2d51f2\+3\+A[a-f0-9]+@[a-f0-9]+$',
 122             'invalid locator from Keep.put("bar"): ' + bar_locator)
 123         self.assertRaises(arvados.errors.NotFoundError,
 124                           keep_client.get,
 125                           unsigned_bar_locator)
 126
 127         # GET from a different user => NotFound
 128         run_test_server.authorize_with('spectator')
 129         self.assertRaises(arvados.errors.NotFoundError,
 130                           arvados.Keep.get,
 131                           bar_locator)
 132
 133         # Unauthenticated GET for a signed locator => NotFound
 134         # Unauthenticated GET for an unsigned locator => NotFound
 135         keep_client.api_token = ''
 136         self.assertRaises(arvados.errors.NotFoundError,
 137                           keep_client.get,
 138                           bar_locator)
 139         self.assertRaises(arvados.errors.NotFoundError,
 140                           keep_client.get,
 141                           unsigned_bar_locator)
 142
 143
 144 # KeepOptionalPermission: starts Keep with --permission-key-file
 145 # but not --enforce-permissions (i.e. generate signatures on PUT
 146 # requests, but do not require them for GET requests)
 147 #
 148 # All of these requests should succeed when permissions are optional:
 149 # * authenticated request, signed locator
 150 # * authenticated request, unsigned locator
 151 # * unauthenticated request, signed locator
 152 # * unauthenticated request, unsigned locator
 153 class KeepOptionalPermission(run_test_server.TestCaseWithServers):
 154     MAIN_SERVER = {}
 155     KEEP_SERVER = {'blob_signing_key': 'abcdefghijk0123456789',
 156                    'enforce_permissions': False}
 157
 158     @classmethod
 159     def setUpClass(cls):
 160         super(KeepOptionalPermission, cls).setUpClass()
 161         run_test_server.authorize_with("admin")
 162         cls.api_client = arvados.api('v1')
 163
 164     def setUp(self):
 165         super(KeepOptionalPermission, self).setUp()
 166         self.keep_client = arvados.KeepClient(api_client=self.api_client,
 167                                               proxy='', local_store='')
 168
 169     def _put_foo_and_check(self):
 170         signed_locator = self.keep_client.put('foo')
 171         self.assertRegexpMatches(
 172             signed_locator,
 173             r'^acbd18db4cc2f85cedef654fccc4a4d8\+3\+A[a-f0-9]+@[a-f0-9]+$',
 174             'invalid locator from Keep.put("foo"): ' + signed_locator)
 175         return signed_locator
 176
 177     def test_KeepAuthenticatedSignedTest(self):
 178         signed_locator = self._put_foo_and_check()
 179         self.assertEqual(self.keep_client.get(signed_locator),
 180                          'foo',
 181                          'wrong content from Keep.get(md5("foo"))')
 182
 183     def test_KeepAuthenticatedUnsignedTest(self):
 184         signed_locator = self._put_foo_and_check()
 185         self.assertEqual(self.keep_client.get("acbd18db4cc2f85cedef654fccc4a4d8"),
 186                          'foo',
 187                          'wrong content from Keep.get(md5("foo"))')
 188
 189     def test_KeepUnauthenticatedSignedTest(self):
 190         # Check that signed GET requests work even when permissions
 191         # enforcement is off.
 192         signed_locator = self._put_foo_and_check()
 193         self.keep_client.api_token = ''
 194         self.assertEqual(self.keep_client.get(signed_locator),
 195                          'foo',
 196                          'wrong content from Keep.get(md5("foo"))')
 197
 198     def test_KeepUnauthenticatedUnsignedTest(self):
 199         # Since --enforce-permissions is not in effect, GET requests
 200         # need not be authenticated.
 201         signed_locator = self._put_foo_and_check()
 202         self.keep_client.api_token = ''
 203         self.assertEqual(self.keep_client.get("acbd18db4cc2f85cedef654fccc4a4d8"),
 204                          'foo',
 205                          'wrong content from Keep.get(md5("foo"))')
 206
 207
 208 class KeepProxyTestCase(run_test_server.TestCaseWithServers):
 209     MAIN_SERVER = {}
 210     KEEP_SERVER = {}
 211     KEEP_PROXY_SERVER = {}
 212
 213     @classmethod
 214     def setUpClass(cls):
 215         super(KeepProxyTestCase, cls).setUpClass()
 216         run_test_server.authorize_with('active')
 217         cls.api_client = arvados.api('v1')
 218
 219     def tearDown(self):
 220         arvados.config.settings().pop('ARVADOS_EXTERNAL_CLIENT', None)
 221         super(KeepProxyTestCase, self).tearDown()
 222
 223     def test_KeepProxyTest1(self):
 224         # Will use ARVADOS_KEEP_PROXY environment variable that is set by
 225         # setUpClass().
