15 import arvados_testutil as tutil
17 import run_test_server
19 class KeepTestCase(run_test_server.TestCaseWithServers):
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='')
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(
36 '^acbd18db4cc2f85cedef654fccc4a4d8\+3',
37 'wrong md5 hash from Keep.put("foo"): ' + foo_locator)
39 # 6 bytes because uploaded 2 copies
40 self.assertEqual(6, self.keep_client.upload_counter.get())
42 self.assertEqual(0, self.keep_client.download_counter.get())
43 self.assertEqual(self.keep_client.get(foo_locator),
45 'wrong content from Keep.get(md5("foo"))')
46 self.assertEqual(3, self.keep_client.download_counter.get())
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(
53 '^7fc7c53b45e53926ba52821140fef396\+6',
54 ('wrong locator from Keep.put(<binarydata>):' + blob_locator))
55 self.assertEqual(self.keep_client.get(blob_locator),
57 'wrong content from Keep.get(md5(<binarydata>))')
59 def test_KeepLongBinaryRWTest(self):
60 blob_str = '\xff\xfe\xfd\xfc\x00\x01\x02\x03'
62 blob_str = blob_str + blob_str
63 blob_locator = self.keep_client.put(blob_str)
64 self.assertRegexpMatches(
66 '^84d90fc0d8175dd5dcfab04b999bc956\+67108864',
67 ('wrong locator from Keep.put(<binarydata>): ' + blob_locator))
68 self.assertEqual(self.keep_client.get(blob_locator),
70 'wrong content from Keep.get(md5(<binarydata>))')
72 def test_KeepSingleCopyRWTest(self):
73 blob_str = '\xff\xfe\xfd\xfc\x00\x01\x02\x03'
74 blob_locator = self.keep_client.put(blob_str, copies=1)
75 self.assertRegexpMatches(
77 '^c902006bc98a3eb4a3663b65ab4a6fab\+8',
78 ('wrong locator from Keep.put(<binarydata>): ' + blob_locator))
79 self.assertEqual(self.keep_client.get(blob_locator),
81 'wrong content from Keep.get(md5(<binarydata>))')
83 def test_KeepEmptyCollectionTest(self):
84 blob_locator = self.keep_client.put('', copies=1)
85 self.assertRegexpMatches(
87 '^d41d8cd98f00b204e9800998ecf8427e\+0',
88 ('wrong locator from Keep.put(""): ' + blob_locator))
90 def test_unicode_must_be_ascii(self):
91 # If unicode type, must only consist of valid ASCII
92 foo_locator = self.keep_client.put(u'foo')
93 self.assertRegexpMatches(
95 '^acbd18db4cc2f85cedef654fccc4a4d8\+3',
96 'wrong md5 hash from Keep.put("foo"): ' + foo_locator)
98 with self.assertRaises(UnicodeEncodeError):
99 # Error if it is not ASCII
100 self.keep_client.put(u'\xe2')
102 with self.assertRaises(arvados.errors.ArgumentError):
103 # Must be a string type
104 self.keep_client.put({})
106 class KeepPermissionTestCase(run_test_server.TestCaseWithServers):
108 KEEP_SERVER = {'blob_signing_key': 'abcdefghijk0123456789',
109 'enforce_permissions': True}
111 def test_KeepBasicRWTest(self):
112 run_test_server.authorize_with('active')
113 keep_client = arvados.KeepClient()
114 foo_locator = keep_client.put('foo')
115 self.assertRegexpMatches(
117 r'^acbd18db4cc2f85cedef654fccc4a4d8\+3\+A[a-f0-9]+@[a-f0-9]+$',
118 'invalid locator from Keep.put("foo"): ' + foo_locator)
119 self.assertEqual(keep_client.get(foo_locator),
121 'wrong content from Keep.get(md5("foo"))')
123 # GET with an unsigned locator => NotFound
124 bar_locator = keep_client.put('bar')
125 unsigned_bar_locator = "37b51d194a7513e45b56f6524f2d51f2+3"
126 self.assertRegexpMatches(
128 r'^37b51d194a7513e45b56f6524f2d51f2\+3\+A[a-f0-9]+@[a-f0-9]+$',
129 'invalid locator from Keep.put("bar"): ' + bar_locator)
130 self.assertRaises(arvados.errors.NotFoundError,
132 unsigned_bar_locator)
134 # GET from a different user => NotFound
135 run_test_server.authorize_with('spectator')
136 self.assertRaises(arvados.errors.NotFoundError,
140 # Unauthenticated GET for a signed locator => NotFound
141 # Unauthenticated GET for an unsigned locator => NotFound
142 keep_client.api_token = ''
143 self.assertRaises(arvados.errors.NotFoundError,
146 self.assertRaises(arvados.errors.NotFoundError,
148 unsigned_bar_locator)
151 # KeepOptionalPermission: starts Keep with --permission-key-file
152 # but not --enforce-permissions (i.e. generate signatures on PUT
153 # requests, but do not require them for GET requests)
155 # All of these requests should succeed when permissions are optional:
156 # * authenticated request, signed locator
157 # * authenticated request, unsigned locator
158 # * unauthenticated request, signed locator
159 # * unauthenticated request, unsigned locator
160 class KeepOptionalPermission(run_test_server.TestCaseWithServers):
162 KEEP_SERVER = {'blob_signing_key': 'abcdefghijk0123456789',
163 'enforce_permissions': False}
167 super(KeepOptionalPermission, cls).setUpClass()
168 run_test_server.authorize_with("admin")
169 cls.api_client = arvados.api('v1')
172 super(KeepOptionalPermission, self).setUp()
173 self.keep_client = arvados.KeepClient(api_client=self.api_client,
174 proxy='', local_store='')
176 def _put_foo_and_check(self):
177 signed_locator = self.keep_client.put('foo')
178 self.assertRegexpMatches(
180 r'^acbd18db4cc2f85cedef654fccc4a4d8\+3\+A[a-f0-9]+@[a-f0-9]+$',
181 'invalid locator from Keep.put("foo"): ' + signed_locator)
182 return signed_locator
184 def test_KeepAuthenticatedSignedTest(self):
185 signed_locator = self._put_foo_and_check()
186 self.assertEqual(self.keep_client.get(signed_locator),
188 'wrong content from Keep.get(md5("foo"))')
190 def test_KeepAuthenticatedUnsignedTest(self):
191 signed_locator = self._put_foo_and_check()
192 self.assertEqual(self.keep_client.get("acbd18db4cc2f85cedef654fccc4a4d8"),
194 'wrong content from Keep.get(md5("foo"))')
196 def test_KeepUnauthenticatedSignedTest(self):
197 # Check that signed GET requests work even when permissions
198 # enforcement is off.
