|
1
|
|
|
import time |
|
2
|
|
|
import pdb |
|
3
|
|
|
import threading |
|
4
|
|
|
import logging |
|
5
|
|
|
from multiprocessing import Pool, Process |
|
6
|
|
|
import pytest |
|
7
|
|
|
from utils import * |
|
8
|
|
|
|
|
9
|
|
|
dim = 128 |
|
10
|
|
|
index_file_size = 10 |
|
11
|
|
|
COMPACT_TIMEOUT = 180 |
|
12
|
|
|
nprobe = 1 |
|
13
|
|
|
top_k = 1 |
|
14
|
|
|
tag = "1970-01-01" |
|
15
|
|
|
nb = 6000 |
|
16
|
|
|
segment_row_count = 5000 |
|
17
|
|
|
entity = gen_entities(1) |
|
|
|
|
|
|
18
|
|
|
entities = gen_entities(nb) |
|
19
|
|
|
raw_vector, binary_entity = gen_binary_entities(1) |
|
|
|
|
|
|
20
|
|
|
raw_vectors, binary_entities = gen_binary_entities(nb) |
|
21
|
|
|
default_fields = gen_default_fields() |
|
|
|
|
|
|
22
|
|
|
field_name = default_float_vec_field_name |
|
|
|
|
|
|
23
|
|
|
default_single_query = { |
|
24
|
|
|
"bool": { |
|
25
|
|
|
"must": [ |
|
26
|
|
|
{"vector": {field_name: {"topk": 10, "query": gen_vectors(1, dim), |
|
|
|
|
|
|
27
|
|
|
"params": {"nprobe": 10}}}} |
|
28
|
|
|
] |
|
29
|
|
|
} |
|
30
|
|
|
} |
|
31
|
|
|
|
|
32
|
|
|
def ip_query(): |
|
33
|
|
|
query = copy.deepcopy(default_single_query) |
|
|
|
|
|
|
34
|
|
|
query["bool"]["must"][0]["vector"][field_name].update({"metric_type": "IP"}) |
|
35
|
|
|
return query |
|
36
|
|
|
|
|
37
|
|
|
|
|
38
|
|
|
class TestCompactBase: |
|
39
|
|
|
""" |
|
40
|
|
|
****************************************************************** |
|
41
|
|
|
The following cases are used to test `compact` function |
|
42
|
|
|
****************************************************************** |
|
43
|
|
|
""" |
|
44
|
|
|
@pytest.mark.timeout(COMPACT_TIMEOUT) |
|
45
|
|
|
def test_compact_collection_name_None(self, connect, collection): |
|
46
|
|
|
''' |
|
47
|
|
|
target: compact collection where collection name is None |
|
48
|
|
|
method: compact with the collection_name: None |
|
49
|
|
|
expected: exception raised |
|
50
|
|
|
''' |
|
51
|
|
|
collection_name = None |
|
52
|
|
|
with pytest.raises(Exception) as e: |
|
53
|
|
|
status = connect.compact(collection_name) |
|
54
|
|
|
|
|
55
|
|
|
@pytest.mark.timeout(COMPACT_TIMEOUT) |
|
56
|
|
|
def test_compact_collection_name_not_existed(self, connect, collection): |
|
57
|
|
|
''' |
|
58
|
|
|
target: compact collection not existed |
|
59
|
|
|
method: compact with a random collection_name, which is not in db |
|
60
|
|
|
expected: exception raised |
|
61
|
|
|
''' |
|
62
|
|
|
collection_name = gen_unique_str("not_existed") |
|
|
|
|
|
|
63
|
|
|
with pytest.raises(Exception) as e: |
|
64
|
|
|
status = connect.compact(collection_name) |
|
65
|
|
|
|
|
66
|
|
|
@pytest.fixture( |
|
67
|
|
|
scope="function", |
|
68
|
|
|
params=gen_invalid_strs() |
|
|
|
|
|
|
69
|
|
|
) |
|
70
|
|
|
def get_collection_name(self, request): |
|
71
|
|
|
yield request.param |
|
72
|
|
|
|
|
73
|
|
|
@pytest.mark.timeout(COMPACT_TIMEOUT) |
|
74
|
|
|
def test_compact_collection_name_invalid(self, connect, get_collection_name): |
|
75
|
|
|
''' |
|
76
|
|
|
target: compact collection with invalid name |
|
77
|
|
|
method: compact with invalid collection_name |
|
78
|
|
|
expected: exception raised |
|
79
|
|
|
''' |
|
80
|
|
|
collection_name = get_collection_name |
|
81
|
|
|
with pytest.raises(Exception) as e: |
|
82
|
|
|
status = connect.compact(collection_name) |
|
83
|
|
|
# assert not status.OK() |
|
84
|
|
|
|
|
85
|
|
|
@pytest.mark.level(2) |
|
86
|
|
|
@pytest.mark.timeout(COMPACT_TIMEOUT) |
|
87
|
|
|
def test_add_entity_and_compact(self, connect, collection): |
|
88
|
|
|
''' |
|
89
|
|
|
target: test add entity and compact |
|
90
|
|
|
method: add entity and compact collection |
|
91
|
|
|
expected: data_size before and after Compact |
|
92
|
|
|
''' |
|
93
|
|
|
# vector = gen_single_vector(dim) |
|
94
|
|
|
ids = connect.insert(collection, entity) |
|
95
|
|
|
assert len(ids) == 1 |
|
96
|
|
|
connect.flush([collection]) |
|
97
|
|
|
# get collection info before compact |
|
98
|
|
|
info = connect.get_collection_stats(collection) |
|
99
|
|
|
logging.getLogger().info(info) |
|
100
|
|
|
size_before = info["partitions"][0]["segments"][0]["data_size"] |
|
101
|
|
|
status = connect.compact(collection) |
|
102
|
|
|
assert status.OK() |
|
103
|
|
|
# get collection info after compact |
|
104
|
|
|
info = connect.get_collection_stats(collection) |
|
105
|
|
|
size_after = info["partitions"][0]["segments"][0]["data_size"] |
|
106
|
|
|
assert(size_before == size_after) |
|
107
|
|
|
|
|
108
|
|
View Code Duplication |
