1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
|
# The MIT License (MIT)
# Copyright (c) Microsoft Corporation. All rights reserved.
import asyncio
import os
import unittest
import uuid
import pytest
import pytest_asyncio
from azure.core.exceptions import ServiceResponseError
import test_config
from azure.cosmos import _partition_health_tracker, documents
from azure.cosmos.aio import CosmosClient
from azure.cosmos.exceptions import CosmosHttpResponseError
from azure.cosmos.http_constants import ResourceType
from test_per_partition_circuit_breaker_mm_async import (create_doc, PK_VALUE, create_errors,
DELETE_ALL_ITEMS_BY_PARTITION_KEY,
validate_unhealthy_partitions as validate_unhealthy_partitions_mm,
perform_write_operation, cleanup_method)
from test_per_partition_circuit_breaker_sm_mrr_async import validate_unhealthy_partitions as validate_unhealthy_partitions_sm_mrr
from _fault_injection_transport_async import FaultInjectionTransportAsync
COLLECTION = "created_collection"
@pytest_asyncio.fixture(scope="class", autouse=True)
async def setup_teardown():
os.environ["AZURE_COSMOS_ENABLE_CIRCUIT_BREAKER"] = "True"
yield
os.environ["AZURE_COSMOS_ENABLE_CIRCUIT_BREAKER"] = "False"
async def create_custom_transport_mm():
custom_transport = FaultInjectionTransportAsync()
is_get_account_predicate = lambda r: FaultInjectionTransportAsync.predicate_is_database_account_call(r)
emulator_as_multi_write_region_account_transformation = \
lambda r, inner: FaultInjectionTransportAsync.transform_topology_mwr(
first_region_name="Write Region",
second_region_name="Read Region",
inner=inner)
custom_transport.add_response_transformation(
is_get_account_predicate,
emulator_as_multi_write_region_account_transformation)
return custom_transport
@pytest.mark.cosmosEmulator
@pytest.mark.asyncio
@pytest.mark.usefixtures("setup_teardown")
class TestCircuitBreakerEmulatorAsync:
host = test_config.TestConfig.host
master_key = test_config.TestConfig.masterKey
connectionPolicy = test_config.TestConfig.connectionPolicy
TEST_DATABASE_ID = test_config.TestConfig.TEST_DATABASE_ID
TEST_CONTAINER_SINGLE_PARTITION_ID = test_config.TestConfig.TEST_MULTI_PARTITION_CONTAINER_ID
async def setup_method_with_custom_transport(self, custom_transport, default_endpoint=host, **kwargs):
client = CosmosClient(default_endpoint, self.master_key, consistency_level="Session",
preferred_locations=["Write Region", "Read Region"],
transport=custom_transport, **kwargs)
db = client.get_database_client(self.TEST_DATABASE_ID)
container = db.get_container_client(self.TEST_CONTAINER_SINGLE_PARTITION_ID)
return {"client": client, "db": db, "col": container}
async def create_custom_transport_sm_mrr(self):
custom_transport = FaultInjectionTransportAsync()
# Inject rule to disallow writes in the read-only region
is_write_operation_in_read_region_predicate = lambda \
r: FaultInjectionTransportAsync.predicate_is_write_operation(r, self.host)
custom_transport.add_fault(
is_write_operation_in_read_region_predicate,
lambda r: FaultInjectionTransportAsync.error_write_forbidden())
# Inject topology transformation that would make Emulator look like a single write region
# account with two read regions
is_get_account_predicate = lambda r: FaultInjectionTransportAsync.predicate_is_database_account_call(r)
emulator_as_multi_region_sm_account_transformation = \
lambda r, inner: FaultInjectionTransportAsync.transform_topology_swr_mrr(
write_region_name="Write Region",
read_region_name="Read Region",
inner=inner)
custom_transport.add_response_transformation(
is_get_account_predicate,
emulator_as_multi_region_sm_account_transformation)
return custom_transport
async def setup_info(self, error, mm=False):
expected_uri = self.host
uri_down = self.host.replace("localhost", "127.0.0.1")