 226         keep_client = arvados.KeepClient(api_client=self.api_client,
 227                                          local_store='')
 228         baz_locator = keep_client.put('baz')
 229         self.assertRegexpMatches(
 230             baz_locator,
 231             '^73feffa4b7f6bb68e44cf984c85f6e88\+3',
 232             'wrong md5 hash from Keep.put("baz"): ' + baz_locator)
 233         self.assertEqual(keep_client.get(baz_locator),
 234                          'baz',
 235                          'wrong content from Keep.get(md5("baz"))')
 236         self.assertTrue(keep_client.using_proxy)
 237
 238     def test_KeepProxyTest2(self):
 239         # Don't instantiate the proxy directly, but set the X-External-Client
 240         # header.  The API server should direct us to the proxy.
 241         arvados.config.settings()['ARVADOS_EXTERNAL_CLIENT'] = 'true'
 242         keep_client = arvados.KeepClient(api_client=self.api_client,
 243                                          proxy='', local_store='')
 244         baz_locator = keep_client.put('baz2')
 245         self.assertRegexpMatches(
 246             baz_locator,
 247             '^91f372a266fe2bf2823cb8ec7fda31ce\+4',
 248             'wrong md5 hash from Keep.put("baz2"): ' + baz_locator)
 249         self.assertEqual(keep_client.get(baz_locator),
 250                          'baz2',
 251                          'wrong content from Keep.get(md5("baz2"))')
 252         self.assertTrue(keep_client.using_proxy)
 253
 254
 255 class KeepClientServiceTestCase(unittest.TestCase, tutil.ApiClientMock):
 256     def get_service_roots(self, api_client):
 257         keep_client = arvados.KeepClient(api_client=api_client)
 258         services = keep_client.weighted_service_roots(arvados.KeepLocator('0'*32))
 259         return [urlparse.urlparse(url) for url in sorted(services)]
 260
 261     def test_ssl_flag_respected_in_roots(self):
 262         for ssl_flag in [False, True]:
 263             services = self.get_service_roots(self.mock_keep_services(
 264                 service_ssl_flag=ssl_flag))
 265             self.assertEqual(
 266                 ('https' if ssl_flag else 'http'), services[0].scheme)
 267
 268     def test_correct_ports_with_ipv6_addresses(self):
 269         service = self.get_service_roots(self.mock_keep_services(
 270             service_type='proxy', service_host='100::1', service_port=10, count=1))[0]
 271         self.assertEqual('100::1', service.hostname)
 272         self.assertEqual(10, service.port)
 273
 274     # test_*_timeout verify that KeepClient instructs pycurl to use
 275     # the appropriate connection and read timeouts. They don't care
 276     # whether pycurl actually exhibits the expected timeout behavior
 277     # -- those tests are in the KeepClientTimeout test class.
 278
 279     def test_get_timeout(self):
 280         api_client = self.mock_keep_services(count=1)
 281         force_timeout = socket.timeout("timed out")
 282         with tutil.mock_keep_responses(force_timeout, 0) as mock:
 283             keep_client = arvados.KeepClient(api_client=api_client)
 284             with self.assertRaises(arvados.errors.KeepReadError):
 285                 keep_client.get('ffffffffffffffffffffffffffffffff')
 286             self.assertEqual(
 287                 mock.responses[0].getopt(pycurl.CONNECTTIMEOUT_MS),
 288                 int(arvados.KeepClient.DEFAULT_TIMEOUT[0]*1000))
 289             self.assertEqual(
 290                 mock.responses[0].getopt(pycurl.TIMEOUT_MS),
 291                 int(arvados.KeepClient.DEFAULT_TIMEOUT[1]*1000))
 292
 293     def test_put_timeout(self):
 294         api_client = self.mock_keep_services(count=1)
 295         force_timeout = socket.timeout("timed out")
 296         with tutil.mock_keep_responses(force_timeout, 0) as mock:
 297             keep_client = arvados.KeepClient(api_client=api_client)
 298             with self.assertRaises(arvados.errors.KeepWriteError):
 299                 keep_client.put('foo')
 300             self.assertEqual(
 301                 mock.responses[0].getopt(pycurl.CONNECTTIMEOUT_MS),
 302                 int(arvados.KeepClient.DEFAULT_TIMEOUT[0]*1000))
 303             self.assertEqual(
 304                 mock.responses[0].getopt(pycurl.TIMEOUT_MS),