@pytest.mark.timeout(COMPACT_TIMEOUT) |
|
|
|
|
|
|
109
|
|
|
def test_insert_and_compact(self, connect, collection): |
|
110
|
|
|
''' |
|
111
|
|
|
target: test add entities and compact |
|
112
|
|
|
method: add entities and compact collection |
|
113
|
|
|
expected: data_size before and after Compact |
|
114
|
|
|
''' |
|
115
|
|
|
# entities = gen_vector(nb, dim) |
|
116
|
|
|
ids = connect.insert(collection, entities) |
|
117
|
|
|
connect.flush([collection]) |
|
118
|
|
|
# get collection info before compact |
|
119
|
|
|
info = connect.get_collection_stats(collection) |
|
120
|
|
|
# assert status.OK() |
|
121
|
|
|
size_before = info["partitions"][0]["segments"][0]["data_size"] |
|
122
|
|
|
status = connect.compact(collection) |
|
123
|
|
|
assert status.OK() |
|
124
|
|
|
# get collection info after compact |
|
125
|
|
|
info = connect.get_collection_stats(collection) |
|
126
|
|
|
# assert status.OK() |
|
127
|
|
|
size_after = info["partitions"][0]["segments"][0]["data_size"] |
|
128
|
|
|
assert(size_before == size_after) |
|
129
|
|
|
|
|
130
|
|
View Code Duplication |
@pytest.mark.timeout(COMPACT_TIMEOUT) |
|
|
|
|
|
|
131
|
|
|
@pytest.mark.skip(reason="delete not support yet") |
|
132
|
|
|
def test_insert_delete_part_and_compact(self, connect, collection): |
|
133
|
|
|
''' |
|
134
|
|
|
target: test add entities, delete part of them and compact |
|
135
|
|
|
method: add entities, delete a few and compact collection |
|
136
|
|
|
expected: status ok, data size maybe is smaller after compact |
|
137
|
|
|
''' |
|
138
|
|
|
ids = connect.insert(collection, entities) |
|
139
|
|
|
assert len(ids) == nb |
|
140
|
|
|
connect.flush([collection]) |
|
141
|
|
|
delete_ids = [ids[0], ids[-1]] |
|
142
|
|
|
status = connect.delete_entity_by_id(collection, delete_ids) |
|
143
|
|
|
assert status.OK() |
|
144
|
|
|
connect.flush([collection]) |
|
145
|
|
|
# get collection info before compact |
|
146
|
|
|
info = connect.get_collection_stats(collection) |
|
147
|
|
|
logging.getLogger().info(info["partitions"]) |
|
148
|
|
|
size_before = info["partitions"][0]["segments"][0]["data_size"] |
|
149
|
|
|
logging.getLogger().info(size_before) |
|
150
|
|
|
status = connect.compact(collection) |
|
151
|
|
|
assert status.OK() |
|
152
|
|
|
# get collection info after compact |
|
153
|
|
|
info = connect.get_collection_stats(collection) |
|
154
|
|
|
logging.getLogger().info(info["partitions"]) |
|
155
|
|
|
size_after = info["partitions"][0]["segments"][0]["data_size"] |
|
156
|
|
|
logging.getLogger().info(size_after) |
|
157
|
|
|
assert(size_before >= size_after) |
|
158
|
|
|
|
|
159
|
|
View Code Duplication |
@pytest.mark.timeout(COMPACT_TIMEOUT) |
|
|
|
|
|
|
160
|
|
|
@pytest.mark.skip(reason="delete not support yet") |
|
161
|
|
|
def test_insert_delete_all_and_compact(self, connect, collection): |
|
162
|
|
|
''' |
|
163
|
|
|
target: test add entities, delete them and compact |
|
164
|
|
|
method: add entities, delete all and compact collection |
|
165
|
|
|
expected: status ok, no data size in collection info because collection is empty |
|
166
|
|
|
''' |
|
167
|
|
|
ids = connect.insert(collection, entities) |
|
168
|
|
|
assert len(ids) == nb |
|
169
|
|
|
connect.flush([collection]) |
|
170
|
|
|
status = connect.delete_entity_by_id(collection, ids) |
|
171
|
|
|
assert status.OK() |
|
172
|
|
|
connect.flush([collection]) |
|
173
|
|
|
# get collection info before compact |
|
174
|
|
|
info = connect.get_collection_stats(collection) |
|
175
|
|
|
status = connect.compact(collection) |
|
176
|
|
|
assert status.OK() |
|
177
|
|
|
# get collection info after compact |
|
178
|
|
|
info = connect.get_collection_stats(collection) |
|
179
|
|
|
logging.getLogger().info(info["partitions"]) |
|
180
|
|
|
assert not info["partitions"][0]["segments"] |
|
181
|
|
|
|
|
182
|
|
|
@pytest.mark.timeout(COMPACT_TIMEOUT) |
|
183
|
|
|
@pytest.mark.skip(reason="delete not support yet") |
|
184
|
|
|
def test_insert_partition_delete_half_and_compact(self, connect, collection): |
|
185
|
|
|
''' |
|
186
|
|
|
target: test add entities into partition, delete them and compact |
|
187
|
|
|
method: add entities, delete half of entities in partition and compact collection |
|
188
|
|
|
expected: status ok, data_size less than the older version |
|
189
|
|
|
''' |
|
190
|
|
|
connect.create_partition(collection, tag) |
|
191
|
|
|
assert connect.has_partition(collection, tag) |
|
192
|
|
|
ids = connect.insert(collection, entities, partition_tag=tag) |
|
193
|
|
|