custom_transport = await create_custom_transport_mm() if mm else await self.create_custom_transport_sm_mrr()
# two documents targeted to same partition, one will always fail and the other will succeed
doc = create_doc()
predicate = lambda r: (FaultInjectionTransportAsync.predicate_is_resource_type(r, ResourceType.Collection) and
FaultInjectionTransportAsync.predicate_is_operation_type(r, documents._OperationType.Delete) and
FaultInjectionTransportAsync.predicate_targets_region(r, uri_down))
custom_transport.add_fault(predicate, error)
custom_setup = await self.setup_method_with_custom_transport(custom_transport, default_endpoint=self.host, multiple_write_locations=mm)
setup = await self.setup_method_with_custom_transport(None, default_endpoint=self.host, multiple_write_locations=mm)
return setup, doc, expected_uri, uri_down, custom_setup, custom_transport, predicate
@pytest.mark.parametrize("error", create_errors())
async def test_write_consecutive_failure_threshold_delete_all_items_by_pk_sm_async(self, setup_teardown, error):
error_lambda = lambda r: asyncio.create_task(FaultInjectionTransportAsync.error_after_delay(0, error))
setup, doc, expected_uri, uri_down, custom_setup, custom_transport, predicate = await self.setup_info(error_lambda)
fault_injection_container = custom_setup['col']
container = setup['col']
global_endpoint_manager = fault_injection_container.client_connection._global_endpoint_manager
# writes should fail in sm mrr with circuit breaker and should not mark unavailable a partition
for i in range(6):
validate_unhealthy_partitions_sm_mrr(global_endpoint_manager, 0)
with pytest.raises((CosmosHttpResponseError, ServiceResponseError)):
await perform_write_operation(
DELETE_ALL_ITEMS_BY_PARTITION_KEY,
container,
fault_injection_container,
str(uuid.uuid4()),
PK_VALUE,
expected_uri,
)
validate_unhealthy_partitions_sm_mrr(global_endpoint_manager, 0)
await cleanup_method([custom_setup, setup])
@pytest.mark.parametrize("error", create_errors())
async def test_write_consecutive_failure_threshold_delete_all_items_by_pk_mm_async(self, setup_teardown, error):
error_lambda = lambda r: asyncio.create_task(FaultInjectionTransportAsync.error_after_delay(0, error))
setup, doc, expected_uri, uri_down, custom_setup, custom_transport, predicate = await self.setup_info(error_lambda, mm=True)
fault_injection_container = custom_setup['col']
container = setup['col']
global_endpoint_manager = fault_injection_container.client_connection._global_endpoint_manager
with pytest.raises((CosmosHttpResponseError, ServiceResponseError)) as exc_info:
await perform_write_operation(DELETE_ALL_ITEMS_BY_PARTITION_KEY,
container,
fault_injection_container,
str(uuid.uuid4()),
PK_VALUE,
expected_uri)
assert exc_info.value == error
validate_unhealthy_partitions_mm(global_endpoint_manager, 0)
# writes should fail but still be tracked
for i in range(4):
with pytest.raises((CosmosHttpResponseError, ServiceResponseError)) as exc_info:
await perform_write_operation(DELETE_ALL_ITEMS_BY_PARTITION_KEY,
container,
fault_injection_container,
str(uuid.uuid4()),
PK_VALUE,
expected_uri)
assert exc_info.value == error
# writes should now succeed because going to the other region
await perform_write_operation(DELETE_ALL_ITEMS_BY_PARTITION_KEY,
container,
fault_injection_container,
str(uuid.uuid4()),
PK_VALUE,
expected_uri)
validate_unhealthy_partitions_mm(global_endpoint_manager, 1)
# remove faults and reduce initial recover time and perform a write
original_unavailable_time = _partition_health_tracker.INITIAL_UNAVAILABLE_TIME
_partition_health_tracker.INITIAL_UNAVAILABLE_TIME = 1
custom_transport.faults = []
try:
await perform_write_operation(DELETE_ALL_ITEMS_BY_PARTITION_KEY,
container,
fault_injection_container,
str(uuid.uuid4()),
PK_VALUE,
uri_down)
finally:
