File: test_worker.py

package info (click to toggle)
dask.distributed 2.10.0+ds.1-3
  • links: PTS, VCS
  • area: main
  • in suites: bullseye, sid
  • size: 7,016 kB
  • sloc: python: 48,042; makefile: 231; sh: 83
file content (1607 lines) | stat: -rw-r--r-- 45,009 bytes parent folder | download
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
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
from concurrent.futures import ThreadPoolExecutor
import importlib
import logging
from numbers import Number
from operator import add
import os
import psutil
import shutil
import sys
from time import sleep
import traceback
import asyncio

import dask
from dask import delayed
from dask.utils import format_bytes
from dask.system import CPU_COUNT
import pytest
from toolz import pluck, sliding_window, first
import tornado
from tornado import gen

from distributed import (
    Client,
    Nanny,
    get_client,
    wait,
    default_client,
    get_worker,
    Reschedule,
    wait,
)
from distributed.compatibility import WINDOWS
from distributed.core import rpc
from distributed.scheduler import Scheduler
from distributed.metrics import time
from distributed.worker import Worker, error_message, logger, parse_memory_limit
from distributed.utils import tmpfile, TimeoutError
from distributed.utils_test import (  # noqa: F401
    cleanup,
    inc,
    mul,
    gen_cluster,
    div,
    dec,
    slowinc,
    gen_test,
    captured_logger,
)
from distributed.utils_test import (  # noqa: F401
    client,
    loop,
    nodebug,
    cluster_fixture,
    s,
    a,
    b,
)


def test_worker_nthreads():
    w = Worker("127.0.0.1", 8019)
    try:
        assert w.executor._max_workers == CPU_COUNT
    finally:
        shutil.rmtree(w.local_directory)


@gen_cluster()
def test_str(s, a, b):
    assert a.address in str(a)
    assert a.address in repr(a)
    assert str(a.nthreads) in str(a)
    assert str(a.nthreads) in repr(a)
    assert str(len(a.executing)) in repr(a)


def test_identity():
    w = Worker("127.0.0.1", 8019)
    ident = w.identity(None)
    assert "Worker" in ident["type"]
    assert ident["scheduler"] == "tcp://127.0.0.1:8019"
    assert isinstance(ident["nthreads"], int)
    assert isinstance(ident["memory_limit"], Number)


@gen_cluster(client=True)
def test_worker_bad_args(c, s, a, b):
    class NoReprObj(object):
        """ This object cannot be properly represented as a string. """

        def __str__(self):
            raise ValueError("I have no str representation.")

        def __repr__(self):
            raise ValueError("I have no repr representation.")

    x = c.submit(NoReprObj, workers=a.address)
    yield wait(x)
    assert not a.executing
    assert a.data

    def bad_func(*args, **kwargs):
        1 / 0

    class MockLoggingHandler(logging.Handler):
        """Mock logging handler to check for expected logs."""

        def __init__(self, *args, **kwargs):
            self.reset()
            logging.Handler.__init__(self, *args, **kwargs)

        def emit(self, record):
            self.messages[record.levelname.lower()].append(record.getMessage())

        def reset(self):
            self.messages = {
                "debug": [],
                "info": [],
                "warning": [],
                "error": [],
                "critical": [],
            }

    hdlr = MockLoggingHandler()
    old_level = logger.level
    logger.setLevel(logging.DEBUG)
    logger.addHandler(hdlr)
    y = c.submit(bad_func, x, k=x, workers=b.address)
    yield wait(y)

    assert not b.executing
    assert y.status == "error"
    # Make sure job died because of bad func and not because of bad
    # argument.
    with pytest.raises(ZeroDivisionError):
        yield y

    if sys.version_info[0] >= 3:
        tb = yield y._traceback()
        assert any(
            "1 / 0" in line for line in pluck(3, traceback.extract_tb(tb)) if line
        )
    assert "Compute Failed" in hdlr.messages["warning"][0]
    logger.setLevel(old_level)

    # Now we check that both workers are still alive.

    xx = c.submit(add, 1, 2, workers=a.address)
    yy = c.submit(add, 3, 4, workers=b.address)

    results = yield c._gather([xx, yy])

    assert tuple(results) == (3, 7)


@pytest.mark.slow
@gen_cluster()
def dont_test_delete_data_with_missing_worker(c, a, b):
    bad = "127.0.0.1:9001"  # this worker doesn't exist
    c.who_has["z"].add(bad)
    c.who_has["z"].add(a.address)
    c.has_what[bad].add("z")
    c.has_what[a.address].add("z")
    a.data["z"] = 5

    cc = rpc(ip=c.ip, port=c.port)

    yield cc.delete_data(keys=["z"])  # TODO: this hangs for a while
    assert "z" not in a.data
    assert not c.who_has["z"]
    assert not c.has_what[bad]
    assert not c.has_what[a.address]

    yield cc.close_rpc()