 305                 int(arvados.KeepClient.DEFAULT_TIMEOUT[1]*1000))
 306
 307     def test_proxy_get_timeout(self):
 308         api_client = self.mock_keep_services(service_type='proxy', count=1)
 309         force_timeout = socket.timeout("timed out")
 310         with tutil.mock_keep_responses(force_timeout, 0) as mock:
 311             keep_client = arvados.KeepClient(api_client=api_client)
 312             with self.assertRaises(arvados.errors.KeepReadError):
 313                 keep_client.get('ffffffffffffffffffffffffffffffff')
 314             self.assertEqual(
 315                 mock.responses[0].getopt(pycurl.CONNECTTIMEOUT_MS),
 316                 int(arvados.KeepClient.DEFAULT_PROXY_TIMEOUT[0]*1000))
 317             self.assertEqual(
 318                 mock.responses[0].getopt(pycurl.TIMEOUT_MS),
 319                 int(arvados.KeepClient.DEFAULT_PROXY_TIMEOUT[1]*1000))
 320
 321     def test_proxy_put_timeout(self):
 322         api_client = self.mock_keep_services(service_type='proxy', count=1)
 323         force_timeout = socket.timeout("timed out")
 324         with tutil.mock_keep_responses(force_timeout, 0) as mock:
 325             keep_client = arvados.KeepClient(api_client=api_client)
 326             with self.assertRaises(arvados.errors.KeepWriteError):
 327                 keep_client.put('foo')
 328             self.assertEqual(
 329                 mock.responses[0].getopt(pycurl.CONNECTTIMEOUT_MS),
 330                 int(arvados.KeepClient.DEFAULT_PROXY_TIMEOUT[0]*1000))
 331             self.assertEqual(
 332                 mock.responses[0].getopt(pycurl.TIMEOUT_MS),
 333                 int(arvados.KeepClient.DEFAULT_PROXY_TIMEOUT[1]*1000))
 334
 335     def check_no_services_error(self, verb, exc_class):
 336         api_client = mock.MagicMock(name='api_client')
 337         api_client.keep_services().accessible().execute.side_effect = (
 338             arvados.errors.ApiError)
 339         keep_client = arvados.KeepClient(api_client=api_client)
 340         with self.assertRaises(exc_class) as err_check:
 341             getattr(keep_client, verb)('d41d8cd98f00b204e9800998ecf8427e+0')
 342         self.assertEqual(0, len(err_check.exception.request_errors()))
 343
 344     def test_get_error_with_no_services(self):
 345         self.check_no_services_error('get', arvados.errors.KeepReadError)
 346
 347     def test_put_error_with_no_services(self):
 348         self.check_no_services_error('put', arvados.errors.KeepWriteError)
 349
 350     def check_errors_from_last_retry(self, verb, exc_class):
 351         api_client = self.mock_keep_services(count=2)
 352         req_mock = tutil.mock_keep_responses(
 353             "retry error reporting test", 500, 500, 403, 403)
 354         with req_mock, tutil.skip_sleep, \
 355                 self.assertRaises(exc_class) as err_check:
 356             keep_client = arvados.KeepClient(api_client=api_client)
 357             getattr(keep_client, verb)('d41d8cd98f00b204e9800998ecf8427e+0',
 358                                        num_retries=3)
 359         self.assertEqual([403, 403], [
 360                 getattr(error, 'status_code', None)
 361                 for error in err_check.exception.request_errors().itervalues()])
 362
 363     def test_get_error_reflects_last_retry(self):
 364         self.check_errors_from_last_retry('get', arvados.errors.KeepReadError)
 365
 366     def test_put_error_reflects_last_retry(self):
 367         self.check_errors_from_last_retry('put', arvados.errors.KeepWriteError)
 368
 369     def test_put_error_does_not_include_successful_puts(self):
 370         data = 'partial failure test'
 371         data_loc = tutil.str_keep_locator(data)
 372         api_client = self.mock_keep_services(count=3)
 373         with tutil.mock_keep_responses(data_loc, 200, 500, 500) as req_mock, \
 374                 self.assertRaises(arvados.errors.KeepWriteError) as exc_check:
 375             keep_client = arvados.KeepClient(api_client=api_client)
 376             keep_client.put(data)
 377         self.assertEqual(2, len(exc_check.exception.request_errors()))
 378
 379     def test_proxy_put_with_no_writable_services(self):
 380         data = 'test with no writable services'
