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):
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)
mock.responses[0].getopt(pycurl.TIMEOUT_MS),
int(arvados.KeepClient.DEFAULT_PROXY_TIMEOUT[1]*1000))
- def test_probe_order_reference_set(self):
+ 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'.
- 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_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)<T+, we haven't
+ # started a request too early.
+ pending = []
+ for pos, expected in enumerate(self.expected_order[i]*3):
+ got = got_order[pos-len(pending)]
+ while got in pending:
+ del pending[pending.index(got)]
+ got = got_order[pos-len(pending)]
+ if got != expected:
+ pending.append(expected)
+ self.assertLess(
+ len(pending), copies,
+ "pending={}, with copies={}, got {}, expected {}".format(
+ pending, 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
- api_client = self.mock_keep_services(count=initial_services)
- keep_client = arvados.KeepClient(api_client=api_client)
+ self.api_client = self.mock_keep_services(count=initial_services)
+ self.keep_client = arvados.KeepClient(api_client=self.api_client)
probes_before = [
- keep_client.weighted_service_roots(arvados.KeepLocator(hash)) for hash in hashes]
+ 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)
data = hashlib.md5(data).hexdigest() + '+1234'
# Arbitrary port number:
aport = random.randint(1024,65535)
- api_client = self.mock_keep_services(service_port=aport, count=16)
+ 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:
def test_put_error_shows_probe_order(self):
self.check_64_zeros_error_order('put', arvados.errors.KeepWriteError)
- 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)
+class KeepClientTimeout(unittest.TestCase, tutil.ApiClientMock):
+ DATA = 'x' * 2**10
- def test_put_error_with_no_services(self):
- self.check_no_services_error('put', arvados.errors.KeepWriteError)
+ class assertTakesBetween(unittest.TestCase):
+ def __init__(self, tmin, tmax):
+ self.tmin = tmin
+ self.tmax = tmax
- 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 __enter__(self):
+ self.t0 = time.time()
- def test_get_error_reflects_last_retry(self):
- self.check_errors_from_last_retry('get', arvados.errors.KeepReadError)
+ def __exit__(self, *args, **kwargs):
+ self.assertGreater(time.time() - self.t0, self.tmin)
+ self.assertLess(time.time() - self.t0, self.tmax)
- def test_put_error_reflects_last_retry(self):
- self.check_errors_from_last_retry('put', arvados.errors.KeepWriteError)
+ 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 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 tearDown(self):
+ self.server.shutdown()
+
+ def keepClient(self, timeouts=(0.1, 1.0)):
+ return arvados.KeepClient(
+ api_client=self.api_client,
+ timeout=timeouts)
+
+ def test_timeout_slow_connect(self):
+ # Can't simulate TCP delays with our own socket. Leave our
+ # stub server running uselessly, and try to connect to an
+ # unroutable IP address instead.
+ self.api_client = self.mock_keep_services(
+ count=1,
+ service_host='240.0.0.0',
+ )
+ with self.assertTakesBetween(0.1, 0.5):
+ with self.assertRaises(arvados.errors.KeepWriteError):
+ self.keepClient((0.1, 1)).put(self.DATA, copies=1, num_retries=0)
+
+ def test_timeout_slow_request(self):
+ self.server.setdelays(request=0.2)
+ self._test_200ms()
+
+ def test_timeout_slow_response(self):
+ self.server.setdelays(response=0.2)
+ self._test_200ms()
+
+ def test_timeout_slow_response_body(self):
+ self.server.setdelays(response_body=0.2)
+ self._test_200ms()
+
+ def _test_200ms(self):
+ """Connect should be t<100ms, request should be 200ms <= t < 300ms"""
+
+ # Allow 100ms to connect, then 1s for response. Everything
+ # should work, and everything should take at least 200ms to
+ # return.
+ kc = self.keepClient((.1, 1))
+ with self.assertTakesBetween(.2, .3):
+ loc = kc.put(self.DATA, copies=1, num_retries=0)
+ with self.assertTakesBetween(.2, .3):
+ self.assertEqual(self.DATA, kc.get(loc, num_retries=0))
+
+ # Allow 1s to connect, then 100ms for response. Nothing should
+ # work, and everything should take at least 100ms to return.
+ kc = self.keepClient((1, .1))
+ with self.assertTakesBetween(.1, .2):
+ with self.assertRaises(arvados.errors.KeepReadError):
+ kc.get(loc, num_retries=0)
+ with self.assertTakesBetween(.1, .2):
+ with self.assertRaises(arvados.errors.KeepWriteError):
+ kc.put(self.DATA, copies=1, num_retries=0)
class KeepClientGatewayTestCase(unittest.TestCase, tutil.ApiClientMock):
projects / arvados.git / blobdiff