X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/d2f68bd1e108c3f2dda2322c427050d019b17e04..ebb2559b3a09636ff687316bbe512e0e8a86b168:/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 be13c55048..85b5bc81f0 100644 --- a/sdk/python/tests/test_keep_client.py +++ b/sdk/python/tests/test_keep_client.py @@ -1,15 +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): @@ -25,14 +29,21 @@ class KeepTestCase(run_test_server.TestCaseWithServers): proxy='', local_store='') def test_KeepBasicRWTest(self): + self.assertEqual(0, self.keep_client.upload_counter.get()) foo_locator = self.keep_client.put('foo') self.assertRegexpMatches( foo_locator, '^acbd18db4cc2f85cedef654fccc4a4d8\+3', 'wrong md5 hash from Keep.put("foo"): ' + foo_locator) + + # 6 bytes because uploaded 2 copies + self.assertEqual(6, self.keep_client.upload_counter.get()) + + self.assertEqual(0, self.keep_client.download_counter.get()) self.assertEqual(self.keep_client.get(foo_locator), 'foo', 'wrong content from Keep.get(md5("foo"))') + self.assertEqual(3, self.keep_client.download_counter.get()) def test_KeepBinaryRWTest(self): blob_str = '\xff\xfe\xf7\x00\x01\x02' @@ -58,6 +69,7 @@ class KeepTestCase(run_test_server.TestCaseWithServers): blob_str, 'wrong content from Keep.get(md5())') + @unittest.skip("unreliable test - please fix and close #8752") def test_KeepSingleCopyRWTest(self): blob_str = '\xff\xfe\xfd\xfc\x00\x01\x02\x03' blob_locator = self.keep_client.put(blob_str, copies=1) @@ -92,6 +104,17 @@ class KeepTestCase(run_test_server.TestCaseWithServers): # Must be a string type self.keep_client.put({}) + def test_KeepHeadTest(self): + locator = self.keep_client.put('test_head') + self.assertRegexpMatches( + locator, + '^b9a772c7049325feb7130fff1f8333e9\+9', + 'wrong md5 hash from Keep.put for "test_head": ' + locator) + self.assertEqual(True, self.keep_client.head(locator)) + self.assertEqual(self.keep_client.get(locator), + 'test_head', + 'wrong content from Keep.get for "test_head"') + class KeepPermissionTestCase(run_test_server.TestCaseWithServers): MAIN_SERVER = {} KEEP_SERVER = {'blob_signing_key': 'abcdefghijk0123456789', @@ -217,8 +240,8 @@ class KeepProxyTestCase(run_test_server.TestCaseWithServers): super(KeepProxyTestCase, self).tearDown() def test_KeepProxyTest1(self): - # Will use ARVADOS_KEEP_PROXY environment variable that is set by - # setUpClass(). + # Will use ARVADOS_KEEP_SERVICES environment variable that + # is set by setUpClass(). keep_client = arvados.KeepClient(api_client=self.api_client, local_store='') baz_locator = keep_client.put('baz') @@ -247,6 +270,22 @@ class KeepProxyTestCase(run_test_server.TestCaseWithServers): 'wrong content from Keep.get(md5("baz2"))') self.assertTrue(keep_client.using_proxy) + def test_KeepProxyTestMultipleURIs(self): + # Test using ARVADOS_KEEP_SERVICES env var overriding any + # existing proxy setting and setting multiple proxies + arvados.config.settings()['ARVADOS_KEEP_SERVICES'] = 'http://10.0.0.1 https://foo.example.org:1234/' + keep_client = arvados.KeepClient(api_client=self.api_client, + local_store='') + uris = [x['_service_root'] for x in keep_client._keep_services] + self.assertEqual(uris, ['http://10.0.0.1/', + 'https://foo.example.org:1234/']) + + def test_KeepProxyTestInvalidURI(self): + arvados.config.settings()['ARVADOS_KEEP_SERVICES'] = 'bad.uri.org' + with self.assertRaises(arvados.errors.ArgumentError): + keep_client = arvados.KeepClient(api_client=self.api_client, + local_store='') + class KeepClientServiceTestCase(unittest.TestCase, tutil.ApiClientMock): def get_service_roots(self, api_client): @@ -267,97 +306,305 @@ class KeepClientServiceTestCase(unittest.TestCase, tutil.ApiClientMock): 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(count=1) - force_timeout = [socket.timeout("timed out")] - with tutil.mock_get(force_timeout) as mock_session: + 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_session.return_value.get.called) self.assertEqual( - arvados.KeepClient.DEFAULT_TIMEOUT, - mock_session.return_value.get.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.LOW_SPEED_TIME), + int(arvados.KeepClient.DEFAULT_TIMEOUT[1])) + self.assertEqual( + mock.responses[0].getopt(pycurl.LOW_SPEED_LIMIT), + int(arvados.KeepClient.DEFAULT_TIMEOUT[2])) def test_put_timeout(self): api_client = self.mock_keep_services(count=1) - force_timeout = [socket.timeout("timed out")] - with tutil.mock_put(force_timeout) as mock_session: + 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_session.return_value.put.called) self.assertEqual( - arvados.KeepClient.DEFAULT_TIMEOUT, - mock_session.return_value.put.