X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/40d92dda0dd3e31cbfd5b887ea8465f9940b7540..04f5f75659e963f02441a566afcc6d22683d0485:/sdk/python/tests/test_keep_client.py diff --git a/sdk/python/tests/test_keep_client.py b/sdk/python/tests/test_keep_client.py index 6198919e8b..4f710fd0c8 100644 --- a/sdk/python/tests/test_keep_client.py +++ b/sdk/python/tests/test_keep_client.py @@ -1,11 +1,14 @@ -# usage example: -# -# ARVADOS_API_TOKEN=abc ARVADOS_API_HOST=arvados.local python -m unittest discover - +import hashlib +import mock import os +import re +import socket import unittest +import urlparse import arvados +import arvados.retry +import arvados_testutil as tutil import run_test_server class KeepTestCase(run_test_server.TestCaseWithServers): @@ -65,6 +68,13 @@ class KeepTestCase(run_test_server.TestCaseWithServers): blob_str, 'wrong content from Keep.get(md5())') + def test_KeepEmptyCollectionTest(self): + blob_locator = self.keep_client.put('', copies=1) + self.assertRegexpMatches( + blob_locator, + '^d41d8cd98f00b204e9800998ecf8427e\+0', + ('wrong locator from Keep.put(""): ' + blob_locator)) + class KeepPermissionTestCase(run_test_server.TestCaseWithServers): MAIN_SERVER = {} @@ -219,3 +229,291 @@ class KeepProxyTestCase(run_test_server.TestCaseWithServers): 'baz2', 'wrong content from Keep.get(md5("baz2"))') self.assertTrue(keep_client.using_proxy) + + +class KeepClientServiceTestCase(unittest.TestCase): + def mock_keep_services(self, *services): + api_client = mock.MagicMock(name='api_client') + api_client.keep_services().accessible().execute.return_value = { + 'items_available': len(services), + 'items': [{ + 'uuid': 'zzzzz-bi6l4-{:015x}'.format(index), + 'owner_uuid': 'zzzzz-tpzed-000000000000000', + 'service_host': host, + 'service_port': port, + 'service_ssl_flag': ssl, + 'service_type': servtype, + } for index, (host, port, ssl, servtype) + in enumerate(services)], + } + return api_client + + def mock_n_keep_disks(self, service_count): + return self.mock_keep_services( + *[("keep0x{:x}".format(index), 80, False, 'disk') + for index in range(service_count)]) + + def get_service_roots(self, *services): + api_client = self.mock_keep_services(*services) + keep_client = arvados.KeepClient(api_client=api_client) + services = keep_client.weighted_service_roots('000000') + return [urlparse.urlparse(url) for url in sorted(services)] + + def test_ssl_flag_respected_in_roots(self): + services = self.get_service_roots(('keep', 10, False, 'disk'), + ('keep', 20, True, 'disk')) + self.assertEqual(10, services[0].port) + self.assertEqual('http', services[0].scheme) + self.assertEqual(20, services[1].port) + self.assertEqual('https', services[1].scheme) + + def test_correct_ports_with_ipv6_addresses(self): + service = self.get_service_roots(('100::1', 10, True, 'proxy'))[0] + self.assertEqual('100::1', service.hostname) + self.assertEqual(10, service.port) + + # test_get_timeout and test_put_timeout test that + # KeepClient.get and KeepClient.put use the appropriate timeouts + # when connected directly to a Keep server (i.e. non-proxy timeout) + + def test_get_timeout(self): + api_client = self.mock_keep_services(('keep', 10, False, 'disk')) + force_timeout = [socket.timeout("timed out")] + with tutil.mock_get(force_timeout) as mock_session: + keep_client = arvados.KeepClient(api_client=api_client) + with self.assertRaises(arvados.errors.KeepReadError): + keep_client.get('ffffffffffffffffffffffffffffffff') + self.assertTrue(mock_session.return_value.get.called) + self.assertEqual( + arvados.KeepClient.DEFAULT_TIMEOUT, + mock_session.return_value.get.call_args[1]['timeout']) + + def test_put_timeout(self): + api_client = self.mock_keep_services(('keep', 10, False, 'disk')) + force_timeout = [socket.timeout("timed out")] + with tutil.mock_put(force_timeout) as mock_session: + keep_client = arvados.KeepClient(api_client=api_client) + with