X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/d7a0d54602493bfdfe4a69887f0c2627ccc6d461..0eb72b526bf8bbb011551ecf019f604e17a534f1:/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 52eb0e9915..c2c4728253 100644 --- a/sdk/python/tests/test_keep_client.py +++ b/sdk/python/tests/test_keep_client.py @@ -1,3 +1,14 @@ +# Copyright (C) The Arvados Authors. All rights reserved. +# +# SPDX-License-Identifier: Apache-2.0 + +from __future__ import absolute_import +from __future__ import division +from future import standard_library +standard_library.install_aliases() +from builtins import str +from builtins import range +from builtins import object import hashlib import mock import os @@ -5,16 +16,16 @@ import pycurl import random import re import socket -import threading +import sys import time import unittest -import urlparse +import urllib.parse import arvados import arvados.retry -import arvados_testutil as tutil -import keepstub -import run_test_server +from . import arvados_testutil as tutil +from . import keepstub +from . import run_test_server class KeepTestCase(run_test_server.TestCaseWithServers): MAIN_SERVER = {} @@ -29,19 +40,26 @@ 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( + self.assertRegex( 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', + b'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' + blob_str = b'\xff\xfe\xf7\x00\x01\x02' blob_locator = self.keep_client.put(blob_str) - self.assertRegexpMatches( + self.assertRegex( blob_locator, '^7fc7c53b45e53926ba52821140fef396\+6', ('wrong locator from Keep.put():' + blob_locator)) @@ -50,32 +68,33 @@ class KeepTestCase(run_test_server.TestCaseWithServers): 'wrong content from Keep.get(md5())') def test_KeepLongBinaryRWTest(self): - blob_str = '\xff\xfe\xfd\xfc\x00\x01\x02\x03' + blob_data = b'\xff\xfe\xfd\xfc\x00\x01\x02\x03' for i in range(0,23): - blob_str = blob_str + blob_str - blob_locator = self.keep_client.put(blob_str) - self.assertRegexpMatches( + blob_data = blob_data + blob_data + blob_locator = self.keep_client.put(blob_data) + self.assertRegex( blob_locator, '^84d90fc0d8175dd5dcfab04b999bc956\+67108864', ('wrong locator from Keep.put(): ' + blob_locator)) self.assertEqual(self.keep_client.get(blob_locator), - blob_str, + blob_data, '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) - self.assertRegexpMatches( + blob_data = b'\xff\xfe\xfd\xfc\x00\x01\x02\x03' + blob_locator = self.keep_client.put(blob_data, copies=1) + self.assertRegex( blob_locator, '^c902006bc98a3eb4a3663b65ab4a6fab\+8', ('wrong locator from Keep.put(): ' + blob_locator)) self.assertEqual(self.keep_client.get(blob_locator), - blob_str, + blob_data, 'wrong content from Keep.get(md5())') def test_KeepEmptyCollectionTest(self): blob_locator = self.keep_client.put('', copies=1) - self.assertRegexpMatches( + self.assertRegex( blob_locator, '^d41d8cd98f00b204e9800998ecf8427e\+0', ('wrong locator from Keep.put(""): ' + blob_locator)) @@ -83,19 +102,31 @@ class KeepTestCase(run_test_server.TestCaseWithServers): 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( + self.assertRegex( 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') + if sys.version_info < (3, 0): + 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 + with self.assertRaises(AttributeError): + # Must be bytes or have an encode() method self.keep_client.put({}) + def test_KeepHeadTest(self): + locator = self.keep_client.put('test_head') + self.assertRegex( + 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), + b'test_head', + 'wrong content from Keep.get for "test_head"') + class KeepPermissionTestCase(run_test_server.TestCaseWithServers): MAIN_SERVER = {} KEEP_SERVER = {'blob_signing_key': 'abcdefghijk0123456789', @@ -105,18 +136,18 @@ class