199 signed_locator = self._put_foo_and_check()
200 self.keep_client.api_token = ''
201 self.assertEqual(self.keep_client.get(signed_locator),
203 'wrong content from Keep.get(md5("foo"))')
205 def test_KeepUnauthenticatedUnsignedTest(self):
206 # Since --enforce-permissions is not in effect, GET requests
207 # need not be authenticated.
208 signed_locator = self._put_foo_and_check()
209 self.keep_client.api_token = ''
210 self.assertEqual(self.keep_client.get("acbd18db4cc2f85cedef654fccc4a4d8"),
212 'wrong content from Keep.get(md5("foo"))')
215 class KeepProxyTestCase(run_test_server.TestCaseWithServers):
218 KEEP_PROXY_SERVER = {}
222 super(KeepProxyTestCase, cls).setUpClass()
223 run_test_server.authorize_with('active')
224 cls.api_client = arvados.api('v1')
227 arvados.config.settings().pop('ARVADOS_EXTERNAL_CLIENT', None)
228 super(KeepProxyTestCase, self).tearDown()
230 def test_KeepProxyTest1(self):
231 # Will use ARVADOS_KEEP_PROXY environment variable that is set by
233 keep_client = arvados.KeepClient(api_client=self.api_client,
235 baz_locator = keep_client.put('baz')
236 self.assertRegexpMatches(
238 '^73feffa4b7f6bb68e44cf984c85f6e88\+3',
239 'wrong md5 hash from Keep.put("baz"): ' + baz_locator)
240 self.assertEqual(keep_client.get(baz_locator),
242 'wrong content from Keep.get(md5("baz"))')
243 self.assertTrue(keep_client.using_proxy)
245 def test_KeepProxyTest2(self):
246 # Don't instantiate the proxy directly, but set the X-External-Client
247 # header. The API server should direct us to the proxy.
248 arvados.config.settings()['ARVADOS_EXTERNAL_CLIENT'] = 'true'
249 keep_client = arvados.KeepClient(api_client=self.api_client,
250 proxy='', local_store='')
251 baz_locator = keep_client.put('baz2')
252 self.assertRegexpMatches(
254 '^91f372a266fe2bf2823cb8ec7fda31ce\+4',
255 'wrong md5 hash from Keep.put("baz2"): ' + baz_locator)
256 self.assertEqual(keep_client.get(baz_locator),
258 'wrong content from Keep.get(md5("baz2"))')
259 self.assertTrue(keep_client.using_proxy)
262 class KeepClientServiceTestCase(unittest.TestCase, tutil.ApiClientMock):
263 def get_service_roots(self, api_client):
264 keep_client = arvados.KeepClient(api_client=api_client)
265 services = keep_client.weighted_service_roots(arvados.KeepLocator('0'*32))
266 return [urlparse.urlparse(url) for url in sorted(services)]
268 def test_ssl_flag_respected_in_roots(self):
269 for ssl_flag in [False, True]:
270 services = self.get_service_roots(self.mock_keep_services(
271 service_ssl_flag=ssl_flag))
273 ('https' if ssl_flag else 'http'), services[0].scheme)
275 def test_correct_ports_with_ipv6_addresses(self):
276 service = self.get_service_roots(self.mock_keep_services(
277 service_type='proxy', service_host='100::1', service_port=10, count=1))[0]
278 self.assertEqual('100::1', service.hostname)
279 self.assertEqual(10, service.port)
281 # test_*_timeout verify that KeepClient instructs pycurl to use
282 # the appropriate connection and read timeouts. They don't care
283 # whether pycurl actually exhibits the expected timeout behavior
284 # -- those tests are in the KeepClientTimeout test class.