 381         data_loc = tutil.str_keep_locator(data)
 382         api_client = self.mock_keep_services(service_type='proxy', read_only=True, count=1)
 383         with tutil.mock_keep_responses(data_loc, 200, 500, 500) as req_mock, \
 384                 self.assertRaises(arvados.errors.KeepWriteError) as exc_check:
 385           keep_client = arvados.KeepClient(api_client=api_client)
 386           keep_client.put(data)
 387         self.assertEqual(True, ("no Keep services available" in str(exc_check.exception)))
 388         self.assertEqual(0, len(exc_check.exception.request_errors()))
 389
 390     def test_oddball_service_get(self):
 391         body = 'oddball service get'
 392         api_client = self.mock_keep_services(service_type='fancynewblobstore')
 393         with tutil.mock_keep_responses(body, 200):
 394             keep_client = arvados.KeepClient(api_client=api_client)
 395             actual = keep_client.get(tutil.str_keep_locator(body))
 396         self.assertEqual(body, actual)
 397
 398     def test_oddball_service_put(self):
 399         body = 'oddball service put'
 400         pdh = tutil.str_keep_locator(body)
 401         api_client = self.mock_keep_services(service_type='fancynewblobstore')
 402         with tutil.mock_keep_responses(pdh, 200):
 403             keep_client = arvados.KeepClient(api_client=api_client)
 404             actual = keep_client.put(body, copies=1)
 405         self.assertEqual(pdh, actual)
 406
 407     def test_oddball_service_writer_count(self):
 408         body = 'oddball service writer count'
 409         pdh = tutil.str_keep_locator(body)
 410         api_client = self.mock_keep_services(service_type='fancynewblobstore',
 411                                              count=4)
 412         headers = {'x-keep-replicas-stored': 3}
 413         with tutil.mock_keep_responses(pdh, 200, 418, 418, 418,
 414                                        **headers) as req_mock:
 415             keep_client = arvados.KeepClient(api_client=api_client)
 416             actual = keep_client.put(body, copies=2)
 417         self.assertEqual(pdh, actual)
 418         self.assertEqual(1, req_mock.call_count)
 419
 420
 421 @tutil.skip_sleep
 422 class KeepClientRendezvousTestCase(unittest.TestCase, tutil.ApiClientMock):
 423
 424     def setUp(self):
 425         # expected_order[i] is the probe order for
 426         # hash=md5(sprintf("%064x",i)) where there are 16 services
 427         # with uuid sprintf("anything-%015x",j) with j in 0..15. E.g.,
 428         # the first probe for the block consisting of 64 "0"
 429         # characters is the service whose uuid is
 430         # "zzzzz-bi6l4-000000000000003", so expected_order[0][0]=='3'.
 431         self.services = 16
 432         self.expected_order = [
 433             list('3eab2d5fc9681074'),
 434             list('097dba52e648f1c3'),
 435             list('c5b4e023f8a7d691'),
 436             list('9d81c02e76a3bf54'),
 437             ]
 438         self.blocks = [
 439             "{:064x}".format(x)
 440             for x in range(len(self.expected_order))]
 441         self.hashes = [
 442             hashlib.md5(self.blocks[x]).hexdigest()
 443             for x in range(len(self.expected_order))]
 444         self.api_client = self.mock_keep_services(count=self.services)
 445         self.keep_client = arvados.KeepClient(api_client=self.api_client)
 446
 447     def test_weighted_service_roots_against_reference_set(self):
 448         # Confirm weighted_service_roots() returns the correct order
 449         for i, hash in enumerate(self.hashes):
 450             roots = self.keep_client.weighted_service_roots(arvados.KeepLocator(hash))
 451             got_order = [
 452                 re.search(r'//\[?keep0x([0-9a-f]+)', root).group(1)
 453                 for root in roots]
 454             self.assertEqual(self.expected_order[i], got_order)
 455
 456     def test_get_probe_order_against_reference_set(self):
 457         self._test_probe_order_against_reference_set(
 458             lambda i: self.keep_client.get(self.hashes[i], num_retries=1))
 459
 460     def test_put_probe_order_against_reference_set(self):
 461         # copies=1 prevents the test from being sensitive to races
 462         # between writer threads.