@gen_cluster(client=True)
def test_upload_file(c, s, a, b):
    assert not os.path.exists(os.path.join(a.local_directory, "foobar.py"))
    assert not os.path.exists(os.path.join(b.local_directory, "foobar.py"))
    assert a.local_directory != b.local_directory

    with rpc(a.address) as aa, rpc(b.address) as bb:
        yield [
            aa.upload_file(filename="foobar.py", data=b"x = 123"),
            bb.upload_file(filename="foobar.py", data="x = 123"),
        ]

    assert os.path.exists(os.path.join(a.local_directory, "foobar.py"))
    assert os.path.exists(os.path.join(b.local_directory, "foobar.py"))

    def g():
        import foobar

        return foobar.x

    future = c.submit(g, workers=a.address)
    result = yield future
    assert result == 123

    yield c.close()
    yield s.close(close_workers=True)
    assert not os.path.exists(os.path.join(a.local_directory, "foobar.py"))


@pytest.mark.skip(reason="don't yet support uploading pyc files")
@gen_cluster(client=True, nthreads=[("127.0.0.1", 1)])
def test_upload_file_pyc(c, s, w):
    with tmpfile() as dirname:
        os.mkdir(dirname)
        with open(os.path.join(dirname, "foo.py"), mode="w") as f:
            f.write("def f():\n    return 123")

        sys.path.append(dirname)
        try:
            import foo

            assert foo.f() == 123
            pyc = importlib.util.cache_from_source(os.path.join(dirname, "foo.py"))
            assert os.path.exists(pyc)
            yield c.upload_file(pyc)

            def g():
                import foo

                return foo.x

            future = c.submit(g)
            result = yield future
            assert result == 123
        finally:
            sys.path.remove(dirname)


@gen_cluster(client=True)
def test_upload_egg(c, s, a, b):
    eggname = "testegg-1.0.0-py3.4.egg"
    if 'AUTOPKGTEST_TMP' in os.environ:
        local_file = os.path.abspath(eggname)
    else:
        local_file = __file__.replace('test_worker.py', eggname)
    assert not os.path.exists(os.path.join(a.local_directory, eggname))
    assert not os.path.exists(os.path.join(b.local_directory, eggname))
    assert a.local_directory != b.local_directory

    yield c.upload_file(filename=local_file)

    assert os.path.exists(os.path.join(a.local_directory, eggname))
    assert os.path.exists(os.path.join(b.local_directory, eggname))

    def g(x):
        import testegg

        return testegg.inc(x)

    future = c.submit(g, 10, workers=a.address)
    result = yield future
    assert result == 10 + 1

    yield c.close()
    yield s.close()
    yield a.close()
    yield b.close()
    assert not os.path.exists(os.path.join(a.local_directory, eggname))


@gen_cluster(client=True)
def test_upload_pyz(c, s, a, b):
    pyzname = "mytest.pyz"
    if 'AUTOPKGTEST_TMP' in os.environ:
        local_file = os.path.abspath(pyzname)
    else:
        local_file = __file__.replace('test_worker.py', pyzname)
    local_file = __file__.replace("test_worker.py", pyzname)
    assert not os.path.exists(os.path.join(a.local_directory, pyzname))
    assert not os.path.exists(os.path.join(b.local_directory, pyzname))
    assert a.local_directory != b.local_directory

    yield c.upload_file(filename=local_file)

    assert os.path.exists(os.path.join(a.local_directory, pyzname))
    assert os.path.exists(os.path.join(b.local_directory, pyzname))

    def g(x):
        from mytest import mytest

        return mytest.inc(x)

    future = c.submit(g, 10, workers=a.address)
    result = yield future
    assert result == 10 + 1

    yield c.close()
    yield s.close()
    yield a.close()
    yield b.close()
    assert not os.path.exists(os.path.join(a.local_directory, pyzname))


@pytest.mark.xfail(reason="Still lose time to network I/O")
@gen_cluster(client=True)
def test_upload_large_file(c, s, a, b):
    pytest.importorskip("crick")
    yield gen.sleep(0.05)
    with rpc(a.address) as aa:
        yield aa.upload_file(filename="myfile.dat", data=b"0" * 100000000)
        yield gen.sleep(0.05)
        assert a.digests["tick-duration"].components[0].max() < 0.050


@gen_cluster()
def test_broadcast(s, a, b):
    with rpc(s.address) as cc:
        results = yield cc.broadcast(msg={"op": "ping"})
        assert results == {a.address: b"pong", b.address: b"pong"}


@gen_test()
def test_worker_with_port_zero():
    s = yield Scheduler(port=8007)
    w = yield Worker(s.address)
    assert isinstance(w.port, int)
    assert w.port > 1024

    yield w.close()


