+ self.assertTrue(keep_client.using_proxy)
+
+
+class KeepClientServiceTestCase(unittest.TestCase, tutil.ApiClientMock):
+ def get_service_roots(self, api_client):
+ keep_client = arvados.KeepClient(api_client=api_client)
+ services = keep_client.weighted_service_roots(arvados.KeepLocator('0'*32))
+ return [urlparse.urlparse(url) for url in sorted(services)]
+
+ def test_ssl_flag_respected_in_roots(self):
+ for ssl_flag in [False, True]:
+ services = self.get_service_roots(self.mock_keep_services(
+ service_ssl_flag=ssl_flag))
+ self.assertEqual(
+ ('https' if ssl_flag else 'http'), services[0].scheme)
+
+ def test_correct_ports_with_ipv6_addresses(self):
+ service = self.get_service_roots(self.mock_keep_services(
+ service_type='proxy', service_host='100::1', service_port=10, count=1))[0]
+ self.assertEqual('100::1', service.hostname)
+ self.assertEqual(10, service.port)
+
+ # test_*_timeout verify that KeepClient instructs pycurl to use
+ # the appropriate connection and read timeouts. They don't care
+ # whether pycurl actually exhibits the expected timeout behavior
+ # -- those tests are in the KeepClientTimeout test class.
+
+ def test_get_timeout(self):
+ api_client = self.mock_keep_services(count=1)
+ force_timeout = socket.timeout("timed out")
+ with tutil.mock_keep_responses(force_timeout, 0) as mock:
+ keep_client = arvados.KeepClient(api_client=api_client)
+ with self.assertRaises(arvados.errors.KeepReadError):
+ keep_client.get('ffffffffffffffffffffffffffffffff')
+ self.assertEqual(
+ mock.responses[0].getopt(pycurl.CONNECTTIMEOUT_MS),
+ int(arvados.KeepClient.DEFAULT_TIMEOUT[0]*1000))
+ self.assertEqual(
+ mock.responses[0].getopt(pycurl.TIMEOUT_MS),
+ int(arvados.KeepClient.DEFAULT_TIMEOUT[1]*1000))
+
+ def test_put_timeout(self):
+ api_client = self.mock_keep_services(count=1)
+ force_timeout = socket.timeout("timed out")
+ with tutil.mock_keep_responses(force_timeout, 0) as mock:
+ keep_client = arvados.KeepClient(api_client=api_client)
+ with self.assertRaises(arvados.errors.KeepWriteError):
+ keep_client.put('foo')
+ self.assertEqual(
+ mock.responses[0].getopt(pycurl.CONNECTTIMEOUT_MS),
+ int(arvados.KeepClient.DEFAULT_TIMEOUT[0]*1000))
+ self.assertEqual(
+ mock.responses[0].getopt(pycurl.TIMEOUT_MS),
+ int(arvados.KeepClient.DEFAULT_TIMEOUT[1]*1000))
+
+ def test_proxy_get_timeout(self):
+ api_client = self.mock_keep_services(service_type='proxy', count=1)
+ force_timeout = socket.timeout("timed out")
+ with tutil.mock_keep_responses(force_timeout, 0) as mock:
+ keep_client = arvados.KeepClient(api_client=api_client)
+ with self.assertRaises(arvados.errors.KeepReadError):
+ keep_client.get('ffffffffffffffffffffffffffffffff')
+ self.assertEqual(
+ mock.responses[0].getopt(pycurl.CONNECTTIMEOUT_MS),
+ int(arvados.KeepClient.DEFAULT_PROXY_TIMEOUT[0]*1000))
+ self.assertEqual(
+ mock.responses[0].getopt(pycurl.TIMEOUT_MS),
+ int(arvados.KeepClient.DEFAULT_PROXY_TIMEOUT[1]*1000))
+
+ def test_proxy_put_timeout(self):
+ api_client = self.mock_keep_services(service_type='proxy', count=1)
+ force_timeout = socket.timeout("timed out")
+ with tutil.mock_keep_responses(force_timeout, 0) as mock:
+ keep_client = arvados.KeepClient(api_client=api_client)
+ with self.assertRaises(arvados.errors.KeepWriteError):
+ keep_client.put('foo')
+ self.assertEqual(
+ mock.responses[0].getopt(pycurl.CONNECTTIMEOUT_MS),
+ int(arvados.KeepClient.DEFAULT_PROXY_TIMEOUT[0]*1000))
+ self.assertEqual(
+ mock.responses[0].getopt(pycurl.TIMEOUT_MS),
+ int(arvados.KeepClient.DEFAULT_PROXY_TIMEOUT[1]*1000))
+
+ def check_no_services_error(self, verb, exc_class):
+ api_client = mock.MagicMock(name='api_client')
+ api_client.keep_services().accessible().execute.side_effect = (
+ arvados.errors.ApiError)
+ keep_client = arvados.KeepClient(api_client=api_client)
+ with self.assertRaises(exc_class) as err_check:
+ getattr(keep_client, verb)('d41d8cd98f00b204e9800998ecf8427e+0')