 463         self._test_probe_order_against_reference_set(
 464             lambda i: self.keep_client.put(self.blocks[i], num_retries=1, copies=1))
 465
 466     def _test_probe_order_against_reference_set(self, op):
 467         for i in range(len(self.blocks)):
 468             with tutil.mock_keep_responses('', *[500 for _ in range(self.services*2)]) as mock, \
 469                  self.assertRaises(arvados.errors.KeepRequestError):
 470                 op(i)
 471             got_order = [
 472                 re.search(r'//\[?keep0x([0-9a-f]+)', resp.getopt(pycurl.URL)).group(1)
 473                 for resp in mock.responses]
 474             self.assertEqual(self.expected_order[i]*2, got_order)
 475
 476     def test_put_probe_order_multiple_copies(self):
 477         for copies in range(2, 4):
 478             for i in range(len(self.blocks)):
 479                 with tutil.mock_keep_responses('', *[500 for _ in range(self.services*3)]) as mock, \
 480                      self.assertRaises(arvados.errors.KeepWriteError):
 481                     self.keep_client.put(self.blocks[i], num_retries=2, copies=copies)
 482                 got_order = [
 483                     re.search(r'//\[?keep0x([0-9a-f]+)', resp.getopt(pycurl.URL)).group(1)
 484                     for resp in mock.responses]
 485                 # With T threads racing to make requests, the position
 486                 # of a given server in the sequence of HTTP requests
 487                 # (got_order) cannot be more than T-1 positions
 488                 # earlier than that server's position in the reference
 489                 # probe sequence (expected_order).
 490                 #
 491                 # Loop invariant: we have accounted for +pos+ expected
 492                 # probes, either by seeing them in +got_order+ or by
 493                 # putting them in +pending+ in the hope of seeing them
 494                 # later. As long as +len(pending)<T+, we haven't
 495                 # started a request too early.
 496                 pending = []
 497                 for pos, expected in enumerate(self.expected_order[i]*3):
 498                     got = got_order[pos-len(pending)]
 499                     while got in pending:
 500                         del pending[pending.index(got)]
 501                         got = got_order[pos-len(pending)]
 502                     if got != expected:
 503                         pending.append(expected)
 504                         self.assertLess(
 505                             len(pending), copies,
 506                             "pending={}, with copies={}, got {}, expected {}".format(
 507                                 pending, copies, repr(got_order), repr(self.expected_order[i]*3)))
 508
 509     def test_probe_waste_adding_one_server(self):
 510         hashes = [
 511             hashlib.md5("{:064x}".format(x)).hexdigest() for x in range(100)]
 512         initial_services = 12
 513         self.api_client = self.mock_keep_services(count=initial_services)
 514         self.keep_client = arvados.KeepClient(api_client=self.api_client)
 515         probes_before = [
 516             self.keep_client.weighted_service_roots(arvados.KeepLocator(hash)) for hash in hashes]
 517         for added_services in range(1, 12):
 518             api_client = self.mock_keep_services(count=initial_services+added_services)
 519             keep_client = arvados.KeepClient(api_client=api_client)
 520             total_penalty = 0
 521             for hash_index in range(len(hashes)):
 522                 probe_after = keep_client.weighted_service_roots(
 523                     arvados.KeepLocator(hashes[hash_index]))
 524                 penalty = probe_after.index(probes_before[hash_index][0])
 525                 self.assertLessEqual(penalty, added_services)
 526                 total_penalty += penalty
 527             # Average penalty per block should not exceed
 528             # N(added)/N(orig) by more than 20%, and should get closer
 529             # to the ideal as we add data points.
 530             expect_penalty = (
 531                 added_services *
 532                 len(hashes) / initial_services)
 533             max_penalty = (
 534                 expect_penalty *
 535                 (120 - added_services)/100)
 536             min_penalty = (
 537                 expect_penalty * 8/10)
 538             self.assertTrue(
 539                 min_penalty <= total_penalty <= max_penalty,
 540                 "With {}+{} services, {} blocks, penalty {} but expected {}..{}".format(