connect.flush([collection]) |
|
194
|
|
|
info = connect.get_collection_stats(collection) |
|
195
|
|
|
logging.getLogger().info(info["partitions"]) |
|
196
|
|
|
|
|
197
|
|
|
delete_ids = ids[:3000] |
|
198
|
|
|
status = connect.delete_entity_by_id(collection, delete_ids) |
|
199
|
|
|
assert status.OK() |
|
200
|
|
|
connect.flush([collection]) |
|
201
|
|
|
# get collection info before compact |
|
202
|
|
|
info = connect.get_collection_stats(collection) |
|
203
|
|
|
logging.getLogger().info(info["partitions"]) |
|
204
|
|
|
status = connect.compact(collection) |
|
205
|
|
|
assert status.OK() |
|
206
|
|
|
# get collection info after compact |
|
207
|
|
|
info_after = connect.get_collection_stats(collection) |
|
208
|
|
|
logging.getLogger().info(info_after["partitions"]) |
|
209
|
|
|
assert info["partitions"][1]["segments"][0]["data_size"] > info_after["partitions"][1]["segments"][0]["data_size"] |
|
210
|
|
|
|
|
211
|
|
View Code Duplication |
@pytest.fixture( |
|
|
|
|
|
|
212
|
|
|
scope="function", |
|
213
|
|
|
params=gen_simple_index() |
|
|
|
|
|
|
214
|
|
|
) |
|
215
|
|
|
def get_simple_index(self, request, connect): |
|
216
|
|
|
if str(connect._cmd("mode")) == "GPU": |
|
217
|
|
|
if not request.param["index_type"] not in ivf(): |
|
|
|
|
|
|
218
|
|
|
pytest.skip("Only support index_type: idmap/ivf") |
|
219
|
|
|
if str(connect._cmd("mode")) == "CPU": |
|
220
|
|
|
if request.param["index_type"] in index_cpu_not_support(): |
|
|
|
|
|
|
221
|
|
|
pytest.skip("CPU not support index_type: ivf_sq8h") |
|
222
|
|
|
return request.param |
|
223
|
|
|
|
|
224
|
|
|
@pytest.mark.skip(reason="create_index not support yet") |
|
225
|
|
|
def test_compact_after_index_created(self, connect, collection, get_simple_index): |
|
226
|
|
|
''' |
|
227
|
|
|
target: test compact collection after index created |
|
228
|
|
|
method: add entities, create index, delete part of entities and compact |
|
229
|
|
|
expected: status ok, index description no change, data size smaller after compact |
|
230
|
|
|
''' |
|
231
|
|
|
count = 10 |
|
232
|
|
|
ids = connect.insert(collection, entities) |
|
233
|
|
|
connect.flush([collection]) |
|
234
|
|
|
connect.create_index(collection, field_name, get_simple_index) |
|
235
|
|
|
connect.flush([collection]) |
|
236
|
|
|
# get collection info before compact |
|
237
|
|
|
info = connect.get_collection_stats(collection) |
|
238
|
|
|
size_before = info["partitions"][0]["segments"][0]["data_size"] |
|
239
|
|
|
logging.getLogger().info(info["partitions"]) |
|
240
|
|
|
delete_ids = [ids[0], ids[-1]] |
|
241
|
|
|
status = connect.delete_entity_by_id(collection, delete_ids) |
|
242
|
|
|
assert status.OK() |
|
243
|
|
|
connect.flush([collection]) |
|
244
|
|
|
status = connect.compact(collection) |
|
245
|
|
|
assert status.OK() |
|
246
|
|
|
# get collection info after compact |
|
247
|
|
|
info = connect.get_collection_stats(collection) |
|
248
|
|
|
logging.getLogger().info(info["partitions"]) |
|
249
|
|
|
size_after = info["partitions"][0]["segments"][0]["data_size"] |
|
250
|
|
|
assert(size_before >= size_after) |
|
251
|
|
|
|
|
252
|
|
View Code Duplication |
@pytest.mark.timeout(COMPACT_TIMEOUT) |
|
|
|
|
|
|
253
|
|
|
def test_add_entity_and_compact_twice(self, connect, collection): |
|
254
|
|
|
''' |
|
255
|
|
|
target: test add entity and compact twice |
|
256
|
|
|
method: add entity and compact collection twice |
|
257
|
|
|
expected: status ok, data size no change |
|
258
|
|
|
''' |
|
259
|
|
|
ids = connect.insert(collection, entity) |
|
260
|
|
|
connect.flush([collection]) |
|
261
|
|
|
# get collection info before compact |
|
262
|
|
|
info = connect.get_collection_stats(collection) |
|
263
|
|
|
size_before = info["partitions"][0]["segments"][0]["data_size"] |
|
264
|
|
|
status = connect.compact(collection) |
|
265
|
|
|
assert status.OK() |
|
266
|
|
|
connect.flush([collection]) |
|
267
|
|
|
# get collection info after compact |
|
268
|
|
|
info = connect.get_collection_stats(collection) |
|
269
|
|
|
size_after = info["partitions"][0]["segments"][0]["data_size"] |
|
270
|
|
|
assert(size_before == size_after) |
|
271
|
|
|
status = connect.compact(collection) |
|
272
|
|
|
assert status.OK() |
|
273
|
|
|
# get collection info after compact twice |
|
274
|
|
|
info = connect.get_collection_stats(collection) |
|
275
|
|
|
size_after_twice = info["partitions"][0]["segments"][0]["data_size"] |
|
276
|
|
|
assert(size_after == size_after_twice) |
|
277
|
|
|
|
|
278
|
|
View Code Duplication |