KeepPermissionTestCase(run_test_server.TestCaseWithServers): run_test_server.authorize_with('active') keep_client = arvados.KeepClient() foo_locator = keep_client.put('foo') - self.assertRegexpMatches( + self.assertRegex( foo_locator, r'^acbd18db4cc2f85cedef654fccc4a4d8\+3\+A[a-f0-9]+@[a-f0-9]+$', 'invalid locator from Keep.put("foo"): ' + foo_locator) self.assertEqual(keep_client.get(foo_locator), - 'foo', + b'foo', 'wrong content from Keep.get(md5("foo"))') # GET with an unsigned locator => NotFound bar_locator = keep_client.put('bar') unsigned_bar_locator = "37b51d194a7513e45b56f6524f2d51f2+3" - self.assertRegexpMatches( + self.assertRegex( bar_locator, r'^37b51d194a7513e45b56f6524f2d51f2\+3\+A[a-f0-9]+@[a-f0-9]+$', 'invalid locator from Keep.put("bar"): ' + bar_locator) @@ -168,7 +199,7 @@ class KeepOptionalPermission(run_test_server.TestCaseWithServers): def _put_foo_and_check(self): signed_locator = self.keep_client.put('foo') - self.assertRegexpMatches( + self.assertRegex( signed_locator, r'^acbd18db4cc2f85cedef654fccc4a4d8\+3\+A[a-f0-9]+@[a-f0-9]+$', 'invalid locator from Keep.put("foo"): ' + signed_locator) @@ -177,13 +208,13 @@ class KeepOptionalPermission(run_test_server.TestCaseWithServers): def test_KeepAuthenticatedSignedTest(self): signed_locator = self._put_foo_and_check() self.assertEqual(self.keep_client.get(signed_locator), - 'foo', + b'foo', 'wrong content from Keep.get(md5("foo"))') def test_KeepAuthenticatedUnsignedTest(self): signed_locator = self._put_foo_and_check() self.assertEqual(self.keep_client.get("acbd18db4cc2f85cedef654fccc4a4d8"), - 'foo', + b'foo', 'wrong content from Keep.get(md5("foo"))') def test_KeepUnauthenticatedSignedTest(self): @@ -192,7 +223,7 @@ class KeepOptionalPermission(run_test_server.TestCaseWithServers): signed_locator = self._put_foo_and_check() self.keep_client.api_token = '' self.assertEqual(self.keep_client.get(signed_locator), - 'foo', + b'foo', 'wrong content from Keep.get(md5("foo"))') def test_KeepUnauthenticatedUnsignedTest(self): @@ -201,7 +232,7 @@ class KeepOptionalPermission(run_test_server.TestCaseWithServers): signed_locator = self._put_foo_and_check() self.keep_client.api_token = '' self.assertEqual(self.keep_client.get("acbd18db4cc2f85cedef654fccc4a4d8"), - 'foo', + b'foo', 'wrong content from Keep.get(md5("foo"))') @@ -221,17 +252,17 @@ 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') - self.assertRegexpMatches( + self.assertRegex( baz_locator, '^73feffa4b7f6bb68e44cf984c85f6e88\+3', 'wrong md5 hash from Keep.put("baz"): ' + baz_locator) self.assertEqual(keep_client.get(baz_locator), - 'baz', + b'baz', 'wrong content from Keep.get(md5("baz"))') self.assertTrue(keep_client.using_proxy) @@ -242,21 +273,37 @@ class KeepProxyTestCase(run_test_server.TestCaseWithServers): keep_client = arvados.KeepClient(api_client=self.api_client, proxy='', local_store='') baz_locator = keep_client.put('baz2') - self.assertRegexpMatches( + self.assertRegex( baz_locator, '^91f372a266fe2bf2823cb8ec7fda31ce\+4', 'wrong md5 hash from Keep.put("baz2"): ' + baz_locator) self.assertEqual(keep_client.get(baz_locator), - 'baz2', + b'baz2', '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): 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)] + return [urllib.parse.urlparse(url) for url in sorted(services)] def test_ssl_flag_respected_in_roots(self): for ssl_flag in [False, True]: @@ -299,7 +346,24 @@ class KeepClientServiceTestCase(unittest.TestCase, tutil.ApiClientMock): 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') + keep_client.put(b'foo') + 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_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)) @@ -327,6 +391,23 @@ class KeepClientServiceTestCase(unittest.TestCase, tutil.ApiClientMock): 