+ self.assertEqual(0, len(err_check.exception.request_errors()))
+
+ def test_get_error_with_no_services(self):
+ self.check_no_services_error('get', arvados.errors.KeepReadError)
+
+ def test_put_error_with_no_services(self):
+ self.check_no_services_error('put', arvados.errors.KeepWriteError)
+
+ def check_errors_from_last_retry(self, verb, exc_class):
+ api_client = self.mock_keep_services(count=2)
+ req_mock = tutil.mock_keep_responses(
+ "retry error reporting test", 500, 500, 403, 403)
+ with req_mock, tutil.skip_sleep, \
+ self.assertRaises(exc_class) as err_check:
+ keep_client = arvados.KeepClient(api_client=api_client)
+ getattr(keep_client, verb)('d41d8cd98f00b204e9800998ecf8427e+0',
+ num_retries=3)
+ self.assertEqual([403, 403], [
+ getattr(error, 'status_code', None)
+ for error in err_check.exception.request_errors().itervalues()])
+
+ def test_get_error_reflects_last_retry(self):
+ self.check_errors_from_last_retry('get', arvados.errors.KeepReadError)
+
+ def test_put_error_reflects_last_retry(self):
+ self.check_errors_from_last_retry('put', arvados.errors.KeepWriteError)
+
+ def test_put_error_does_not_include_successful_puts(self):
+ data = 'partial failure test'
+ data_loc = '{}+{}'.format(hashlib.md5(data).hexdigest(), len(data))
+ api_client = self.mock_keep_services(count=3)
+ with tutil.mock_keep_responses(data_loc, 200, 500, 500) as req_mock, \
+ self.assertRaises(arvados.errors.KeepWriteError) as exc_check:
+ keep_client = arvados.KeepClient(api_client=api_client)
+ keep_client.put(data)
+ self.assertEqual(2, len(exc_check.exception.request_errors()))
+
+ def test_proxy_put_with_no_writable_services(self):
+ data = 'test with no writable services'
+ data_loc = '{}+{}'.format(hashlib.md5(data).hexdigest(), len(data))
+ api_client = self.mock_keep_services(service_type='proxy', read_only=True, count=1)
+ with tutil.mock_keep_responses(data_loc, 200, 500, 500) as req_mock, \
+ self.assertRaises(arvados.errors.KeepWriteError) as exc_check:
+ keep_client = arvados.KeepClient(api_client=api_client)
+ keep_client.put(data)
+ self.assertEqual(True, ("no Keep services available" in str(exc_check.exception)))
+ self.assertEqual(0, len(exc_check.exception.request_errors()))
+
+
+@tutil.skip_sleep
+class KeepClientRendezvousTestCase(unittest.TestCase, tutil.ApiClientMock):
+
+ def setUp(self):
+ # expected_order[i] is the probe order for
+ # hash=md5(sprintf("%064x",i)) where there are 16 services
+ # with uuid sprintf("anything-%015x",j) with j in 0..15. E.g.,
+ # the first probe for the block consisting of 64 "0"
+ # characters is the service whose uuid is
+ # "zzzzz-bi6l4-000000000000003", so expected_order[0][0]=='3'.
+ self.services = 16
+ self.expected_order = [
+ list('3eab2d5fc9681074'),
+ list('097dba52e648f1c3'),
+ list('c5b4e023f8a7d691'),
+ list('9d81c02e76a3bf54'),
+ ]
+ self.blocks = [
+ "{:064x}".format(x)
+ for x in range(len(self.expected_order))]
+ self.hashes = [
+ hashlib.md5(self.blocks[x]).hexdigest()
+ for x in range(len(self.expected_order))]
+ self.api_client = self.mock_keep_services(count=self.services)
+ self.keep_client = arvados.KeepClient(api_client=self.api_client)
+
+ def test_weighted_service_roots_against_reference_set(self):
+ # Confirm weighted_service_roots() returns the correct order
+ for i, hash in enumerate(self.hashes):
+ roots = self.keep_client.weighted_service_roots(arvados.KeepLocator(hash))
+ got_order = [
+ re.search(r'//\[?keep0x([0-9a-f]+)', root).group(1)
+ for root in roots]
+ self.assertEqual(self.expected_order[i], got_order)
+
+ def test_get_probe_order_against_reference_set(self):
+ self._test_probe_order_against_reference_set(
+ lambda i: self.keep_client.get(self.hashes[i], num_retries=1))
+
+ def test_put_probe_order_against_reference_set(self):
+ # copies=1 prevents the test from being sensitive to races
+ # between writer threads.