@pytest.mark.timeout(COMPACT_TIMEOUT) |
|
|
|
|
|
|
279
|
|
|
@pytest.mark.skip(reason="delete not support yet") |
|
280
|
|
|
def test_insert_delete_part_and_compact_twice(self, connect, collection): |
|
281
|
|
|
''' |
|
282
|
|
|
target: test add entities, delete part of them and compact twice |
|
283
|
|
|
method: add entities, delete part and compact collection twice |
|
284
|
|
|
expected: status ok, data size smaller after first compact, no change after second |
|
285
|
|
|
''' |
|
286
|
|
|
ids = connect.insert(collection, entities) |
|
287
|
|
|
connect.flush([collection]) |
|
288
|
|
|
delete_ids = [ids[0], ids[-1]] |
|
289
|
|
|
status = connect.delete_entity_by_id(collection, delete_ids) |
|
290
|
|
|
assert status.OK() |
|
291
|
|
|
connect.flush([collection]) |
|
292
|
|
|
# get collection info before compact |
|
293
|
|
|
info = connect.get_collection_stats(collection) |
|
294
|
|
|
size_before = info["partitions"][0]["segments"][0]["data_size"] |
|
295
|
|
|
status = connect.compact(collection) |
|
296
|
|
|
assert status.OK() |
|
297
|
|
|
# get collection info after compact |
|
298
|
|
|
info = connect.get_collection_stats(collection) |
|
299
|
|
|
size_after = info["partitions"][0]["segments"][0]["data_size"] |
|
300
|
|
|
assert(size_before >= size_after) |
|
301
|
|
|
status = connect.compact(collection) |
|
302
|
|
|
assert status.OK() |
|
303
|
|
|
# get collection info after compact twice |
|
304
|
|
|
info = connect.get_collection_stats(collection) |
|
305
|
|
|
size_after_twice = info["partitions"][0]["segments"][0]["data_size"] |
|
306
|
|
|
assert(size_after == size_after_twice) |
|
307
|
|
|
|
|
308
|
|
|
@pytest.mark.timeout(COMPACT_TIMEOUT) |
|
309
|
|
|
def test_compact_multi_collections(self, connect): |
|
310
|
|
|
''' |
|
311
|
|
|
target: test compact works or not with multiple collections |
|
312
|
|
|
method: create 50 collections, add entities into them and compact in turn |
|
313
|
|
|
expected: status ok |
|
314
|
|
|
''' |
|
315
|
|
|
nq = 100 |
|
316
|
|
|
num_collections = 50 |
|
317
|
|
|
entities = gen_entities(nq) |
|
|
|
|
|
|
318
|
|
|
collection_list = [] |
|
319
|
|
|
for i in range(num_collections): |
|
320
|
|
|
collection_name = gen_unique_str("test_compact_multi_collection_%d" % i) |
|
|
|
|
|
|
321
|
|
|
collection_list.append(collection_name) |
|
322
|
|
|
connect.create_collection(collection_name, default_fields) |
|
323
|
|
|
time.sleep(6) |
|
324
|
|
|
for i in range(num_collections): |
|
325
|
|
|
ids = connect.insert(collection_list[i], entities) |
|
326
|
|
|
status = connect.compact(collection_list[i]) |
|
327
|
|
|
assert status.OK() |
|
328
|
|
|
|
|
329
|
|
View Code Duplication |
@pytest.mark.level(2) |
|
|
|
|
|
|
330
|
|
|
@pytest.mark.timeout(COMPACT_TIMEOUT) |
|
331
|
|
|
def test_add_entity_after_compact(self, connect, collection): |
|
332
|
|
|
''' |
|
333
|
|
|
target: test add entity after compact |
|
334
|
|
|
method: after compact operation, add entity |
|
335
|
|
|
expected: status ok, entity added |
|
336
|
|
|
''' |
|
337
|
|
|
ids = connect.insert(collection, entities) |
|
338
|
|
|
assert len(ids) == nb |
|
339
|
|
|
connect.flush([collection]) |
|
340
|
|
|
# get collection info before compact |
|
341
|
|
|
info = connect.get_collection_stats(collection) |
|
342
|
|
|
size_before = info["partitions"][0]["segments"][0]["data_size"] |
|
343
|
|
|
status = connect.compact(collection) |
|
344
|
|
|
assert status.OK() |
|
345
|
|
|
# get collection info after compact |
|
346
|
|
|
info = connect.get_collection_stats(collection) |
|
347
|
|
|
size_after = info["partitions"][0]["segments"][0]["data_size"] |
|
348
|
|
|
assert(size_before == size_after) |
|
349
|
|
|
ids = connect.insert(collection, entity) |
|
350
|
|
|
connect.flush([collection]) |
|
351
|
|
|
res = connect.count_entities(collection) |
|
352
|
|
|
assert res == nb+1 |
|
353
|
|
|
|
|
354
|
|
|
@pytest.mark.skip(reason="delete not support yet") |
|
355
|
|
|
@pytest.mark.timeout(COMPACT_TIMEOUT) |
|
356
|
|
|
def test_index_creation_after_compact(self, connect, collection, get_simple_index): |
|
357
|
|
|
''' |
|
358
|
|
|
target: test index creation after compact |
|
359
|
|
|
method: after compact operation, create index |
|
360
|
|
|
expected: status ok, index description no change |
|
361
|
|
|
''' |
|
362
|
|
|
ids = connect.insert(collection, entities) |
|
363
|
|
|
connect.flush([collection]) |
|
364
|
|
|
status = connect.delete_entity_by_id(collection, ids[:10]) |
|
365
|
|
|
assert status.OK() |
|
366
|
|
|