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): api_client = self.mock_keep_services(service_type='proxy', count=1) force_timeout = socket.timeout("timed out") @@ -356,6 +437,9 @@ class KeepClientServiceTestCase(unittest.TestCase, tutil.ApiClientMock): def test_get_error_with_no_services(self): self.check_no_services_error('get', arvados.errors.KeepReadError) + 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) @@ -370,11 +454,14 @@ class KeepClientServiceTestCase(unittest.TestCase, tutil.ApiClientMock): num_retries=3) self.assertEqual([403, 403], [ getattr(error, 'status_code', None) - for error in err_check.exception.request_errors().itervalues()]) + for error in err_check.exception.request_errors().values()]) 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) @@ -400,7 +487,7 @@ class KeepClientServiceTestCase(unittest.TestCase, tutil.ApiClientMock): self.assertEqual(0, len(exc_check.exception.request_errors())) def test_oddball_service_get(self): - body = 'oddball service get' + body = b'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) @@ -408,7 +495,7 @@ class KeepClientServiceTestCase(unittest.TestCase, tutil.ApiClientMock): self.assertEqual(body, actual) def test_oddball_service_put(self): - body = 'oddball service put' + body = b'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): @@ -417,7 +504,7 @@ class KeepClientServiceTestCase(unittest.TestCase, tutil.ApiClientMock): self.assertEqual(pdh, actual) def test_oddball_service_writer_count(self): - body = 'oddball service writer count' + body = b'oddball service writer count' pdh = tutil.str_keep_locator(body) api_client = self.mock_keep_services(service_type='fancynewblobstore', count=4) @@ -448,7 +535,7 @@ class KeepClientRendezvousTestCase(unittest.TestCase, tutil.ApiClientMock): list('9d81c02e76a3bf54'), ] self.blocks = [ - "{:064x}".format(x) + "{:064x}".format(x).encode() for x in range(len(self.expected_order))] self.hashes = [ hashlib.md5(self.blocks[x]).hexdigest() @@ -469,6 +556,10 @@ class KeepClientRendezvousTestCase(unittest.TestCase, tutil.ApiClientMock): 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. @@ -481,7 +572,7 @@ class KeepClientRendezvousTestCase(unittest.TestCase, tutil.ApiClientMock): self.assertRaises(arvados.errors.KeepRequestError): op(i) got_order = [ - re.search(r'//\[?keep0x([0-9a-f]+)', resp.getopt(pycurl.URL)).group(1) + re.search(r'//\[?keep0x([0-9a-f]+)', resp.getopt(pycurl.URL).decode()).group(1) for resp in mock.responses] self.assertEqual(self.expected_order[i]*2, got_order) @@ -492,7 +583,7 @@ class KeepClientRendezvousTestCase(unittest.TestCase, tutil.ApiClientMock): 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) + re.search(r'//\[?keep0x([0-9a-f]+)', resp.getopt(pycurl.URL).decode()).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 @@ -520,7 +611,7 @@ class KeepClientRendezvousTestCase(unittest.TestCase, tutil.ApiClientMock): def test_probe_waste_adding_one_server(self): hashes = [ - hashlib.md5("{:064x}".format(x)).hexdigest() for x in range(100)] + hashlib.md5("{:064x}".format(x).encode()).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) @@ -558,7 +649,7 @@ class KeepClientRendezvousTestCase(unittest.TestCase, tutil.ApiClientMock): max_penalty)) def check_64_zeros_error_order(self, verb, exc_class): - data = '0' * 64 + data = b'0' * 64 if verb == 'get': data = tutil.str_keep_locator(data) # Arbitrary port number: @@ -567,9 +658,9 @@ class KeepClientRendezvousTestCase(unittest.TestCase, tutil.ApiClientMock): 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 + curl_mock.return_value = tutil.FakeCurl.make(code=500, body=b'') getattr(keep_client, verb)(data) - urls = [urlparse.urlparse(url) + urls = [urllib.parse.