 541                     initial_services,
 542                     added_services,
 543                     len(hashes),
 544                     total_penalty,
 545                     min_penalty,
 546                     max_penalty))
 547
 548     def check_64_zeros_error_order(self, verb, exc_class):
 549         data = '0' * 64
 550         if verb == 'get':
 551             data = tutil.str_keep_locator(data)
 552         # Arbitrary port number:
 553         aport = random.randint(1024,65535)
 554         api_client = self.mock_keep_services(service_port=aport, count=self.services)
 555         keep_client = arvados.KeepClient(api_client=api_client)
 556         with mock.patch('pycurl.Curl') as curl_mock, \
 557              self.assertRaises(exc_class) as err_check:
 558             curl_mock.return_value.side_effect = socket.timeout
 559             getattr(keep_client, verb)(data)
 560         urls = [urlparse.urlparse(url)
 561                 for url in err_check.exception.request_errors()]
 562         self.assertEqual([('keep0x' + c, aport) for c in '3eab2d5fc9681074'],
 563                          [(url.hostname, url.port) for url in urls])
 564
 565     def test_get_error_shows_probe_order(self):
 566         self.check_64_zeros_error_order('get', arvados.errors.KeepReadError)
 567
 568     def test_put_error_shows_probe_order(self):
 569         self.check_64_zeros_error_order('put', arvados.errors.KeepWriteError)
 570
 571
 572 class KeepClientTimeout(unittest.TestCase, tutil.ApiClientMock):
 573     DATA = 'x' * 2**10
 574
 575     class assertTakesBetween(unittest.TestCase):
 576         def __init__(self, tmin, tmax):
 577             self.tmin = tmin
 578             self.tmax = tmax
 579
 580         def __enter__(self):
 581             self.t0 = time.time()
 582
 583         def __exit__(self, *args, **kwargs):
 584             self.assertGreater(time.time() - self.t0, self.tmin)
 585             self.assertLess(time.time() - self.t0, self.tmax)
 586
 587     def setUp(self):
 588         sock = socket.socket()
 589         sock.bind(('0.0.0.0', 0))
 590         self.port = sock.getsockname()[1]
 591         sock.close()
 592         self.server = keepstub.Server(('0.0.0.0', self.port), keepstub.Handler)
 593         self.thread = threading.Thread(target=self.server.serve_forever)
 594         self.thread.daemon = True # Exit thread if main proc exits
 595         self.thread.start()
 596         self.api_client = self.mock_keep_services(
 597             count=1,
 598             service_host='localhost',
 599             service_port=self.port,
 600         )
 601
 602     def tearDown(self):
 603         self.server.shutdown()
 604
 605     def keepClient(self, timeouts=(0.1, 1.0)):
 606         return arvados.KeepClient(
 607             api_client=self.api_client,
 608             timeout=timeouts)
 609
 610     def test_timeout_slow_connect(self):
 611         # Can't simulate TCP delays with our own socket. Leave our
 612         # stub server running uselessly, and try to connect to an
 613         # unroutable IP address instead.
 614         self.api_client = self.mock_keep_services(
 615             count=1,
 616             service_host='240.0.0.0',
 617         )
 618         with self.assertTakesBetween(0.1, 0.5):
 619             with self.assertRaises(arvados.errors.KeepWriteError):
 620                 self.keepClient((0.1, 1)).put(self.DATA, copies=1, num_retries=0)
 621
 622     def test_timeout_slow_request(self):
 623         self.server.setdelays(request=0.2)
 624         self._test_200ms()
 625
 626     def test_timeout_slow_response(self):
 627         self.server.setdelays(response=0.2)
 628         self._test_200ms()
 629
 630     def test_timeout_slow_response_body(self):
 631         self.server.setdelays(response_body=0.2)
 632         self._test_200ms()
 633
 634     def _test_200ms(self):
 635         """Connect should be t<100ms, request should be 200ms <= t < 300ms"""
 636
 637         # Allow 100ms to connect, then 1s for response. Everything
 638         # should work, and everything should take at least 200ms to
 639         # return.
 640         kc = self.keepClient((.1, 1))
 641         with self.assertTakesBetween(.2, .3):
 642             loc = kc.put(self.DATA, copies=1, num_retries=0)
 643         with self.assertTakesBetween(.2, .3):
 644             self.assertEqual(self.DATA, kc.get(loc, num_retries=0))
 645
 646         # Allow 1s to connect, then 100ms for response. Nothing should
 647         # work, and everything should take at least 100ms to return.