286 def test_get_timeout(self):
287 api_client = self.mock_keep_services(count=1)
288 force_timeout = socket.timeout("timed out")
289 with tutil.mock_keep_responses(force_timeout, 0) as mock:
290 keep_client = arvados.KeepClient(api_client=api_client)
291 with self.assertRaises(arvados.errors.KeepReadError):
292 keep_client.get('ffffffffffffffffffffffffffffffff')
294 mock.responses[0].getopt(pycurl.CONNECTTIMEOUT_MS),
295 int(arvados.KeepClient.DEFAULT_TIMEOUT[0]*1000))
297 mock.responses[0].getopt(pycurl.LOW_SPEED_TIME),
298 int(arvados.KeepClient.DEFAULT_TIMEOUT[1]))
300 mock.responses[0].getopt(pycurl.LOW_SPEED_LIMIT),
301 int(arvados.KeepClient.DEFAULT_TIMEOUT[2]))
303 def test_put_timeout(self):
304 api_client = self.mock_keep_services(count=1)
305 force_timeout = socket.timeout("timed out")
306 with tutil.mock_keep_responses(force_timeout, 0) as mock:
307 keep_client = arvados.KeepClient(api_client=api_client)
308 with self.assertRaises(arvados.errors.KeepWriteError):
309 keep_client.put('foo')
311 mock.responses[0].getopt(pycurl.CONNECTTIMEOUT_MS),
312 int(arvados.KeepClient.DEFAULT_TIMEOUT[0]*1000))
314 mock.responses[0].getopt(pycurl.LOW_SPEED_TIME),
315 int(arvados.KeepClient.DEFAULT_TIMEOUT[1]))
317 mock.responses[0].getopt(pycurl.LOW_SPEED_LIMIT),
318 int(arvados.KeepClient.DEFAULT_TIMEOUT[2]))
320 def test_proxy_get_timeout(self):
321 api_client = self.mock_keep_services(service_type='proxy', count=1)
322 force_timeout = socket.timeout("timed out")
323 with tutil.mock_keep_responses(force_timeout, 0) as mock:
324 keep_client = arvados.KeepClient(api_client=api_client)
325 with self.assertRaises(arvados.errors.KeepReadError):
326 keep_client.get('ffffffffffffffffffffffffffffffff')
328 mock.responses[0].getopt(pycurl.CONNECTTIMEOUT_MS),
329 int(arvados.KeepClient.DEFAULT_PROXY_TIMEOUT[0]*1000))
331 mock.responses[0].getopt(pycurl.LOW_SPEED_TIME),
332 int(arvados.KeepClient.DEFAULT_PROXY_TIMEOUT[1]))
334 mock.responses[0].getopt(pycurl.LOW_SPEED_LIMIT),
335 int(arvados.KeepClient.DEFAULT_PROXY_TIMEOUT[2]))
337 def test_proxy_put_timeout(self):
338 api_client = self.mock_keep_services(service_type='proxy', count=1)
339 force_timeout = socket.timeout("timed out")
340 with tutil.mock_keep_responses(force_timeout, 0) as mock:
341 keep_client = arvados.KeepClient(api_client=api_client)
342 with self.assertRaises(arvados.errors.KeepWriteError):
343 keep_client.put('foo')
345 mock.responses[0].getopt(pycurl.CONNECTTIMEOUT_MS),
346 int(arvados.KeepClient.DEFAULT_PROXY_TIMEOUT[0]*1000))
348 mock.responses[0].getopt(pycurl.LOW_SPEED_TIME),
349 int(arvados.KeepClient.DEFAULT_PROXY_TIMEOUT[1]))
351 mock.responses[0].getopt(pycurl.LOW_SPEED_LIMIT),
352 int(arvados.KeepClient.DEFAULT_PROXY_TIMEOUT[2]))
354 def check_no_services_error(self, verb, exc_class):
355 api_client = mock.MagicMock(name='api_client')
356 api_client.keep_services().accessible().execute.side_effect = (
357 arvados.errors.ApiError)
358 keep_client = arvados.KeepClient(api_client=api_client)
359 with self.assertRaises(exc_class) as err_check:
360 getattr(keep_client, verb)('d41d8cd98f00b204e9800998ecf8427e+0')
361 self.assertEqual(0, len(err_check.exception.request_errors()))
363 def test_get_error_with_no_services(self):
364 self.check_no_services_error('get', arvados.errors.KeepReadError)
366 def test_put_error_with_no_services(self):
367 self.check_no_services_error('put', arvados.errors.KeepWriteError)
369 def check_errors_from_last_retry(self, verb, exc_class):
370 api_client = self.mock_keep_services(count=2)
371 req_mock = tutil.mock_keep_responses(
372 "retry error reporting test", 500, 500, 403, 403)
373 with req_mock, tutil.skip_sleep, \
374 self.assertRaises(exc_class) as err_check:
375 keep_client = arvados.KeepClient(api_client=api_client)
376 getattr(keep_client, verb)('d41d8cd98f00b204e9800998ecf8427e+0',
378 self.assertEqual([403, 403], [
379 getattr(error, 'status_code', None)
380 for error in err_check.exception.request_errors().itervalues()])
382 def test_get_error_reflects_last_retry(self):
383 self.check_errors_from_last_retry('get', arvados.errors.KeepReadError)
385 def test_put_error_reflects_last_retry(self):
386 self.check_errors_from_last_retry('put', arvados.errors.KeepWriteError)
388 def test_put_error_does_not_include_successful_puts(self):
389 data = 'partial failure test'
390 data_loc = tutil.str_keep_locator(data)
391 api_client = self.mock_keep_services(count=3)
392 with tutil.mock_keep_responses(data_loc, 200, 500, 500) as req_mock, \
393 self.assertRaises(arvados.errors.KeepWriteError) as exc_check:
394 keep_client = arvados.KeepClient(api_client=api_client)
395 keep_client.put(data)