connect.flush([collection]) |
|
367
|
|
|
status = connect.compact(collection) |
|
368
|
|
|
assert status.OK() |
|
369
|
|
|
status = connect.create_index(collection, field_name, get_simple_index) |
|
370
|
|
|
assert status.OK() |
|
371
|
|
|
# status, result = connect.get_index_info(collection) |
|
372
|
|
|
|
|
373
|
|
|
@pytest.mark.timeout(COMPACT_TIMEOUT) |
|
374
|
|
|
@pytest.mark.skip(reason="delete not support yet") |
|
375
|
|
|
def test_delete_entities_after_compact(self, connect, collection): |
|
376
|
|
|
''' |
|
377
|
|
|
target: test delete entities after compact |
|
378
|
|
|
method: after compact operation, delete entities |
|
379
|
|
|
expected: status ok, entities deleted |
|
380
|
|
|
''' |
|
381
|
|
|
ids = connect.insert(collection, entities) |
|
382
|
|
|
assert len(ids) == nb |
|
383
|
|
|
connect.flush([collection]) |
|
384
|
|
|
status = connect.compact(collection) |
|
385
|
|
|
assert status.OK() |
|
386
|
|
|
connect.flush([collection]) |
|
387
|
|
|
status = connect.delete_entity_by_id(collection, ids) |
|
388
|
|
|
assert status.OK() |
|
389
|
|
|
connect.flush([collection]) |
|
390
|
|
|
assert connect.count_entities(collection) == 0 |
|
391
|
|
|
|
|
392
|
|
|
@pytest.mark.skip(reason="search not support yet") |
|
393
|
|
|
@pytest.mark.timeout(COMPACT_TIMEOUT) |
|
394
|
|
|
def test_search_after_compact(self, connect, collection): |
|
395
|
|
|
''' |
|
396
|
|
|
target: test search after compact |
|
397
|
|
|
method: after compact operation, search vector |
|
398
|
|
|
expected: status ok |
|
399
|
|
|
''' |
|
400
|
|
|
ids = connect.insert(collection, entities) |
|
401
|
|
|
assert len(ids) == nb |
|
402
|
|
|
connect.flush([collection]) |
|
403
|
|
|
status = connect.compact(collection) |
|
404
|
|
|
assert status.OK() |
|
405
|
|
|
query = copy.deepcopy(default_single_query) |
|
|
|
|
|
|
406
|
|
|
query["bool"]["must"][0]["vector"][field_name]["query"] = [entity[-1]["values"][0], entities[-1]["values"][0], |
|
407
|
|
|
entities[-1]["values"][-1]] |
|
408
|
|
|
res = connect.search(collection, query) |
|
409
|
|
|
logging.getLogger().debug(res) |
|
410
|
|
|
assert len(res) == len(query["bool"]["must"][0]["vector"][field_name]["query"]) |
|
|
|
|
|
|
411
|
|
|
assert res[0]._distances[0] > epsilon |
|
|
|
|
|
|
412
|
|
|
assert res[1]._distances[0] < epsilon |
|
413
|
|
|
assert res[2]._distances[0] < epsilon |
|
414
|
|
|
|
|
415
|
|
|
# TODO: enable |
|
416
|
|
|
@pytest.mark.skip(reason="delete not support yet") |
|
417
|
|
|
def _test_compact_server_crashed_recovery(self, connect, collection): |
|
418
|
|
|
''' |
|
419
|
|
|
target: test compact when server crashed unexpectedly and restarted |
|
420
|
|
|
method: add entities, delete and compact collection; server stopped and restarted during compact |
|
421
|
|
|
expected: status ok, request recovered |
|
422
|
|
|
''' |
|
423
|
|
|
entities = gen_vector(nb * 100, dim) |
|
|
|
|
|
|
424
|
|
|
status, ids = connect.insert(collection, entities) |
|
425
|
|
|
assert status.OK() |
|
426
|
|
|
status = connect.flush([collection]) |
|
427
|
|
|
assert status.OK() |
|
428
|
|
|
delete_ids = ids[0:1000] |
|
429
|
|
|
status = connect.delete_entity_by_id(collection, delete_ids) |
|
430
|
|
|
assert status.OK() |
|
431
|
|
|
status = connect.flush([collection]) |
|
432
|
|
|
assert status.OK() |
|
433
|
|
|
# start to compact, kill and restart server |
|
434
|
|
|
logging.getLogger().info("compact starting...") |
|
435
|
|
|
status = connect.compact(collection) |
|
436
|
|
|
# pdb.set_trace() |
|
437
|
|
|
assert status.OK() |
|
438
|
|
|
# get collection info after compact |
|
439
|
|
|
status, info = connect.get_collection_stats(collection) |
|
440
|
|
|
assert status.OK() |
|
441
|
|
|
assert info["partitions"][0].count == nb * 100 - 1000 |
|
442
|
|
|
|
|
443
|
|
|
|
|
444
|
|
|
class TestCompactJAC: |
|
445
|
|
|
""" |
|
446
|
|
|
****************************************************************** |
|
447
|
|
|
The following cases are used to test `compact` function |
|
448
|
|
|
****************************************************************** |
|
449
|
|
|
""" |
|
450
|
|
View Code Duplication |
@pytest.mark.timeout(COMPACT_TIMEOUT) |
|
|
|
|
|
|
451
|
|
|
def test_add_entity_and_compact(self, connect, jac_collection): |
|
452
|
|
|
''' |
|
453
|
|
|
target: test add binary vector and compact |
|
454
|
|
|
method: add vector and compact collection |
|
455
|
|
|
expected: status ok, vector added |
|
456
|
|
|
''' |
|
457
|
|
|
ids = connect.insert(jac_collection, binary_entity) |
|
458