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.LOW_SPEED_TIME), + int(arvados.KeepClient.DEFAULT_TIMEOUT[1])) + self.assertEqual( + mock.responses[0].getopt(pycurl.LOW_SPEED_LIMIT), + int(arvados.KeepClient.DEFAULT_TIMEOUT[2])) + + def test_head_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.head('ffffffffffffffffffffffffffffffff') + self.assertEqual( + mock.responses[0].getopt(pycurl.CONNECTTIMEOUT_MS), + int(arvados.KeepClient.DEFAULT_TIMEOUT[0]*1000)) + self.assertEqual( + mock.responses[0].getopt(pycurl.LOW_SPEED_TIME), + int(arvados.KeepClient.DEFAULT_TIMEOUT[1])) + self.assertEqual( + mock.responses[0].getopt(pycurl.LOW_SPEED_LIMIT), + int(arvados.KeepClient.DEFAULT_TIMEOUT[2])) 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(service_type='proxy', count=1) - force_timeout = [socket.timeout("timed out")] - with tutil.mock_get(force_timeout) as mock_session: + 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_session.return_value.get.called) self.assertEqual( - arvados.KeepClient.DEFAULT_PROXY_TIMEOUT, - mock_session.return_value.get.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.LOW_SPEED_TIME), + int(arvados.KeepClient.DEFAULT_PROXY_TIMEOUT[1])) + self.assertEqual( + mock.responses[0].getopt(pycurl.LOW_SPEED_LIMIT), + int(arvados.KeepClient.DEFAULT_PROXY_TIMEOUT[2])) + + def test_proxy_head_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.head('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.LOW_SPEED_TIME), + int(arvados.KeepClient.DEFAULT_PROXY_TIMEOUT[1])) + self.assertEqual( + mock.responses[0].getopt(pycurl.LOW_SPEED_LIMIT), + int(arvados.KeepClient.DEFAULT_PROXY_TIMEOUT[2])) 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(service_type='proxy', count=1) - force_timeout = [socket.timeout("timed out")] - with tutil.mock_put(force_timeout) as mock_session: + 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_session.return_value.put.called) self.assertEqual( - arvados.KeepClient.DEFAULT_PROXY_TIMEOUT, - mock_session.return_value.put.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.LOW_SPEED_TIME), + int(arvados.KeepClient.DEFAULT_PROXY_TIMEOUT[1])) + self.assertEqual( + mock.responses[0].getopt(pycurl.LOW_SPEED_LIMIT), + int(arvados.KeepClient.DEFAULT_PROXY_TIMEOUT[2])) + + 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_probe_order_reference_set(self): + def test_head_error_with_no_services(self): + self.check_no_services_error('head', 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_head_error_reflects_last_retry(self): + self.check_errors_from_last_retry('head', 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 = tutil.str_keep_locator(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 = tutil.str_keep_locator(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())) + + def test_oddball_service_get(self): + body = 'oddball service get' + api_client = self.mock_keep_services(service_type='fancynewblobstore') + with tutil.mock_keep_responses(body, 200): + keep_client = arvados.KeepClient(api_client=api_client) + actual = keep_client.get(tutil.str_keep_locator(body)) + self.assertEqual(body, actual) + + def test_oddball_service_put(self): + body = 'oddball service put' + pdh = tutil.str_keep_locator(body) + api_client = self.mock_keep_services(service_type='fancynewblobstore') + with tutil.mock_keep_responses(pdh, 200): + keep_client = arvados.KeepClient(api_client=api_client) + actual = keep_client.put(body, copies=1) + self.assertEqual(pdh, actual) + + def test_oddball_service_writer_count(self): + body = 'oddball service writer count' + pdh = tutil.str_keep_locator(body) + api_client = self.mock_keep_services(service_type='fancynewblobstore', + count=4) + headers = {'x-keep-replicas-stored': 3} + with tutil.mock_keep_responses(pdh, 200, 418, 418, 418, + **headers) as req_mock: + keep_client = arvados.KeepClient(api_client=api_client) + actual = keep_client.put(body, copies=2) + self.assertEqual(pdh, actual) + self.assertEqual(1, req_mock.call_count) + + +@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_keep_services(count=16) - keep_client = arvados.KeepClient(api_client=api_client) - for i, hash in enumerate(hashes): - roots = keep_client.weighted_service_roots(arvados.KeepLocator(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_head_probe_order_against_reference_set(self): + self._test_probe_order_against_reference_set( + lambda i: self.keep_client.head(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)