396 self.assertEqual(2, len(exc_check.exception.request_errors()))
398 def test_proxy_put_with_no_writable_services(self):
399 data = 'test with no writable services'
400 data_loc = tutil.str_keep_locator(data)
401 api_client = self.mock_keep_services(service_type='proxy', read_only=True, count=1)
402 with tutil.mock_keep_responses(data_loc, 200, 500, 500) as req_mock, \
403 self.assertRaises(arvados.errors.KeepWriteError) as exc_check:
404 keep_client = arvados.KeepClient(api_client=api_client)
405 keep_client.put(data)
406 self.assertEqual(True, ("no Keep services available" in str(exc_check.exception)))
407 self.assertEqual(0, len(exc_check.exception.request_errors()))
409 def test_oddball_service_get(self):
410 body = 'oddball service get'
411 api_client = self.mock_keep_services(service_type='fancynewblobstore')
412 with tutil.mock_keep_responses(body, 200):
413 keep_client = arvados.KeepClient(api_client=api_client)
414 actual = keep_client.get(tutil.str_keep_locator(body))
415 self.assertEqual(body, actual)
417 def test_oddball_service_put(self):
418 body = 'oddball service put'
419 pdh = tutil.str_keep_locator(body)
420 api_client = self.mock_keep_services(service_type='fancynewblobstore')
421 with tutil.mock_keep_responses(pdh, 200):
422 keep_client = arvados.KeepClient(api_client=api_client)
423 actual = keep_client.put(body, copies=1)
424 self.assertEqual(pdh, actual)
426 def test_oddball_service_writer_count(self):
427 body = 'oddball service writer count'
428 pdh = tutil.str_keep_locator(body)
429 api_client = self.mock_keep_services(service_type='fancynewblobstore',
431 headers = {'x-keep-replicas-stored': 3}
432 with tutil.mock_keep_responses(pdh, 200, 418, 418, 418,
433 **headers) as req_mock:
434 keep_client = arvados.KeepClient(api_client=api_client)
435 actual = keep_client.put(body, copies=2)
436 self.assertEqual(pdh, actual)
437 self.assertEqual(1, req_mock.call_count)
441 class KeepClientRendezvousTestCase(unittest.TestCase, tutil.ApiClientMock):
444 # expected_order[i] is the probe order for
445 # hash=md5(sprintf("%064x",i)) where there are 16 services
446 # with uuid sprintf("anything-%015x",j) with j in 0..15. E.g.,
447 # the first probe for the block consisting of 64 "0"
448 # characters is the service whose uuid is
449 # "zzzzz-bi6l4-000000000000003", so expected_order[0][0]=='3'.
451 self.expected_order = [
452 list('3eab2d5fc9681074'),
453 list('097dba52e648f1c3'),
454 list('c5b4e023f8a7d691'),
455 list('9d81c02e76a3bf54'),
459 for x in range(len(self.expected_order))]
461 hashlib.md5(self.blocks[x]).hexdigest()
462 for x in range(len(self.expected_order))]
463 self.api_client = self.mock_keep_services(count=self.services)
464 self.keep_client = arvados.KeepClient(api_client=self.api_client)
466 def test_weighted_service_roots_against_reference_set(self):
467 # Confirm weighted_service_roots() returns the correct order
468 for i, hash in enumerate(self.hashes):
469 roots = self.keep_client.weighted_service_roots(arvados.KeepLocator(hash))
471 re.search(r'//\[?keep0x([0-9a-f]+)', root).group(1)
473 self.assertEqual(self.expected_order[i], got_order)
475 def test_get_probe_order_against_reference_set(self):
476 self._test_probe_order_against_reference_set(
477 lambda i: self.keep_client.get(self.hashes[i], num_retries=1))
479 def test_put_probe_order_against_reference_set(self):
480 # copies=1 prevents the test from being sensitive to races
481 # between writer threads.
482 self._test_probe_order_against_reference_set(
483 lambda i: self.keep_client.put(self.blocks[i], num_retries=1, copies=1))
485 def _test_probe_order_against_reference_set(self, op):
486 for i in range(len(self.blocks)):
487 with tutil.mock_keep_responses('', *[500 for _ in range(self.services*2)]) as mock, \
488 self.assertRaises(arvados.errors.KeepRequestError):
491 re.search(r'//\[?keep0x([0-9a-f]+)', resp.getopt(pycurl.URL)).group(1)
492 for resp in mock.responses]
493 self.assertEqual(self.expected_order[i]*2, got_order)
495 def test_put_probe_order_multiple_copies(self):
496 for copies in range(2, 4):
497 for i in range(len(self.blocks)):
498 with tutil.mock_keep_responses('', *[500 for _ in range(self.services*3)]) as mock, \
499 self.assertRaises(arvados.errors.KeepWriteError):
500 self.keep_client.put(self.blocks[i], num_retries=2, copies=copies)
502 re.search(r'//\[?keep0x([0-9a-f]+)', resp.getopt(pycurl.URL)).group(1)
503 for resp in mock.responses]
504 # With T threads racing to make requests, the position
505 # of a given server in the sequence of HTTP requests
506 # (got_order) cannot be more than T-1 positions
507 # earlier than that server's position in the reference
508 # probe sequence (expected_order).