|
|
|
assert len(ids) == 1 |
|
459
|
|
|
connect.flush([jac_collection]) |
|
460
|
|
|
# get collection info before compact |
|
461
|
|
|
info = connect.get_collection_stats(jac_collection) |
|
462
|
|
|
size_before = info["partitions"][0]["segments"][0]["data_size"] |
|
463
|
|
|
status = connect.compact(jac_collection) |
|
464
|
|
|
assert status.OK() |
|
465
|
|
|
# get collection info after compact |
|
466
|
|
|
info = connect.get_collection_stats(jac_collection) |
|
467
|
|
|
size_after = info["partitions"][0]["segments"][0]["data_size"] |
|
468
|
|
|
assert(size_before == size_after) |
|
469
|
|
|
|
|
470
|
|
View Code Duplication |
@pytest.mark.timeout(COMPACT_TIMEOUT) |
|
|
|
|
|
|
471
|
|
|
def test_insert_and_compact(self, connect, jac_collection): |
|
472
|
|
|
''' |
|
473
|
|
|
target: test add entities with binary vector and compact |
|
474
|
|
|
method: add entities and compact collection |
|
475
|
|
|
expected: status ok, entities added |
|
476
|
|
|
''' |
|
477
|
|
|
ids = connect.insert(jac_collection, binary_entities) |
|
478
|
|
|
assert len(ids) == nb |
|
479
|
|
|
connect.flush([jac_collection]) |
|
480
|
|
|
# get collection info before compact |
|
481
|
|
|
info = connect.get_collection_stats(jac_collection) |
|
482
|
|
|
size_before = info["partitions"][0]["segments"][0]["data_size"] |
|
483
|
|
|
status = connect.compact(jac_collection) |
|
484
|
|
|
assert status.OK() |
|
485
|
|
|
# get collection info after compact |
|
486
|
|
|
info = connect.get_collection_stats(jac_collection) |
|
487
|
|
|
size_after = info["partitions"][0]["segments"][0]["data_size"] |
|
488
|
|
|
assert(size_before == size_after) |
|
489
|
|
|
|
|
490
|
|
View Code Duplication |
@pytest.mark.timeout(COMPACT_TIMEOUT) |
|
|
|
|
|
|
491
|
|
|
@pytest.mark.skip(reason="delete not support yet") |
|
492
|
|
|
def test_insert_delete_part_and_compact(self, connect, jac_collection): |
|
493
|
|
|
''' |
|
494
|
|
|
target: test add entities, delete part of them and compact |
|
495
|
|
|
method: add entities, delete a few and compact collection |
|
496
|
|
|
expected: status ok, data size is smaller after compact |
|
497
|
|
|
''' |
|
498
|
|
|
ids = connect.insert(jac_collection, binary_entities) |
|
499
|
|
|
assert len(ids) == nb |
|
500
|
|
|
connect.flush([jac_collection]) |
|
501
|
|
|
delete_ids = [ids[0], ids[-1]] |
|
502
|
|
|
status = connect.delete_entity_by_id(jac_collection, delete_ids) |
|
503
|
|
|
assert status.OK() |
|
504
|
|
|
connect.flush([jac_collection]) |
|
505
|
|
|
# get collection info before compact |
|
506
|
|
|
info = connect.get_collection_stats(jac_collection) |
|
507
|
|
|
logging.getLogger().info(info["partitions"]) |
|
508
|
|
|
size_before = info["partitions"][0]["segments"][0]["data_size"] |
|
509
|
|
|
logging.getLogger().info(size_before) |
|
510
|
|
|
status = connect.compact(jac_collection) |
|
511
|
|
|
assert status.OK() |
|
512
|
|
|
# get collection info after compact |
|
513
|
|
|
info = connect.get_collection_stats(jac_collection) |
|
514
|
|
|
logging.getLogger().info(info["partitions"]) |
|
515
|
|
|
size_after = info["partitions"][0]["segments"][0]["data_size"] |
|
516
|
|
|
logging.getLogger().info(size_after) |
|
517
|
|
|
assert(size_before >= size_after) |
|
518
|
|
|
|
|
519
|
|
View Code Duplication |
@pytest.mark.timeout(COMPACT_TIMEOUT) |
|
|
|
|
|
|
520
|
|
|
@pytest.mark.skip(reason="delete not support yet") |
|
521
|
|
|
def test_insert_delete_all_and_compact(self, connect, jac_collection): |
|
522
|
|
|
''' |
|
523
|
|
|
target: test add entities, delete them and compact |
|
524
|
|
|
method: add entities, delete all and compact collection |
|
525
|
|
|
expected: status ok, no data size in collection info because collection is empty |
|
526
|
|
|
''' |
|
527
|
|
|
ids = connect.insert(jac_collection, binary_entities) |
|
528
|
|
|
assert len(ids) == nb |
|
529
|
|
|
connect.flush([jac_collection]) |
|
530
|
|
|
status = connect.delete_entity_by_id(jac_collection, ids) |
|
531
|
|
|
assert status.OK() |
|
532
|
|
|
connect.flush([jac_collection]) |
|
533
|
|
|
# get collection info before compact |
|
534
|
|
|
info = connect.get_collection_stats(jac_collection) |
|
535
|
|
|
status = connect.compact(jac_collection) |
|
536
|
|
|
assert status.OK() |
|
537
|
|
|
# get collection info after compact |
|
538
|
|
|
info = connect.get_collection_stats(jac_collection) |
|
539
|
|
|
assert status.OK() |
|
540
|
|
|
logging.getLogger().info(info["partitions"]) |
|
541
|
|
|
assert not info["partitions"][0]["segments"] |