+ self._test_probe_order_against_reference_set(
+ lambda i: self.keep_client.put(self.blocks[i], num_retries=1, copies=1))
+
+ def _test_probe_order_against_reference_set(self, op):
+ for i in range(len(self.blocks)):
+ with tutil.mock_keep_responses('', *[500 for _ in range(self.services*2)]) as mock, \
+ self.assertRaises(arvados.errors.KeepRequestError):
+ op(i)
+ got_order = [
+ re.search(r'//\[?keep0x([0-9a-f]+)', resp.getopt(pycurl.URL)).group(1)
+ for resp in mock.responses]
+ self.assertEqual(self.expected_order[i]*2, got_order)
+
+ def test_put_probe_order_multiple_copies(self):
+ for copies in range(2, 4):
+ for i in range(len(self.blocks)):
+ with tutil.mock_keep_responses('', *[500 for _ in range(self.services*3)]) as mock, \
+ self.assertRaises(arvados.errors.KeepWriteError):
+ self.keep_client.put(self.blocks[i], num_retries=2, copies=copies)
+ got_order = [
+ re.search(r'//\[?keep0x([0-9a-f]+)', resp.getopt(pycurl.URL)).group(1)
+ for resp in mock.responses]
+ for pos, expected in enumerate(self.expected_order[i]*3):
+ # With C threads racing to make requests, the
+ # position of a given server in the sequence of
+ # HTTP requests (got_order) should be within C-1
+ # positions of that server's position in the
+ # reference probe sequence (expected_order).
+ close_enough = False
+ for diff in range(1-copies, copies):
+ if 0 <= pos+diff < len(got_order):
+ if expected == got_order[pos+diff]:
+ close_enough = True
+ self.assertEqual(
+ True, close_enough,
+ "With copies={}, got {}, expected {}".format(
+ copies, repr(got_order), repr(self.expected_order[i]*3)))
+
+ def test_probe_waste_adding_one_server(self):
+ hashes = [
+ hashlib.md5("{:064x}".format(x)).hexdigest() for x in range(100)]
+ initial_services = 12
+ self.api_client = self.mock_keep_services(count=initial_services)
+ self.keep_client = arvados.KeepClient(api_client=self.api_client)
+ probes_before = [
+ self.keep_client.weighted_service_roots(arvados.KeepLocator(hash)) for hash in hashes]
+ for added_services in range(1, 12):
+ api_client = self.mock_keep_services(count=initial_services+added_services)
+ keep_client = arvados.KeepClient(api_client=api_client)
+ total_penalty = 0
+ for hash_index in range(len(hashes)):
+ probe_after = keep_client.weighted_service_roots(
+ arvados.KeepLocator(hashes[hash_index]))
+ penalty = probe_after.index(probes_before[hash_index][0])
+ self.assertLessEqual(penalty, added_services)
+ total_penalty += penalty
+ # Average penalty per block should not exceed
+ # N(added)/N(orig) by more than 20%, and should get closer
+ # to the ideal as we add data points.
+ expect_penalty = (
+ added_services *
+ len(hashes) / initial_services)
+ max_penalty = (
+ expect_penalty *
+ (120 - added_services)/100)
+ min_penalty = (
+ expect_penalty * 8/10)
+ self.assertTrue(
+ min_penalty <= total_penalty <= max_penalty,
+ "With {}+{} services, {} blocks, penalty {} but expected {}..{}".format(
+ initial_services,
+ added_services,
+ len(hashes),
+ total_penalty,
+ min_penalty,
+ max_penalty))
+
+ def check_64_zeros_error_order(self, verb, exc_class):
+ data = '0' * 64
+ if verb == 'get':
+ data = hashlib.md5(data).hexdigest() + '+1234'
+ # Arbitrary port number:
+ aport = random.randint(1024,65535)
+ api_client = self.mock_keep_services(service_port=aport, count=self.services)
+ keep_client = arvados.KeepClient(api_client=api_client)
+ with mock.patch('pycurl.Curl') as curl_mock, \
+ self.assertRaises(exc_class) as err_check:
+ curl_mock.return_value.side_effect = socket.timeout
+ getattr(keep_client, verb)(data)
+ urls = [urlparse.urlparse(url)
+ for url in err_check.exception.request_errors()]
+ self.assertEqual([('keep0x' + c, aport) for c in '3eab2d5fc9681074'],
+ [(url.hostname, url.port) for url in urls])
+
+ def test_get_error_shows_probe_order(self):
+ self.check_64_zeros_error_order('get', arvados.errors.KeepReadError)
+
+ def test_put_error_shows_probe_order(self):
+ self.check_64_zeros_error_order('put', arvados.errors.KeepWriteError)
+
+
+class KeepClientTimeout(unittest.TestCase, tutil.ApiClientMock):
+ DATA = 'x' * 2**10
+
+ class assertTakesBetween(unittest.TestCase):
+ def __init__(self, tmin, tmax):
+ self.tmin = tmin
+ self.tmax = tmax
+
+ def __enter__(self):
+ self.t0 = time.time()
+
+ def __exit__(self, *args, **kwargs):
+ self.assertGreater(time.time() - self.t0, self.tmin)
+ self.assertLess(time.time() - self.t0, self.tmax)