510 # Loop invariant: we have accounted for +pos+ expected
511 # probes, either by seeing them in +got_order+ or by
512 # putting them in +pending+ in the hope of seeing them
513 # later. As long as +len(pending)<T+, we haven't
514 # started a request too early.
516 for pos, expected in enumerate(self.expected_order[i]*3):
517 got = got_order[pos-len(pending)]
518 while got in pending:
519 del pending[pending.index(got)]
520 got = got_order[pos-len(pending)]
522 pending.append(expected)
524 len(pending), copies,
525 "pending={}, with copies={}, got {}, expected {}".format(
526 pending, copies, repr(got_order), repr(self.expected_order[i]*3)))
528 def test_probe_waste_adding_one_server(self):
530 hashlib.md5("{:064x}".format(x)).hexdigest() for x in range(100)]
531 initial_services = 12
532 self.api_client = self.mock_keep_services(count=initial_services)
533 self.keep_client = arvados.KeepClient(api_client=self.api_client)
535 self.keep_client.weighted_service_roots(arvados.KeepLocator(hash)) for hash in hashes]
536 for added_services in range(1, 12):
537 api_client = self.mock_keep_services(count=initial_services+added_services)
538 keep_client = arvados.KeepClient(api_client=api_client)
540 for hash_index in range(len(hashes)):
541 probe_after = keep_client.weighted_service_roots(
542 arvados.KeepLocator(hashes[hash_index]))
543 penalty = probe_after.index(probes_before[hash_index][0])
544 self.assertLessEqual(penalty, added_services)
545 total_penalty += penalty
546 # Average penalty per block should not exceed
547 # N(added)/N(orig) by more than 20%, and should get closer
548 # to the ideal as we add data points.
551 len(hashes) / initial_services)
554 (120 - added_services)/100)
556 expect_penalty * 8/10)
558 min_penalty <= total_penalty <= max_penalty,
559 "With {}+{} services, {} blocks, penalty {} but expected {}..{}".format(
567 def check_64_zeros_error_order(self, verb, exc_class):
570 data = tutil.str_keep_locator(data)
571 # Arbitrary port number:
572 aport = random.randint(1024,65535)
573 api_client = self.mock_keep_services(service_port=aport, count=self.services)
574 keep_client = arvados.KeepClient(api_client=api_client)
575 with mock.patch('pycurl.Curl') as curl_mock, \
576 self.assertRaises(exc_class) as err_check:
577 curl_mock.return_value.side_effect = socket.timeout
578 getattr(keep_client, verb)(data)
579 urls = [urlparse.urlparse(url)
580 for url in err_check.exception.request_errors()]
581 self.assertEqual([('keep0x' + c, aport) for c in '3eab2d5fc9681074'],
582 [(url.hostname, url.port) for url in urls])
584 def test_get_error_shows_probe_order(self):
585 self.check_64_zeros_error_order('get', arvados.errors.KeepReadError)
587 def test_put_error_shows_probe_order(self):
588 self.check_64_zeros_error_order('put', arvados.errors.KeepWriteError)
591 class KeepClientTimeout(unittest.TestCase, tutil.ApiClientMock):
592 # BANDWIDTH_LOW_LIM must be less than len(DATA) so we can transfer
593 # 1s worth of data and then trigger bandwidth errors before running
596 BANDWIDTH_LOW_LIM = 1024
599 class assertTakesBetween(unittest.TestCase):
600 def __init__(self, tmin, tmax):
605 self.t0 = time.time()
607 def __exit__(self, *args, **kwargs):
608 # Round times to milliseconds, like CURL. Otherwise, we
609 # fail when CURL reaches a 1s timeout at 0.9998s.
610 delta = round(time.time() - self.t0, 3)
611 self.assertGreaterEqual(delta, self.tmin)
612 self.assertLessEqual(delta, self.tmax)
614 class assertTakesGreater(unittest.TestCase):
615 def __init__(self, tmin):
619 self.t0 = time.time()
621 def __exit__(self, *args, **kwargs):
622 delta = round(time.time() - self.t0, 3)
623 self.assertGreaterEqual(delta, self.tmin)
626 sock = socket.socket()
627 sock.bind(('0.0.0.0', 0))
628 self.port = sock.getsockname()[1]
630 self.server = keepstub.Server(('0.0.0.0', self.port), keepstub.Handler)
631 self.thread = threading.Thread(target=self.server.serve_forever)
632 self.thread.daemon = True # Exit thread if main proc exits
634 self.api_client = self.mock_keep_services(
636 service_host='localhost',
637 service_port=self.port,
641 self.server.shutdown()
643 def keepClient(self, timeouts=(0.1, TIMEOUT_TIME, BANDWIDTH_LOW_LIM)):
644 return arvados.KeepClient(
645 api_client=self.api_client,
648 def test_timeout_slow_connect(self):
649 # Can't simulate TCP delays with our own socket. Leave our
650 # stub server running uselessly, and try to connect to an
651 # unroutable IP address instead.