self.assertRaises(arvados.errors.KeepWriteError): + keep_client.put('foo') + self.assertTrue(mock_session.return_value.put.called) + self.assertEqual( + arvados.KeepClient.DEFAULT_TIMEOUT, + mock_session.return_value.put.call_args[1]['timeout']) + + def test_proxy_get_timeout(self): + # Force a timeout, verifying that the requests.get or + # requests.put method was called with the proxy_timeout + # setting rather than the default timeout. + api_client = self.mock_keep_services(('keep', 10, False, 'proxy')) + force_timeout = [socket.timeout("timed out")] + with tutil.mock_get(force_timeout) as mock_session: + keep_client = arvados.KeepClient(api_client=api_client) + with self.assertRaises(arvados.errors.KeepReadError): + keep_client.get('ffffffffffffffffffffffffffffffff') + self.assertTrue(mock_session.return_value.get.called) + self.assertEqual( + arvados.KeepClient.DEFAULT_PROXY_TIMEOUT, + mock_session.return_value.get.call_args[1]['timeout']) + + def test_proxy_put_timeout(self): + # Force a timeout, verifying that the requests.get or + # requests.put method was called with the proxy_timeout + # setting rather than the default timeout. + api_client = self.mock_keep_services(('keep', 10, False, 'proxy')) + force_timeout = [socket.timeout("timed out")] + with tutil.mock_put(force_timeout) as mock_session: + keep_client = arvados.KeepClient(api_client=api_client) + with self.assertRaises(arvados.errors.KeepWriteError): + keep_client.put('foo') + self.assertTrue(mock_session.return_value.put.called) + self.assertEqual( + arvados.KeepClient.DEFAULT_PROXY_TIMEOUT, + mock_session.return_value.put.call_args[1]['timeout']) + + def test_probe_order_reference_set(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'. + expected_order = [ + list('3eab2d5fc9681074'), + list('097dba52e648f1c3'), + list('c5b4e023f8a7d691'), + list('9d81c02e76a3bf54'), + ] + hashes = [ + hashlib.md5("{:064x}".format(x)).hexdigest() + for x in range(len(expected_order))] + api_client = self.mock_n_keep_disks(16) + keep_client = arvados.KeepClient(api_client=api_client) + for i, hash in enumerate(hashes): + roots = keep_client.weighted_service_roots(hash) + got_order = [ + re.search(r'//\[?keep0x([0-9a-f]+)', root).group(1) + for root in roots] + self.assertEqual(expected_order[i], got_order) + + def test_probe_waste_adding_one_server(self): + hashes = [ + hashlib.md5("{:064x}".format(x)).hexdigest() for x in range(100)] + initial_services = 12 + api_client = self.mock_n_keep_disks(initial_services) + keep_client = arvados.KeepClient(api_client=api_client) + probes_before = [ + keep_client.weighted_service_roots(hash) for hash in hashes] + for added_services in range(1, 12): + api_client = self.mock_n_keep_disks(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( + 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)) + + +class KeepClientRetryTestMixin(object): + # Testing with a local Keep store won't exercise the retry behavior. + # Instead, our strategy is: + # * Create a client with one proxy specified (pointed at a black + # hole), so there's no need to instantiate an API client, and + # all HTTP requests come from one place. + # * Mock httplib's request method to provide simulated responses. + # This lets us test the retry logic extensively without relying on any + # supporting servers, and prevents side effects in case something hiccups. + # To use this mixin, define DEFAULT_EXPECT, DEFAULT_EXCEPTION, and + # run_method(). + # + # Test classes must define TEST_PATCHER to a method that mocks + # out appropriate methods in the client. + + PROXY_ADDR = 'http://[%s]:65535/' % (tutil.TEST_HOST,) + TEST_DATA = 'testdata' + TEST_LOCATOR = 'ef654c40ab4f1747fc699915d4f70902+8' + + def setUp(self): + self.client_kwargs = {'proxy': self.PROXY_ADDR, 'local_store': ''} + + def new_client(self, **caller_kwargs): + kwargs = self.client_kwargs.copy() + kwargs.update(caller_kwargs) + return arvados.KeepClient(**kwargs) + + def run_method(self, *args, **kwargs): + raise NotImplementedError("test subclasses must define run_method") + + def check_success(self, expected=None, *args, **kwargs): + if expected is None: + expected = self.DEFAULT_EXPECT + self.assertEqual(expected, self.run_method(*args, **kwargs)) + + def check_exception(self, error_class=None, *args, **kwargs): + if error_class is None: + error_class = self.DEFAULT_EXCEPTION + self.assertRaises(error_class, self.run_method, *args, **kwargs) + + def test_immediate_success(self): + with self.TEST_PATCHER(self.DEFAULT_EXPECT, 200): + self.check_success() + + def test_retry_then_success(self): + with self.TEST_PATCHER(self.DEFAULT_EXPECT, 500, 200): + self.check_success(num_retries=3) + + def test_no_default_retry(self): + with self.TEST_PATCHER(self.DEFAULT_EXPECT, 500, 200): + self.check_exception() + + def test_no_retry_after_permanent_error(self): + with self.TEST_PATCHER(self.DEFAULT_EXPECT, 403, 200): + self.check_exception(num_retries=3) + + def test_error_after_retries_exhausted(self): + with self.TEST_PATCHER(self.DEFAULT_EXPECT, 500, 500, 200): + self.check_exception(num_retries=1) + + def test_num_retries_instance_fallback(self): + self.client_kwargs['num_retries'] = 3 + with self.TEST_PATCHER(self.DEFAULT_EXPECT, 500, 200): + self.check_success() + + +@tutil.skip_sleep +class KeepClientRetryGetTestCase(KeepClientRetryTestMixin, unittest.TestCase): + DEFAULT_EXPECT = KeepClientRetryTestMixin.TEST_DATA + DEFAULT_EXCEPTION = arvados.errors.KeepReadError + HINTED_LOCATOR = KeepClientRetryTestMixin.TEST_LOCATOR + '+K@xyzzy' + TEST_PATCHER = staticmethod(tutil.mock_get_responses) + + def run_method(self, locator=KeepClientRetryTestMixin.TEST_LOCATOR, + *args, **kwargs): + return self.new_client().get(locator, *args, **kwargs) + + def test_specific_exception_when_not_found(self): + with tutil.mock_get_responses(self.DEFAULT_EXPECT, 404, 200): + self.check_exception(arvados.errors.NotFoundError, num_retries=3) + + def test_general_exception_with_mixed_errors(self): + # get should raise a NotFoundError if no server returns the block, + # and a high threshold of servers report that it's not found. + # This test rigs up 50/50 disagreement between two servers, and + # checks that it does not become a NotFoundError. + client = self.new_client() + with tutil.mock_get_responses(self.DEFAULT_EXPECT, 404, 500): + with self.assertRaises(arvados.errors.KeepReadError) as exc_check: + client.get(self.HINTED_LOCATOR) + self.assertNotIsInstance( + exc_check.exception, arvados.errors.NotFoundError, + "mixed errors raised NotFoundError") + + def test_hint_server_can_succeed_without_retries(self): + with tutil.mock_get_responses(self.DEFAULT_EXPECT, 404, 200, 500): + self.check_success(locator=self.HINTED_LOCATOR) + + def test_try_next_server_after_timeout(self): + with tutil.mock_get([ + socket.timeout("timed out"), + tutil.fake_requests_response(200, self.DEFAULT_EXPECT)]): + self.check_success(locator=self.HINTED_LOCATOR) + + def test_retry_data_with_wrong_checksum(self): + with tutil.mock_get((tutil.fake_requests_response(200, s) for s in ['baddata', self.TEST_DATA])): + self.check_success(locator=self.HINTED_LOCATOR) + + +@tutil.skip_sleep +class KeepClientRetryPutTestCase(KeepClientRetryTestMixin, unittest.TestCase): + DEFAULT_EXPECT = KeepClientRetryTestMixin.TEST_LOCATOR + DEFAULT_EXCEPTION = arvados.errors.KeepWriteError + TEST_PATCHER = staticmethod(tutil.mock_put_responses) + + def run_method(self, data=KeepClientRetryTestMixin.TEST_DATA, + copies=1, *args, **kwargs): + return self.new_client().put(data, copies, *args, **kwargs) + + def test_do_not_send_multiple_copies_to_same_server(self): + with tutil.mock_put_responses(self.DEFAULT_EXPECT, 200): + self.check_exception(copies=2, num_retries=3)