_partition_health_tracker.INITIAL_UNAVAILABLE_TIME = original_unavailable_time
validate_unhealthy_partitions_mm(global_endpoint_manager, 0)
await cleanup_method([custom_setup, setup])
@pytest.mark.parametrize("error", create_errors())
async def test_write_failure_rate_threshold_delete_all_items_by_pk_mm_async(self, setup_teardown, error):
error_lambda = lambda r: asyncio.create_task(FaultInjectionTransportAsync.error_after_delay(0, error))
setup, doc, expected_uri, uri_down, custom_setup, custom_transport, predicate = await self.setup_info(error_lambda, mm=True)
fault_injection_container = custom_setup['col']
container = setup['col']
global_endpoint_manager = fault_injection_container.client_connection._global_endpoint_manager
# lower minimum requests for testing
_partition_health_tracker.MINIMUM_REQUESTS_FOR_FAILURE_RATE = 10
os.environ["AZURE_COSMOS_FAILURE_PERCENTAGE_TOLERATED"] = "80"
try:
# writes should fail but still be tracked and mark unavailable a partition after crossing threshold
for i in range(10):
validate_unhealthy_partitions_mm(global_endpoint_manager, 0)
if i == 4 or i == 8:
# perform some successful creates to reset consecutive counter
# remove faults and perform a write
custom_transport.faults = []
await fault_injection_container.upsert_item(body=doc)
custom_transport.add_fault(predicate,
lambda r: asyncio.create_task(FaultInjectionTransportAsync.error_after_delay(
0,
error
)))
else:
with pytest.raises((CosmosHttpResponseError, ServiceResponseError)) as exc_info:
await perform_write_operation(DELETE_ALL_ITEMS_BY_PARTITION_KEY,
container,
fault_injection_container,
str(uuid.uuid4()),
PK_VALUE,
expected_uri)
assert exc_info.value == error
validate_unhealthy_partitions_mm(global_endpoint_manager, 1)
finally:
os.environ["AZURE_COSMOS_FAILURE_PERCENTAGE_TOLERATED"] = "90"
# restore minimum requests
_partition_health_tracker.MINIMUM_REQUESTS_FOR_FAILURE_RATE = 100
await cleanup_method([custom_setup, setup])
@pytest.mark.parametrize("error", create_errors())
async def test_write_failure_rate_threshold_delete_all_items_by_pk_sm_async(self, setup_teardown, error):
error_lambda = lambda r: asyncio.create_task(FaultInjectionTransportAsync.error_after_delay(0, error))
setup, doc, expected_uri, uri_down, custom_setup, custom_transport, predicate = await self.setup_info(error_lambda)
fault_injection_container = custom_setup['col']
container = setup['col']
global_endpoint_manager = fault_injection_container.client_connection._global_endpoint_manager
# lower minimum requests for testing
_partition_health_tracker.MINIMUM_REQUESTS_FOR_FAILURE_RATE = 10
os.environ["AZURE_COSMOS_FAILURE_PERCENTAGE_TOLERATED"] = "80"
try:
# writes should fail but still be tracked and mark unavailable a partition after crossing threshold
for i in range(10):
validate_unhealthy_partitions_sm_mrr(global_endpoint_manager, 0)
if i == 4 or i == 8:
# perform some successful creates to reset consecutive counter
# remove faults and perform a write
custom_transport.faults = []
await fault_injection_container.upsert_item(body=doc)
custom_transport.add_fault(predicate,
lambda r: asyncio.create_task(FaultInjectionTransportAsync.error_after_delay(
0,
error
)))
else:
with pytest.raises((CosmosHttpResponseError, ServiceResponseError)) as exc_info:
await perform_write_operation(DELETE_ALL_ITEMS_BY_PARTITION_KEY,
container,
fault_injection_container,
str(uuid.uuid4()),
PK_VALUE,
expected_uri)
assert exc_info.value == error
validate_unhealthy_partitions_sm_mrr(global_endpoint_manager, 0)
finally:
os.environ["AZURE_COSMOS_FAILURE_PERCENTAGE_TOLERATED"] = "90"
# restore minimum requests
_partition_health_tracker.MINIMUM_REQUESTS_FOR_FAILURE_RATE = 100
await cleanup_method([custom_setup, setup])
# test cosmos client timeout
if __name__ == '__main__':
unittest.main()
|