 648         kc = self.keepClient((1, .1))
 649         with self.assertTakesBetween(.1, .2):
 650             with self.assertRaises(arvados.errors.KeepReadError):
 651                 kc.get(loc, num_retries=0)
 652         with self.assertTakesBetween(.1, .2):
 653             with self.assertRaises(arvados.errors.KeepWriteError):
 654                 kc.put(self.DATA, copies=1, num_retries=0)
 655
 656
 657 class KeepClientGatewayTestCase(unittest.TestCase, tutil.ApiClientMock):
 658     def mock_disks_and_gateways(self, disks=3, gateways=1):
 659         self.gateways = [{
 660                 'uuid': 'zzzzz-bi6l4-gateway{:08d}'.format(i),
 661                 'owner_uuid': 'zzzzz-tpzed-000000000000000',
 662                 'service_host': 'gatewayhost{}'.format(i),
 663                 'service_port': 12345,
 664                 'service_ssl_flag': True,
 665                 'service_type': 'gateway:test',
 666         } for i in range(gateways)]
 667         self.gateway_roots = [
 668             "https://{service_host}:{service_port}/".format(**gw)
 669             for gw in self.gateways]
 670         self.api_client = self.mock_keep_services(
 671             count=disks, additional_services=self.gateways)
 672         self.keepClient = arvados.KeepClient(api_client=self.api_client)
 673
 674     @mock.patch('pycurl.Curl')
 675     def test_get_with_gateway_hint_first(self, MockCurl):
 676         MockCurl.return_value = tutil.FakeCurl.make(
 677             code=200, body='foo', headers={'Content-Length': 3})
 678         self.mock_disks_and_gateways()
 679         locator = 'acbd18db4cc2f85cedef654fccc4a4d8+3+K@' + self.gateways[0]['uuid']
 680         self.assertEqual('foo', self.keepClient.get(locator))
 681         self.assertEqual(self.gateway_roots[0]+locator,
 682                          MockCurl.return_value.getopt(pycurl.URL))
 683
 684     @mock.patch('pycurl.Curl')
 685     def test_get_with_gateway_hints_in_order(self, MockCurl):
 686         gateways = 4
 687         disks = 3
 688         mocks = [
 689             tutil.FakeCurl.make(code=404, body='')
 690             for _ in range(gateways+disks)
 691         ]
 692         MockCurl.side_effect = tutil.queue_with(mocks)
 693         self.mock_disks_and_gateways(gateways=gateways, disks=disks)
 694         locator = '+'.join(['acbd18db4cc2f85cedef654fccc4a4d8+3'] +
 695                            ['K@'+gw['uuid'] for gw in self.gateways])
 696         with self.assertRaises(arvados.errors.NotFoundError):
 697             self.keepClient.get(locator)
 698         # Gateways are tried first, in the order given.
 699         for i, root in enumerate(self.gateway_roots):
 700             self.assertEqual(root+locator,
 701                              mocks[i].getopt(pycurl.URL))
 702         # Disk services are tried next.
 703         for i in range(gateways, gateways+disks):
 704             self.assertRegexpMatches(
 705                 mocks[i].getopt(pycurl.URL),
 706                 r'keep0x')
 707
 708     @mock.patch('pycurl.Curl')
 709     def test_get_with_remote_proxy_hint(self, MockCurl):
 710         MockCurl.return_value = tutil.FakeCurl.make(
 711             code=200, body='foo', headers={'Content-Length': 3})
 712         self.mock_disks_and_gateways()
 713         locator = 'acbd18db4cc2f85cedef654fccc4a4d8+3+K@xyzzy'
 714         self.assertEqual('foo', self.keepClient.get(locator))
 715         self.assertEqual('https://keep.xyzzy.arvadosapi.com/'+locator,
 716                          MockCurl.return_value.getopt(pycurl.URL))
 717
 718
 719 class KeepClientRetryTestMixin(object):
 720     # Testing with a local Keep store won't exercise the retry behavior.
 721     # Instead, our strategy is:
 722     # * Create a client with one proxy specified (pointed at a black
 723     #   hole), so there's no need to instantiate an API client, and
 724     #   all HTTP requests come from one place.
 725     # * Mock httplib's request method to provide simulated responses.
 726     # This lets us test the retry logic extensively without relying on any
 727     # supporting servers, and prevents side effects in case something hiccups.
 728     # To use this mixin, define DEFAULT_EXPECT, DEFAULT_EXCEPTION, and
 729     # run_method().
 730     #
 731     # Test classes must define TEST_PATCHER to a method that mocks
 732     # out appropriate methods in the client.