652 self.api_client = self.mock_keep_services(
654 service_host='240.0.0.0',
656 with self.assertTakesBetween(0.1, 0.5):
657 with self.assertRaises(arvados.errors.KeepWriteError):
658 self.keepClient().put(self.DATA, copies=1, num_retries=0)
660 def test_low_bandwidth_no_delays_success(self):
661 self.server.setbandwidth(2*self.BANDWIDTH_LOW_LIM)
662 kc = self.keepClient()
663 loc = kc.put(self.DATA, copies=1, num_retries=0)
664 self.assertEqual(self.DATA, kc.get(loc, num_retries=0))
666 def test_too_low_bandwidth_no_delays_failure(self):
667 # Check that lessening bandwidth corresponds to failing
668 kc = self.keepClient()
669 loc = kc.put(self.DATA, copies=1, num_retries=0)
670 self.server.setbandwidth(0.5*self.BANDWIDTH_LOW_LIM)
671 with self.assertTakesGreater(self.TIMEOUT_TIME):
672 with self.assertRaises(arvados.errors.KeepReadError) as e:
673 kc.get(loc, num_retries=0)
674 with self.assertTakesGreater(self.TIMEOUT_TIME):
675 with self.assertRaises(arvados.errors.KeepWriteError):
676 kc.put(self.DATA, copies=1, num_retries=0)
678 def test_low_bandwidth_with_server_response_delay_failure(self):
679 kc = self.keepClient()
680 loc = kc.put(self.DATA, copies=1, num_retries=0)
681 self.server.setbandwidth(self.BANDWIDTH_LOW_LIM)
682 self.server.setdelays(response=self.TIMEOUT_TIME)
683 with self.assertTakesGreater(self.TIMEOUT_TIME):
684 with self.assertRaises(arvados.errors.KeepReadError) as e:
685 kc.get(loc, num_retries=0)
686 with self.assertTakesGreater(self.TIMEOUT_TIME):
687 with self.assertRaises(arvados.errors.KeepWriteError):
688 kc.put(self.DATA, copies=1, num_retries=0)
690 def test_low_bandwidth_with_server_mid_delay_failure(self):
691 kc = self.keepClient()
692 loc = kc.put(self.DATA, copies=1, num_retries=0)
693 self.server.setbandwidth(self.BANDWIDTH_LOW_LIM)
694 self.server.setdelays(mid_write=self.TIMEOUT_TIME, mid_read=self.TIMEOUT_TIME)
695 with self.assertTakesGreater(self.TIMEOUT_TIME):
696 with self.assertRaises(arvados.errors.KeepReadError) as e:
697 kc.get(loc, num_retries=0)
698 with self.assertTakesGreater(self.TIMEOUT_TIME):
699 with self.assertRaises(arvados.errors.KeepWriteError):
700 kc.put(self.DATA, copies=1, num_retries=0)
702 def test_timeout_slow_request(self):
703 loc = self.keepClient().put(self.DATA, copies=1, num_retries=0)
704 self.server.setdelays(request=.2)
705 self._test_connect_timeout_under_200ms(loc)
706 self.server.setdelays(request=2)
707 self._test_response_timeout_under_2s(loc)
709 def test_timeout_slow_response(self):
710 loc = self.keepClient().put(self.DATA, copies=1, num_retries=0)
711 self.server.setdelays(response=.2)
712 self._test_connect_timeout_under_200ms(loc)
713 self.server.setdelays(response=2)
714 self._test_response_timeout_under_2s(loc)
716 def test_timeout_slow_response_body(self):
717 loc = self.keepClient().put(self.DATA, copies=1, num_retries=0)
718 self.server.setdelays(response_body=.2)
719 self._test_connect_timeout_under_200ms(loc)
720 self.server.setdelays(response_body=2)
721 self._test_response_timeout_under_2s(loc)
723 def _test_connect_timeout_under_200ms(self, loc):
724 # Allow 100ms to connect, then 1s for response. Everything
725 # should work, and everything should take at least 200ms to
727 kc = self.keepClient(timeouts=(.1, 1))
728 with self.assertTakesBetween(.2, .3):
729 kc.put(self.DATA, copies=1, num_retries=0)
730 with self.assertTakesBetween(.2, .3):
731 self.assertEqual(self.DATA, kc.get(loc, num_retries=0))
733 def _test_response_timeout_under_2s(self, loc):
734 # Allow 10s to connect, then 1s for response. Nothing should
735 # work, and everything should take at least 1s to return.