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]) @@ -581,11 +672,11 @@ class KeepClientRendezvousTestCase(unittest.TestCase, tutil.ApiClientMock): self.check_64_zeros_error_order('put', arvados.errors.KeepWriteError) -class KeepClientTimeout(unittest.TestCase, tutil.ApiClientMock): +class KeepClientTimeout(keepstub.StubKeepServers, unittest.TestCase): # BANDWIDTH_LOW_LIM must be less than len(DATA) so we can transfer # 1s worth of data and then trigger bandwidth errors before running # out of data. - DATA = 'x'*2**11 + DATA = b'x'*2**11 BANDWIDTH_LOW_LIM = 1024 TIMEOUT_TIME = 1.0 @@ -615,24 +706,6 @@ class KeepClientTimeout(unittest.TestCase, tutil.ApiClientMock): delta = round(time.time() - self.t0, 3) self.assertGreaterEqual(delta, self.tmin) - def setUp(self): - sock = socket.socket() - sock.bind(('0.0.0.0', 0)) - self.port = sock.getsockname()[1] - sock.close() - self.server = keepstub.Server(('0.0.0.0', self.port), keepstub.Handler) - self.thread = threading.Thread(target=self.server.serve_forever) - self.thread.daemon = True # Exit thread if main proc exits - self.thread.start() - self.api_client = self.mock_keep_services( - count=1, - service_host='localhost', - service_port=self.port, - ) - - def tearDown(self): - self.server.shutdown() - def keepClient(self, timeouts=(0.1, TIMEOUT_TIME, BANDWIDTH_LOW_LIM)): return arvados.KeepClient( api_client=self.api_client, @@ -679,6 +752,9 @@ class KeepClientTimeout(unittest.TestCase, tutil.ApiClientMock): with self.assertTakesGreater(self.TIMEOUT_TIME): with self.assertRaises(arvados.errors.KeepWriteError): kc.put(self.DATA, copies=1, num_retries=0) + with self.assertTakesGreater(self.TIMEOUT_TIME): + with self.assertRaises(arvados.errors.KeepReadError) as e: + kc.head(loc, num_retries=0) def test_low_bandwidth_with_server_mid_delay_failure(self): kc = self.keepClient() @@ -727,10 +803,10 @@ class KeepClientTimeout(unittest.TestCase, tutil.ApiClientMock): # Allow 10s to connect, then 1s for response. Nothing should # work, and everything should take at least 1s to return. kc = self.keepClient(timeouts=(10, 1)) - with self.assertTakesBetween(1, 1.9): + with self.assertTakesBetween(1, 9): with self.assertRaises(arvados.errors.KeepReadError): kc.get(loc, num_retries=0) - with self.assertTakesBetween(1, 1.9): + with self.assertTakesBetween(1, 9): with self.assertRaises(arvados.errors.KeepWriteError): kc.put(self.DATA, copies=1, num_retries=0) @@ -758,9 +834,10 @@ class KeepClientGatewayTestCase(unittest.TestCase, tutil.ApiClientMock): code=200, body='foo', headers={'Content-Length': 3}) self.mock_disks_and_gateways() locator = 'acbd18db4cc2f85cedef654fccc4a4d8+3+K@' + self.gateways[0]['uuid'] - self.assertEqual('foo', self.keepClient.get(locator)) + self.assertEqual(b'foo', self.keepClient.get(locator)) self.assertEqual(self.gateway_roots[0]+locator, - MockCurl.return_value.getopt(pycurl.URL)) + MockCurl.return_value.getopt(pycurl.URL).decode()) + self.assertEqual(True, self.keepClient.head(locator)) @mock.patch('pycurl.Curl') def test_get_with_gateway_hints_in_order(self, MockCurl): @@ -779,22 +856,56 @@ class KeepClientGatewayTestCase(unittest.TestCase, tutil.ApiClientMock): # Gateways are tried first, in the order given. for i, root in enumerate(self.gateway_roots): self.assertEqual(root+locator, - mocks[i].getopt(pycurl.URL)) + mocks[i].getopt(pycurl.URL).decode()) + # Disk services are tried next. + for i in range(gateways, gateways+disks): + self.assertRegex( + mocks[i].getopt(pycurl.URL).decode(), + r'keep0x') + + @mock.patch('pycurl.Curl') + def test_head_with_gateway_hints_in_order(self, MockCurl): + gateways = 4 + disks = 3 + mocks = [ + tutil.FakeCurl.make(code=404, body=b'') + for _ in range(gateways+disks) + ] + MockCurl.side_effect = tutil.queue_with(mocks) + self.mock_disks_and_gateways(gateways=gateways, disks=disks) + locator = '+'.join(['acbd18db4cc2f85cedef654fccc4a4d8+3'] + + ['K@'+gw['uuid'] for gw in self.gateways]) + with self.assertRaises(arvados.errors.NotFoundError): + self.keepClient.head(locator) + # Gateways are tried first, in the order given. + for i, root in enumerate(self.gateway_roots): + self.assertEqual(root+locator, + mocks[i].getopt(pycurl.URL).decode()) # Disk services are tried next. for i in range(gateways, gateways+disks): - self.assertRegexpMatches( - mocks[i].getopt(pycurl.URL), + self.assertRegex( + mocks[i].getopt(pycurl.URL).decode(), r'keep0x') @mock.patch('pycurl.Curl') def test_get_with_remote_proxy_hint(self, MockCurl): MockCurl.return_value = tutil.FakeCurl.make( - code=200, body='foo', headers={'Content-Length': 3}) + code=200, body=b'foo', headers={'Content-Length': 3}) self.mock_disks_and_gateways() locator = 'acbd18db4cc2f85cedef654fccc4a4d8+3+K@xyzzy' - self.assertEqual('foo', self.keepClient.get(locator)) + self.assertEqual(b'foo', self.keepClient.get(locator)) self.assertEqual('https://keep.xyzzy.arvadosapi.com/'+locator, - MockCurl.return_value.getopt(pycurl.URL)) + MockCurl.return_value.getopt(pycurl.URL).decode()) + + @mock.patch('pycurl.Curl') + def test_head_with_remote_proxy_hint(self, MockCurl): + MockCurl.return_value = tutil.FakeCurl.make( + code=200, body=b'foo', headers={'Content-Length': 3}) + self.mock_disks_and_gateways() + locator = 'acbd18db4cc2f85cedef654fccc4a4d8+3+K@xyzzy' + self.assertEqual(True, self.keepClient.head(locator)) + self.assertEqual('https://keep.xyzzy.arvadosapi.com/'+locator, + MockCurl.return_value.getopt(pycurl.URL).decode()) class KeepClientRetryTestMixin(object): @@ -813,7 +924,7 @@ class KeepClientRetryTestMixin(object): # out appropriate methods in the client. PROXY_ADDR = 'http://[%s]:65535/' % (tutil.TEST_HOST,) - TEST_DATA = 'testdata' + TEST_DATA = b'testdata' TEST_LOCATOR = 'ef654c40ab4f1747fc699915d4f70902+8' def setUp(self): @@ -845,6 +956,10 @@ class KeepClientRetryTestMixin(object): with self.TEST_PATCHER(self.DEFAULT_EXPECT, 500, 200): self.check_success(num_retries=3) + def test_exception_then_success(self): + with self.TEST_PATCHER(self.DEFAULT_EXPECT, Exception('mock err'), 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() @@ -907,6 +1022,43 @@ class KeepClientRetryGetTestCase(KeepClientRetryTestMixin, unittest.TestCase): (self.DEFAULT_EXPECT, 200)): self.check_success(locator=self.HINTED_LOCATOR) +@tutil.skip_sleep +class KeepClientRetryHeadTestCase(KeepClientRetryTestMixin, unittest.TestCase): + DEFAULT_EXPECT = True + DEFAULT_EXCEPTION = arvados.errors.KeepReadError + HINTED_LOCATOR = KeepClientRetryTestMixin.TEST_LOCATOR + '+K@xyzzy' + TEST_PATCHER = staticmethod(tutil.mock_keep_responses) + + def run_method(self, locator=KeepClientRetryTestMixin.TEST_LOCATOR, + *args, **kwargs): + return self.new_client().head(locator, *args, **kwargs) + + def test_specific_exception_when_not_found(self): + with tutil.mock_keep_responses(self.DEFAULT_EXPECT, 404, 200): + self.check_exception(arvados.errors.NotFoundError, num_retries=3) + + def test_general_exception_with_mixed_errors(self): + # head 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_keep_responses(self.DEFAULT_EXPECT, 404, 500): + with self.assertRaises(arvados.errors.KeepReadError) as exc_check: + client.head(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_keep_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_keep_responses( + (socket.timeout("timed out"), 200), + (self.DEFAULT_EXPECT, 200)): + self.check_success(locator=self.HINTED_LOCATOR) @tutil.skip_sleep class KeepClientRetryPutTestCase(KeepClientRetryTestMixin, unittest.TestCase): @@ -921,3 +1073,108 @@ class