|
542
|
|
|
|
|
543
|
|
View Code Duplication |
@pytest.mark.timeout(COMPACT_TIMEOUT) |
|
|
|
|
|
|
544
|
|
|
def test_add_entity_and_compact_twice(self, connect, jac_collection): |
|
545
|
|
|
''' |
|
546
|
|
|
target: test add entity and compact twice |
|
547
|
|
|
method: add entity and compact collection twice |
|
548
|
|
|
expected: status ok |
|
549
|
|
|
''' |
|
550
|
|
|
ids = connect.insert(jac_collection, binary_entity) |
|
551
|
|
|
assert len(ids) == 1 |
|
552
|
|
|
connect.flush([jac_collection]) |
|
553
|
|
|
# get collection info before compact |
|
554
|
|
|
info = connect.get_collection_stats(jac_collection) |
|
555
|
|
|
size_before = info["partitions"][0]["segments"][0]["data_size"] |
|
556
|
|
|
status = connect.compact(jac_collection) |
|
557
|
|
|
assert status.OK() |
|
558
|
|
|
# get collection info after compact |
|
559
|
|
|
info = connect.get_collection_stats(jac_collection) |
|
560
|
|
|
size_after = info["partitions"][0]["segments"][0]["data_size"] |
|
561
|
|
|
assert(size_before == size_after) |
|
562
|
|
|
status = connect.compact(jac_collection) |
|
563
|
|
|
assert status.OK() |
|
564
|
|
|
# get collection info after compact twice |
|
565
|
|
|
info = connect.get_collection_stats(jac_collection) |
|
566
|
|
|
size_after_twice = info["partitions"][0]["segments"][0]["data_size"] |
|
567
|
|
|
assert(size_after == size_after_twice) |
|
568
|
|
|
|
|
569
|
|
View Code Duplication |
@pytest.mark.timeout(COMPACT_TIMEOUT) |
|
|
|
|
|
|
570
|
|
|
@pytest.mark.skip(reason="delete not support yet") |
|
571
|
|
|
def test_insert_delete_part_and_compact_twice(self, connect, jac_collection): |
|
572
|
|
|
''' |
|
573
|
|
|
target: test add entities, delete part of them and compact twice |
|
574
|
|
|
method: add entities, delete part and compact collection twice |
|
575
|
|
|
expected: status ok, data size smaller after first compact, no change after second |
|
576
|
|
|
''' |
|
577
|
|
|
ids = connect.insert(jac_collection, binary_entities) |
|
578
|
|
|
assert len(ids) == nb |
|
579
|
|
|
connect.flush([jac_collection]) |
|
580
|
|
|
delete_ids = [ids[0], ids[-1]] |
|
581
|
|
|
status = connect.delete_entity_by_id(jac_collection, delete_ids) |
|
582
|
|
|
assert status.OK() |
|
583
|
|
|
connect.flush([jac_collection]) |
|
584
|
|
|
# get collection info before compact |
|
585
|
|
|
info = connect.get_collection_stats(jac_collection) |
|
586
|
|
|
size_before = info["partitions"][0]["segments"][0]["data_size"] |
|
587
|
|
|
status = connect.compact(jac_collection) |
|
588
|
|
|
assert status.OK() |
|
589
|
|
|
# get collection info after compact |
|
590
|
|
|
info = connect.get_collection_stats(jac_collection) |
|
591
|
|
|
size_after = info["partitions"][0]["segments"][0]["data_size"] |
|
592
|
|
|
assert(size_before >= size_after) |
|
593
|
|
|
status = connect.compact(jac_collection) |
|
594
|
|
|
assert status.OK() |
|
595
|
|
|
# get collection info after compact twice |
|
596
|
|
|
info = connect.get_collection_stats(jac_collection) |
|
597
|
|
|
size_after_twice = info["partitions"][0]["segments"][0]["data_size"] |
|
598
|
|
|
assert(size_after == size_after_twice) |
|
599
|
|
|
|
|
600
|
|
|
@pytest.mark.timeout(COMPACT_TIMEOUT) |
|
601
|
|
|
@pytest.mark.skip(reason="delete not support yet") |
|
602
|
|
|
def test_compact_multi_collections(self, connect): |
|
603
|
|
|
''' |
|
604
|
|
|
target: test compact works or not with multiple collections |
|
605
|
|
|
method: create 10 collections, add entities into them and compact in turn |
|
606
|
|
|
expected: status ok |
|
607
|
|
|
''' |
|
608
|
|
|
nq = 100 |
|
609
|
|
|
num_collections = 10 |
|
610
|
|
|
tmp, entities = gen_binary_entities(nq) |
|
|
|
|
|
|
611
|
|
|
collection_list = [] |
|
612
|
|
|
for i in range(num_collections): |
|
613
|
|
|
collection_name = gen_unique_str("test_compact_multi_collection_%d" % i) |
|
|
|
|
|
|
614
|
|
|
collection_list.append(collection_name) |
|
615
|
|
|
fields = update_fields_metric_type(default_fields, "JACCARD") |
|
|
|
|
|
|
616
|
|
|
connect.create_collection(collection_name, fields) |
|
617
|
|
|
for i in range(num_collections): |
|
618
|
|
|
ids = connect.insert(collection_list[i], entities) |
|
619
|
|
|
assert len(ids) == nq |
|
620
|
|
|
status = connect.delete_entity_by_id(collection_list[i], [ids[0], ids[-1]]) |
|
621
|
|
|
assert status.OK() |
|
622
|
|
|
connect.flush([collection_list[i]]) |
|
623
|
|
|
status = connect.compact(collection_list[i]) |
|
624
|
|
|
assert status.OK() |
|
625
|
|
|
status = connect.drop_collection(collection_list[i]) |