736 kc = self.keepClient(timeouts=(10, 1))
737 with self.assertTakesBetween(1, 1.9):
738 with self.assertRaises(arvados.errors.KeepReadError):
739 kc.get(loc, num_retries=0)
740 with self.assertTakesBetween(1, 1.9):
741 with self.assertRaises(arvados.errors.KeepWriteError):
742 kc.put(self.DATA, copies=1, num_retries=0)
745 class KeepClientGatewayTestCase(unittest.TestCase, tutil.ApiClientMock):
746 def mock_disks_and_gateways(self, disks=3, gateways=1):
748 'uuid': 'zzzzz-bi6l4-gateway{:08d}'.format(i),
749 'owner_uuid': 'zzzzz-tpzed-000000000000000',
750 'service_host': 'gatewayhost{}'.format(i),
751 'service_port': 12345,
752 'service_ssl_flag': True,
753 'service_type': 'gateway:test',
754 } for i in range(gateways)]
755 self.gateway_roots = [
756 "https://{service_host}:{service_port}/".format(**gw)
757 for gw in self.gateways]
758 self.api_client = self.mock_keep_services(
759 count=disks, additional_services=self.gateways)
760 self.keepClient = arvados.KeepClient(api_client=self.api_client)
762 @mock.patch('pycurl.Curl')
763 def test_get_with_gateway_hint_first(self, MockCurl):
764 MockCurl.return_value = tutil.FakeCurl.make(
765 code=200, body='foo', headers={'Content-Length': 3})
766 self.mock_disks_and_gateways()
767 locator = 'acbd18db4cc2f85cedef654fccc4a4d8+3+K@' + self.gateways[0]['uuid']
768 self.assertEqual('foo', self.keepClient.get(locator))
769 self.assertEqual(self.gateway_roots[0]+locator,
770 MockCurl.return_value.getopt(pycurl.URL))
772 @mock.patch('pycurl.Curl')
773 def test_get_with_gateway_hints_in_order(self, MockCurl):
777 tutil.FakeCurl.make(code=404, body='')
778 for _ in range(gateways+disks)
780 MockCurl.side_effect = tutil.queue_with(mocks)
781 self.mock_disks_and_gateways(gateways=gateways, disks=disks)
782 locator = '+'.join(['acbd18db4cc2f85cedef654fccc4a4d8+3'] +
783 ['K@'+gw['uuid'] for gw in self.gateways])
784 with self.assertRaises(arvados.errors.NotFoundError):
785 self.keepClient.get(locator)
786 # Gateways are tried first, in the order given.
787 for i, root in enumerate(self.gateway_roots):
788 self.assertEqual(root+locator,
789 mocks[i].getopt(pycurl.URL))
790 # Disk services are tried next.
791 for i in range(gateways, gateways+disks):
792 self.assertRegexpMatches(
793 mocks[i].getopt(pycurl.URL),
796 @mock.patch('pycurl.Curl')
797 def test_get_with_remote_proxy_hint(self, MockCurl):
798 MockCurl.return_value = tutil.FakeCurl.make(
799 code=200, body='foo', headers={'Content-Length': 3})
800 self.mock_disks_and_gateways()
801 locator = 'acbd18db4cc2f85cedef654fccc4a4d8+3+K@xyzzy'
802 self.assertEqual('foo', self.keepClient.get(locator))
803 self.assertEqual('https://keep.xyzzy.arvadosapi.com/'+locator,
804 MockCurl.return_value.getopt(pycurl.URL))
807 class KeepClientRetryTestMixin(object):
808 # Testing with a local Keep store won't exercise the retry behavior.
809 # Instead, our strategy is:
810 # * Create a client with one proxy specified (pointed at a black
811 # hole), so there's no need to instantiate an API client, and
812 # all HTTP requests come from one place.
813 # * Mock httplib's request method to provide simulated responses.
814 # This lets us test the retry logic extensively without relying on any
815 # supporting servers, and prevents side effects in case something hiccups.
816 # To use this mixin, define DEFAULT_EXPECT, DEFAULT_EXCEPTION, and
819 # Test classes must define TEST_PATCHER to a method that mocks
820 # out appropriate methods in the client.
822 PROXY_ADDR = 'http://[%s]:65535/' % (tutil.TEST_HOST,)
823 TEST_DATA = 'testdata'
824 TEST_LOCATOR = 'ef654c40ab4f1747fc699915d4f70902+8'
827 self.client_kwargs = {'proxy': self.PROXY_ADDR, 'local_store': ''}
829 def new_client(self, **caller_kwargs):
830 kwargs = self.client_kwargs.copy()
831 kwargs.update(caller_kwargs)
832 return arvados.KeepClient(**kwargs)
834 def run_method(self, *args, **kwargs):
835 raise NotImplementedError("test subclasses must define run_method")
837 def check_success(self, expected=None, *args, **kwargs):
839 expected = self.DEFAULT_EXPECT
840 self.assertEqual(expected, self.run_method(*args, **kwargs))
842 def check_exception(self, error_class=None, *args, **kwargs):
843 if error_class is None:
844 error_class = self.DEFAULT_EXCEPTION
845 self.assertRaises(error_class, self.run_method, *args, **kwargs)
847 def test_immediate_success(self):
848 with self.TEST_PATCHER(self.DEFAULT_EXPECT, 200):
851 def test_retry_then_success(self):
852 with self.TEST_PATCHER(self.DEFAULT_EXPECT, 500, 200):
853 self.check_success(num_retries=3)