KeepClientRetryPutTestCase(KeepClientRetryTestMixin, unittest.TestCase): def test_do_not_send_multiple_copies_to_same_server(self): with tutil.mock_keep_responses(self.DEFAULT_EXPECT, 200): self.check_exception(copies=2, num_retries=3) + + +class AvoidOverreplication(unittest.TestCase, tutil.ApiClientMock): + + class FakeKeepService(object): + def __init__(self, delay, will_succeed=False, will_raise=None, replicas=1): + self.delay = delay + self.will_succeed = will_succeed + self.will_raise = will_raise + self._result = {} + self._result['headers'] = {} + self._result['headers']['x-keep-replicas-stored'] = str(replicas) + self._result['body'] = 'foobar' + + def put(self, data_hash, data, timeout): + time.sleep(self.delay) + if self.will_raise is not None: + raise self.will_raise + return self.will_succeed + + def last_result(self): + if self.will_succeed: + return self._result + + def finished(self): + return False + + def setUp(self): + self.copies = 3 + self.pool = arvados.KeepClient.KeepWriterThreadPool( + data = 'foo', + data_hash = 'acbd18db4cc2f85cedef654fccc4a4d8+3', + max_service_replicas = self.copies, + copies = self.copies + ) + + def test_only_write_enough_on_success(self): + for i in range(10): + ks = self.FakeKeepService(delay=i/10.0, will_succeed=True) + self.pool.add_task(ks, None) + self.pool.join() + self.assertEqual(self.pool.done(), self.copies) + + def test_only_write_enough_on_partial_success(self): + for i in range(5): + ks = self.FakeKeepService(delay=i/10.0, will_succeed=False) + self.pool.add_task(ks, None) + ks = self.FakeKeepService(delay=i/10.0, will_succeed=True) + self.pool.add_task(ks, None) + self.pool.join() + self.assertEqual(self.pool.done(), self.copies) + + def test_only_write_enough_when_some_crash(self): + for i in range(5): + ks = self.FakeKeepService(delay=i/10.0, will_raise=Exception()) + self.pool.add_task(ks, None) + ks = self.FakeKeepService(delay=i/10.0, will_succeed=True) + self.pool.add_task(ks, None) + self.pool.join() + self.assertEqual(self.pool.done(), self.copies) + + def test_fail_when_too_many_crash(self): + for i in range(self.copies+1): + ks = self.FakeKeepService(delay=i/10.0, will_raise=Exception()) + self.pool.add_task(ks, None) + for i in range(self.copies-1): + ks = self.FakeKeepService(delay=i/10.0, will_succeed=True) + self.pool.add_task(ks, None) + self.pool.join() + self.assertEqual(self.pool.done(), self.copies-1) + + +@tutil.skip_sleep +class RetryNeedsMultipleServices(unittest.TestCase, tutil.ApiClientMock): + # Test put()s that need two distinct servers to succeed, possibly + # requiring multiple passes through the retry loop. + + def setUp(self): + self.api_client = self.mock_keep_services(count=2) + self.keep_client = arvados.KeepClient(api_client=self.api_client) + + def test_success_after_exception(self): + with tutil.mock_keep_responses( + 'acbd18db4cc2f85cedef654fccc4a4d8+3', + Exception('mock err'), 200, 200) as req_mock: + self.keep_client.put('foo', num_retries=1, copies=2) + self.assertEqual(3, req_mock.call_count) + + def test_success_after_retryable_error(self): + with tutil.mock_keep_responses( + 'acbd18db4cc2f85cedef654fccc4a4d8+3', + 500, 200, 200) as req_mock: + self.keep_client.put('foo', num_retries=1, copies=2) + self.assertEqual(3, req_mock.call_count) + + def test_fail_after_final_error(self): + # First retry loop gets a 200 (can't achieve replication by + # storing again on that server) and a 400 (can't retry that + # server at all), so we shouldn't try a third request. + with tutil.mock_keep_responses( + 'acbd18db4cc2f85cedef654fccc4a4d8+3', + 200, 400, 200) as req_mock: + with self.assertRaises(arvados.errors.KeepWriteError): + self.keep_client.put('foo', num_retries=1, copies=2) + self.assertEqual(2, req_mock.call_count)