|
626
|
|
|
assert status.OK() |
|
627
|
|
|
|
|
628
|
|
View Code Duplication |
@pytest.mark.level(2) |
|
|
|
|
|
|
629
|
|
|
@pytest.mark.timeout(COMPACT_TIMEOUT) |
|
630
|
|
|
def test_add_entity_after_compact(self, connect, jac_collection): |
|
631
|
|
|
''' |
|
632
|
|
|
target: test add entity after compact |
|
633
|
|
|
method: after compact operation, add entity |
|
634
|
|
|
expected: status ok, entity added |
|
635
|
|
|
''' |
|
636
|
|
|
ids = connect.insert(jac_collection, binary_entities) |
|
637
|
|
|
connect.flush([jac_collection]) |
|
638
|
|
|
# get collection info before compact |
|
639
|
|
|
info = connect.get_collection_stats(jac_collection) |
|
640
|
|
|
size_before = info["partitions"][0]["segments"][0]["data_size"] |
|
641
|
|
|
status = connect.compact(jac_collection) |
|
642
|
|
|
assert status.OK() |
|
643
|
|
|
# get collection info after compact |
|
644
|
|
|
info = connect.get_collection_stats(jac_collection) |
|
645
|
|
|
size_after = info["partitions"][0]["segments"][0]["data_size"] |
|
646
|
|
|
assert(size_before == size_after) |
|
647
|
|
|
ids = connect.insert(jac_collection, binary_entity) |
|
648
|
|
|
connect.flush([jac_collection]) |
|
649
|
|
|
res = connect.count_entities(jac_collection) |
|
650
|
|
|
assert res == nb + 1 |
|
651
|
|
|
|
|
652
|
|
|
@pytest.mark.timeout(COMPACT_TIMEOUT) |
|
653
|
|
|
@pytest.mark.skip(reason="delete not support yet") |
|
654
|
|
|
def test_delete_entities_after_compact(self, connect, jac_collection): |
|
655
|
|
|
''' |
|
656
|
|
|
target: test delete entities after compact |
|
657
|
|
|
method: after compact operation, delete entities |
|
658
|
|
|
expected: status ok, entities deleted |
|
659
|
|
|
''' |
|
660
|
|
|
ids = connect.insert(jac_collection, binary_entities) |
|
661
|
|
|
connect.flush([jac_collection]) |
|
662
|
|
|
status = connect.compact(jac_collection) |
|
663
|
|
|
assert status.OK() |
|
664
|
|
|
connect.flush([jac_collection]) |
|
665
|
|
|
status = connect.delete_entity_by_id(jac_collection, ids) |
|
666
|
|
|
assert status.OK() |
|
667
|
|
|
connect.flush([jac_collection]) |
|
668
|
|
|
res = connect.count_entities(jac_collection) |
|
669
|
|
|
assert res == 0 |
|
670
|
|
|
|
|
671
|
|
|
@pytest.mark.skip(reason="search not support yet") |
|
672
|
|
|
@pytest.mark.timeout(COMPACT_TIMEOUT) |
|
673
|
|
|
def test_search_after_compact(self, connect, jac_collection): |
|
674
|
|
|
''' |
|
675
|
|
|
target: test search after compact |
|
676
|
|
|
method: after compact operation, search vector |
|
677
|
|
|
expected: status ok |
|
678
|
|
|
''' |
|
679
|
|
|
ids = connect.insert(jac_collection, binary_entities) |
|
680
|
|
|
assert len(ids) == nb |
|
681
|
|
|
connect.flush([jac_collection]) |
|
682
|
|
|
status = connect.compact(jac_collection) |
|
683
|
|
|
assert status.OK() |
|
684
|
|
|
query_vecs = [raw_vectors[0]] |
|
685
|
|
|
distance = jaccard(query_vecs[0], raw_vectors[0]) |
|
|
|
|
|
|
686
|
|
|
query = copy.deepcopy(default_single_query) |
|
|
|
|
|
|
687
|
|
|
query["bool"]["must"][0]["vector"][field_name]["query"] = [binary_entities[-1]["values"][0], |
|
688
|
|
|
binary_entities[-1]["values"][-1]] |
|
689
|
|
|
res = connect.search(jac_collection, query) |
|
690
|
|
|
assert abs(res[0]._distances[0]-distance) <= epsilon |
|
|
|
|
|
|
691
|
|
|
|
|
692
|
|
|
# TODO: |
|
693
|
|
|
@pytest.mark.skip(reason="search not support yet") |
|
694
|
|
|
@pytest.mark.timeout(COMPACT_TIMEOUT) |
|
695
|
|
|
def test_search_after_compact_ip(self, connect, collection): |
|
696
|
|
|
''' |
|
697
|
|
|
target: test search after compact |
|
698
|
|
|
method: after compact operation, search vector |
|
699
|
|
|
expected: status ok |
|
700
|
|
|
''' |
|
701
|
|
|
ids = connect.insert(collection, entities) |
|
702
|
|
|
assert len(ids) == nb |
|
703
|
|
|
connect.flush([collection]) |
|
704
|
|
|
status = connect.compact(collection) |
|
705
|
|
|
query = ip_query() |
|
706
|
|
|
query["bool"]["must"][0]["vector"][field_name]["query"] = [entity[-1]["values"][0], entities[-1]["values"][0], |
|
707
|
|
|
entities[-1]["values"][-1]] |
|
708
|
|
|
res = connect.search(collection, query) |
|
709
|
|
|
logging.getLogger().info(res) |
|
710
|
|
|
assert len(res) == len(query["bool"]["must"][0]["vector"][field_name]["query"]) |
|
|
|
|
|
|
711
|
|
|
assert res[0]._distances[0] < 1 - epsilon |
|
|
|
|
|
|
712
|
|
|
assert res[1]._distances[0] > 1 - epsilon |
|
713
|
|
|
assert res[2]._distances[0] > 1 - epsilon |
|
714
|
|
|
|
milvus-io /
milvus
Loading history...