@pytest.mark.slow
def test_worker_waits_for_scheduler(loop):
    @gen.coroutine
    def f():
        w = Worker("127.0.0.1", 8007)
        try:
            yield asyncio.wait_for(w, 3)
        except TimeoutError:
            pass
        else:
            assert False
        assert w.status not in ("closed", "running")
        yield w.close(timeout=0.1)

    loop.run_sync(f)


@gen_cluster(client=True, nthreads=[("127.0.0.1", 1)])
def test_worker_task_data(c, s, w):
    x = delayed(2)
    xx = c.persist(x)
    yield wait(xx)
    assert w.data[x.key] == 2


def test_error_message():
    class MyException(Exception):
        def __init__(self, a, b):
            self.args = (a + b,)

        def __str__(self):
            return "MyException(%s)" % self.args

    msg = error_message(MyException("Hello", "World!"))
    assert "Hello" in str(msg["exception"])

    max_error_len = 100
    with dask.config.set({"distributed.admin.max-error-length": max_error_len}):
        msg = error_message(RuntimeError("-" * max_error_len))
        assert len(msg["text"]) <= max_error_len
        assert len(msg["text"]) < max_error_len * 2
        msg = error_message(RuntimeError("-" * max_error_len * 20))
        cut_text = msg["text"].replace("('Long error message', '", "")[:-2]
        assert len(cut_text) == max_error_len

    max_error_len = 1000000
    with dask.config.set({"distributed.admin.max-error-length": max_error_len}):
        msg = error_message(RuntimeError("-" * max_error_len * 2))
        cut_text = msg["text"].replace("('Long error message', '", "")[:-2]
        assert len(cut_text) == max_error_len
        assert len(msg["text"]) > 10100  # default + 100


@gen_cluster()
def test_gather(s, a, b):
    b.data["x"] = 1
    b.data["y"] = 2
    with rpc(a.address) as aa:
        resp = yield aa.gather(who_has={"x": [b.address], "y": [b.address]})
        assert resp["status"] == "OK"

        assert a.data["x"] == b.data["x"]
        assert a.data["y"] == b.data["y"]


@pytest.mark.asyncio
async def test_io_loop(cleanup):
    async with Scheduler(port=0) as s:
        async with Worker(s.address, loop=s.loop) as w:
            assert w.io_loop is s.loop


@gen_cluster(client=True, nthreads=[])
def test_spill_to_disk(c, s):
    np = pytest.importorskip("numpy")
    w = yield Worker(
        s.address,
        loop=s.loop,
        memory_limit=1200 / 0.6,
        memory_pause_fraction=None,
        memory_spill_fraction=None,
    )

    x = c.submit(np.random.randint, 0, 255, size=500, dtype="u1", key="x")
    yield wait(x)
    y = c.submit(np.random.randint, 0, 255, size=500, dtype="u1", key="y")
    yield wait(y)

    assert set(w.data) == {x.key, y.key}
    assert set(w.data.memory) == {x.key, y.key}
    assert set(w.data.fast) == set(w.data.memory)

    z = c.submit(np.random.randint, 0, 255, size=500, dtype="u1", key="z")
    yield wait(z)
    assert set(w.data) == {x.key, y.key, z.key}
    assert set(w.data.memory) == {y.key, z.key}
    assert set(w.data.disk) == {x.key} or set(w.data.slow) == {x.key, y.key}
    assert set(w.data.fast) == set(w.data.memory)
    assert set(w.data.slow) == set(w.data.disk)

    yield x
    assert set(w.data.memory) == {x.key, z.key}
    assert set(w.data.disk) == {y.key} or set(w.data.slow) == {x.key, y.key}
    assert set(w.data.fast) == set(w.data.memory)
    assert set(w.data.slow) == set(w.data.disk)
    yield w.close()


@gen_cluster(client=True)
def test_access_key(c, s, a, b):
    def f(i):
        from distributed.worker import thread_state

        return thread_state.key

    futures = [c.submit(f, i, key="x-%d" % i) for i in range(20)]
    results = yield c._gather(futures)
    assert list(results) == ["x-%d" % i for i in range(20)]


@gen_cluster(client=True)
def test_run_dask_worker(c, s, a, b):
    def f(dask_worker=None):
        return dask_worker.id

    response = yield c._run(f)
    assert response == {a.address: a.id, b.address: b.id}


@gen_cluster(client=True)
def test_run_coroutine_dask_worker(c, s, a, b):
    if sys.version_info < (3,) and tornado.version_info < (4, 5):
        pytest.skip("test needs Tornado 4.5+ on Python 2.7")

    @gen.coroutine
    def f(dask_worker=None):
        yield gen.sleep(0.001)
        raise gen.Return(dask_worker.id)

    response = yield c.run(f)
    assert response == {a.address: a.id, b.address: b.id}