855 def test_exception_then_success(self):
856 with self.TEST_PATCHER(self.DEFAULT_EXPECT, Exception('mock err'), 200):
857 self.check_success(num_retries=3)
859 def test_no_default_retry(self):
860 with self.TEST_PATCHER(self.DEFAULT_EXPECT, 500, 200):
861 self.check_exception()
863 def test_no_retry_after_permanent_error(self):
864 with self.TEST_PATCHER(self.DEFAULT_EXPECT, 403, 200):
865 self.check_exception(num_retries=3)
867 def test_error_after_retries_exhausted(self):
868 with self.TEST_PATCHER(self.DEFAULT_EXPECT, 500, 500, 200):
869 self.check_exception(num_retries=1)
871 def test_num_retries_instance_fallback(self):
872 self.client_kwargs['num_retries'] = 3
873 with self.TEST_PATCHER(self.DEFAULT_EXPECT, 500, 200):
878 class KeepClientRetryGetTestCase(KeepClientRetryTestMixin, unittest.TestCase):
879 DEFAULT_EXPECT = KeepClientRetryTestMixin.TEST_DATA
880 DEFAULT_EXCEPTION = arvados.errors.KeepReadError
881 HINTED_LOCATOR = KeepClientRetryTestMixin.TEST_LOCATOR + '+K@xyzzy'
882 TEST_PATCHER = staticmethod(tutil.mock_keep_responses)
884 def run_method(self, locator=KeepClientRetryTestMixin.TEST_LOCATOR,
886 return self.new_client().get(locator, *args, **kwargs)
888 def test_specific_exception_when_not_found(self):
889 with tutil.mock_keep_responses(self.DEFAULT_EXPECT, 404, 200):
890 self.check_exception(arvados.errors.NotFoundError, num_retries=3)
892 def test_general_exception_with_mixed_errors(self):
893 # get should raise a NotFoundError if no server returns the block,
894 # and a high threshold of servers report that it's not found.
895 # This test rigs up 50/50 disagreement between two servers, and
896 # checks that it does not become a NotFoundError.
897 client = self.new_client()
898 with tutil.mock_keep_responses(self.DEFAULT_EXPECT, 404, 500):
899 with self.assertRaises(arvados.errors.KeepReadError) as exc_check:
900 client.get(self.HINTED_LOCATOR)
901 self.assertNotIsInstance(
902 exc_check.exception, arvados.errors.NotFoundError,
903 "mixed errors raised NotFoundError")
905 def test_hint_server_can_succeed_without_retries(self):
906 with tutil.mock_keep_responses(self.DEFAULT_EXPECT, 404, 200, 500):
907 self.check_success(locator=self.HINTED_LOCATOR)
909 def test_try_next_server_after_timeout(self):
910 with tutil.mock_keep_responses(
911 (socket.timeout("timed out"), 200),
912 (self.DEFAULT_EXPECT, 200)):
913 self.check_success(locator=self.HINTED_LOCATOR)
915 def test_retry_data_with_wrong_checksum(self):
916 with tutil.mock_keep_responses(
918 (self.DEFAULT_EXPECT, 200)):
919 self.check_success(locator=self.HINTED_LOCATOR)
923 class KeepClientRetryPutTestCase(KeepClientRetryTestMixin, unittest.TestCase):
924 DEFAULT_EXPECT = KeepClientRetryTestMixin.TEST_LOCATOR
925 DEFAULT_EXCEPTION = arvados.errors.KeepWriteError
926 TEST_PATCHER = staticmethod(tutil.mock_keep_responses)
928 def run_method(self, data=KeepClientRetryTestMixin.TEST_DATA,
929 copies=1, *args, **kwargs):
930 return self.new_client().put(data, copies, *args, **kwargs)
932 def test_do_not_send_multiple_copies_to_same_server(self):
933 with tutil.mock_keep_responses(self.DEFAULT_EXPECT, 200):
934 self.check_exception(copies=2, num_retries=3)
938 class RetryNeedsMultipleServices(unittest.TestCase, tutil.ApiClientMock):
939 # Test put()s that need two distinct servers to succeed, possibly
940 # requiring multiple passes through the retry loop.
943 self.api_client = self.mock_keep_services(count=2)
944 self.keep_client = arvados.KeepClient(api_client=self.api_client)
946 def test_success_after_exception(self):
947 with tutil.mock_keep_responses(
948 'acbd18db4cc2f85cedef654fccc4a4d8+3',
949 Exception('mock err'), 200, 200) as req_mock:
950 self.keep_client.put('foo', num_retries=1, copies=2)
951 self.assertTrue(3, req_mock.call_count)
953 def test_success_after_retryable_error(self):
954 with tutil.mock_keep_responses(
955 'acbd18db4cc2f85cedef654fccc4a4d8+3',
956 500, 200, 200) as req_mock:
957 self.keep_client.put('foo', num_retries=1, copies=2)
958 self.assertTrue(3, req_mock.call_count)
960 def test_fail_after_final_error(self):
961 # First retry loop gets a 200 (can't achieve replication by
962 # storing again on that server) and a 400 (can't retry that
963 # server at all), so we shouldn't try a third request.
964 with tutil.mock_keep_responses(
965 'acbd18db4cc2f85cedef654fccc4a4d8+3',
966 200, 400, 200) as req_mock:
967 with self.assertRaises(arvados.errors.KeepWriteError):
968 self.keep_client.put('foo', num_retries=1, copies=2)
969 self.assertTrue(2, req_mock.call_count)