X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/cf1db3981ad0c6416d2572ceeb512743a9df24c7..209d83a0ce0bd22669a39f49daddca432babeae7:/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 982e4b44a8..90468924a6 100644 --- a/sdk/python/tests/test_keep_client.py +++ b/sdk/python/tests/test_keep_client.py @@ -1,14 +1,19 @@ import hashlib import mock import os +import pycurl +import random import re import socket +import threading +import time import unittest import urlparse import arvados import arvados.retry import arvados_testutil as tutil +import keepstub import run_test_server class KeepTestCase(run_test_server.TestCaseWithServers): @@ -75,6 +80,21 @@ class KeepTestCase(run_test_server.TestCaseWithServers): '^d41d8cd98f00b204e9800998ecf8427e\+0', ('wrong locator from Keep.put(""): ' + blob_locator)) + def test_unicode_must_be_ascii(self): + # If unicode type, must only consist of valid ASCII + foo_locator = self.keep_client.put(u'foo') + self.assertRegexpMatches( + foo_locator, + '^acbd18db4cc2f85cedef654fccc4a4d8\+3', + 'wrong md5 hash from Keep.put("foo"): ' + foo_locator) + + with self.assertRaises(UnicodeEncodeError): + # Error if it is not ASCII + self.keep_client.put(u'\xe2') + + with self.assertRaises(arvados.errors.ArgumentError): + # Must be a string type + self.keep_client.put({}) class KeepPermissionTestCase(run_test_server.TestCaseWithServers): MAIN_SERVER = {} @@ -188,11 +208,12 @@ class KeepOptionalPermission(run_test_server.TestCaseWithServers): class KeepProxyTestCase(run_test_server.TestCaseWithServers): MAIN_SERVER = {} KEEP_SERVER = {} - KEEP_PROXY_SERVER = {'auth': 'admin'} + KEEP_PROXY_SERVER = {} @classmethod def setUpClass(cls): super(KeepProxyTestCase, cls).setUpClass() + run_test_server.authorize_with('active') cls.api_client = arvados.api('v1') def tearDown(self): @@ -231,145 +252,245 @@ class KeepProxyTestCase(run_test_server.TestCaseWithServers): 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) +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('000000') + 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): - 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) + 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(('100::1', 10, True, 'proxy'))[0] + 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_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) + # 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(('keep', 10, False, 'disk')) - keep_client = arvados.KeepClient(api_client=api_client) - force_timeout = [socket.timeout("timed out")] - with mock.patch('requests.get', side_effect=force_timeout) as mock_request: + 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.assertTrue(mock_request.called) self.assertEqual( - arvados.KeepClient.DEFAULT_TIMEOUT, - mock_request.call_args[1]['timeout']) + 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(('keep', 10, False, 'disk')) - keep_client = arvados.KeepClient(api_client=api_client) - force_timeout = [socket.timeout("timed out")] - with mock.patch('requests.put', side_effect=force_timeout) as mock_request: + 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.assertTrue(mock_request.called) self.assertEqual( - arvados.KeepClient.DEFAULT_TIMEOUT, - mock_request.call_args[1]['timeout']) + 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): - # 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')) - keep_client = arvados.KeepClient(api_client=api_client) - force_timeout = [socket.timeout("timed out")] - with mock.patch('requests.get', side_effect=force_timeout) as mock_request: + 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.assertTrue(mock_request.called) self.assertEqual( - arvados.KeepClient.DEFAULT_PROXY_TIMEOUT, - mock_request.call_args[1]['timeout']) + 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): - # 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')) - keep_client = arvados.KeepClient(api_client=api_client) - force_timeout = [socket.timeout("timed out")] - with mock.patch('requests.put', side_effect=force_timeout) as mock_request: + 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.assertTrue(mock_request.called) self.assertEqual( - arvados.KeepClient.DEFAULT_PROXY_TIMEOUT, - mock_request.call_args[1]['timeout']) + 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_probe_order_reference_set(self): + 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'. - expected_order = [ + self.services = 16 + self.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) + 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(expected_order[i], got_order) + 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] + # With T threads racing to make requests, the position + # of a given server in the sequence of HTTP requests + # (got_order) cannot be more than T-1 positions + # earlier than that server's position in the reference + # probe sequence (expected_order). + # + # Loop invariant: we have accounted for +pos+ expected + # probes, either by seeing them in +got_order+ or by + # putting them in +pending+ in the hope of seeing them + # later. As long as +len(pending)