 733
 734     PROXY_ADDR = 'http://[%s]:65535/' % (tutil.TEST_HOST,)
 735     TEST_DATA = 'testdata'
 736     TEST_LOCATOR = 'ef654c40ab4f1747fc699915d4f70902+8'
 737
 738     def setUp(self):
 739         self.client_kwargs = {'proxy': self.PROXY_ADDR, 'local_store': ''}
 740
 741     def new_client(self, **caller_kwargs):
 742         kwargs = self.client_kwargs.copy()
 743         kwargs.update(caller_kwargs)
 744         return arvados.KeepClient(**kwargs)
 745
 746     def run_method(self, *args, **kwargs):
 747         raise NotImplementedError("test subclasses must define run_method")
 748
 749     def check_success(self, expected=None, *args, **kwargs):
 750         if expected is None:
 751             expected = self.DEFAULT_EXPECT
 752         self.assertEqual(expected, self.run_method(*args, **kwargs))
 753
 754     def check_exception(self, error_class=None, *args, **kwargs):
 755         if error_class is None:
 756             error_class = self.DEFAULT_EXCEPTION
 757         self.assertRaises(error_class, self.run_method, *args, **kwargs)
 758
 759     def test_immediate_success(self):
 760         with self.TEST_PATCHER(self.DEFAULT_EXPECT, 200):
 761             self.check_success()
 762
 763     def test_retry_then_success(self):
 764         with self.TEST_PATCHER(self.DEFAULT_EXPECT, 500, 200):
 765             self.check_success(num_retries=3)
 766
 767     def test_no_default_retry(self):
 768         with self.TEST_PATCHER(self.DEFAULT_EXPECT, 500, 200):
 769             self.check_exception()
 770
 771     def test_no_retry_after_permanent_error(self):
 772         with self.TEST_PATCHER(self.DEFAULT_EXPECT, 403, 200):
 773             self.check_exception(num_retries=3)
 774
 775     def test_error_after_retries_exhausted(self):
 776         with self.TEST_PATCHER(self.DEFAULT_EXPECT, 500, 500, 200):
 777             self.check_exception(num_retries=1)
 778
 779     def test_num_retries_instance_fallback(self):
 780         self.client_kwargs['num_retries'] = 3
 781         with self.TEST_PATCHER(self.DEFAULT_EXPECT, 500, 200):
 782             self.check_success()
 783
 784
 785 @tutil.skip_sleep
 786 class KeepClientRetryGetTestCase(KeepClientRetryTestMixin, unittest.TestCase):
 787     DEFAULT_EXPECT = KeepClientRetryTestMixin.TEST_DATA
 788     DEFAULT_EXCEPTION = arvados.errors.KeepReadError
 789     HINTED_LOCATOR = KeepClientRetryTestMixin.TEST_LOCATOR + '+K@xyzzy'
 790     TEST_PATCHER = staticmethod(tutil.mock_keep_responses)
 791
 792     def run_method(self, locator=KeepClientRetryTestMixin.TEST_LOCATOR,
 793                    *args, **kwargs):
 794         return self.new_client().get(locator, *args, **kwargs)
 795
 796     def test_specific_exception_when_not_found(self):
 797         with tutil.mock_keep_responses(self.DEFAULT_EXPECT, 404, 200):
 798             self.check_exception(arvados.errors.NotFoundError, num_retries=3)
 799
 800     def test_general_exception_with_mixed_errors(self):
 801         # get should raise a NotFoundError if no server returns the block,
 802         # and a high threshold of servers report that it's not found.
 803         # This test rigs up 50/50 disagreement between two servers, and
 804         # checks that it does not become a NotFoundError.
 805         client = self.new_client()
 806         with tutil.mock_keep_responses(self.DEFAULT_EXPECT, 404, 500):
 807             with self.assertRaises(arvados.errors.KeepReadError) as exc_check:
 808                 client.get(self.HINTED_LOCATOR)
 809             self.assertNotIsInstance(
 810                 exc_check.exception, arvados.errors.NotFoundError,
 811                 "mixed errors raised NotFoundError")
 812
 813     def test_hint_server_can_succeed_without_retries(self):
 814         with tutil.mock_keep_responses(self.DEFAULT_EXPECT, 404, 200, 500):
 815             self.check_success(locator=self.HINTED_LOCATOR)
 816
 817     def test_try_next_server_after_timeout(self):
 818         with tutil.mock_keep_responses(
 819                 (socket.timeout("timed out"), 200),
 820                 (self.DEFAULT_EXPECT, 200)):
 821             self.check_success(locator=self.HINTED_LOCATOR)
 822
 823     def test_retry_data_with_wrong_checksum(self):
 824         with tutil.mock_keep_responses(
 825                 ('baddata', 200),
 826                 (self.DEFAULT_EXPECT, 200)):
 827             self.check_success(locator=self.HINTED_LOCATOR)
 828
 829
 830 @tutil.skip_sleep
 831 class KeepClientRetryPutTestCase(KeepClientRetryTestMixin, unittest.TestCase):
 832     DEFAULT_EXPECT = KeepClientRetryTestMixin.TEST_LOCATOR
 833     DEFAULT_EXCEPTION = arvados.errors.KeepWriteError
 834     TEST_PATCHER = staticmethod(tutil.mock_keep_responses)
 835
 836     def run_method(self, data=KeepClientRetryTestMixin.TEST_DATA,
 837                    copies=1, *args, **kwargs):
 838         return self.new_client().put(data, copies, *args, **kwargs)
 839
 840     def test_do_not_send_multiple_copies_to_same_server(self):
 841         with tutil.mock_keep_responses(self.DEFAULT_EXPECT, 200):
 842             self.check_exception(copies=2, num_retries=3)