@gen_cluster(client=True, nthreads=[])
def test_Executor(c, s):
    with ThreadPoolExecutor(2) as e:
        w = Worker(s.address, executor=e)
        assert w.executor is e
        w = yield w

        future = c.submit(inc, 1)
        result = yield future
        assert result == 2

        assert e._threads  # had to do some work

        yield w.close()


@pytest.mark.skip(
    reason="Other tests leak memory, so process-level checks trigger immediately"
)
@gen_cluster(
    client=True,
    nthreads=[("127.0.0.1", 1)],
    timeout=30,
    worker_kwargs={"memory_limit": 10e6},
)
def test_spill_by_default(c, s, w):
    da = pytest.importorskip("dask.array")
    x = da.ones(int(10e6 * 0.7), chunks=1e6, dtype="u1")
    y = c.persist(x)
    yield wait(y)
    assert len(w.data.disk)  # something is on disk
    del x, y


@gen_cluster(nthreads=[("127.0.0.1", 1)], worker_kwargs={"reconnect": False})
def test_close_on_disconnect(s, w):
    yield s.close()

    start = time()
    while w.status != "closed":
        yield gen.sleep(0.01)
        assert time() < start + 5


def test_memory_limit_auto():
    a = Worker("127.0.0.1", 8099, nthreads=1)
    b = Worker("127.0.0.1", 8099, nthreads=2)
    c = Worker("127.0.0.1", 8099, nthreads=100)
    d = Worker("127.0.0.1", 8099, nthreads=200)

    assert isinstance(a.memory_limit, Number)
    assert isinstance(b.memory_limit, Number)

    if CPU_COUNT > 1:
        assert a.memory_limit < b.memory_limit

    assert c.memory_limit == d.memory_limit


@gen_cluster(client=True)
def test_inter_worker_communication(c, s, a, b):
    [x, y] = yield c._scatter([1, 2], workers=a.address)

    future = c.submit(add, x, y, workers=b.address)
    result = yield future
    assert result == 3


@gen_cluster(client=True)
def test_clean(c, s, a, b):
    x = c.submit(inc, 1, workers=a.address)
    y = c.submit(inc, x, workers=b.address)

    yield y

    collections = [
        a.tasks,
        a.task_state,
        a.startstops,
        a.data,
        a.nbytes,
        a.durations,
        a.priorities,
        a.types,
        a.threads,
    ]
    for c in collections:
        assert c

    x.release()
    y.release()

    while x.key in a.task_state:
        yield gen.sleep(0.01)

    for c in collections:
        assert not c


@pytest.mark.skipif(sys.version_info[:2] == (3, 4), reason="mul bytes fails")
@gen_cluster(client=True)
def test_message_breakup(c, s, a, b):
    n = 100000
    a.target_message_size = 10 * n
    b.target_message_size = 10 * n
    xs = [c.submit(mul, b"%d" % i, n, workers=a.address) for i in range(30)]
    y = c.submit(lambda *args: None, xs, workers=b.address)
    yield y

    assert 2 <= len(b.incoming_transfer_log) <= 20
    assert 2 <= len(a.outgoing_transfer_log) <= 20

    assert all(msg["who"] == b.address for msg in a.outgoing_transfer_log)
    assert all(msg["who"] == a.address for msg in a.incoming_transfer_log)


@gen_cluster(client=True)
def test_types(c, s, a, b):
    assert not a.types
    assert not b.types
    x = c.submit(inc, 1, workers=a.address)
    yield wait(x)
    assert a.types[x.key] == int

    y = c.submit(inc, x, workers=b.address)
    yield wait(y)
    assert b.types == {x.key: int, y.key: int}

    yield c._cancel(y)

    start = time()
    while y.key in b.data:
        yield gen.sleep(0.01)
        assert time() < start + 5

    assert y.key not in b.types


@gen_cluster()
def test_system_monitor(s, a, b):
    assert b.monitor
    b.monitor.update()


@gen_cluster(
    client=True, nthreads=[("127.0.0.1", 2, {"resources": {"A": 1}}), ("127.0.0.1", 1)]
)
def test_restrictions(c, s, a, b):
    # Resource restrictions
    x = c.submit(inc, 1, resources={"A": 1})
    yield x
    assert a.resource_restrictions == {x.key: {"A": 1}}
    yield c._cancel(x)

    while x.key in a.task_state:
        yield gen.sleep(0.01)

    assert a.resource_restrictions == {}


@pytest.mark.xfail
@gen_cluster(client=True)
def test_clean_nbytes(c, s, a, b):
    L = [delayed(inc)(i) for i in range(10)]
    for i in range(5):
        L = [delayed(add)(x, y) for x, y in sliding_window(2, L)]
    total = delayed(sum)(L)

    future = c.compute(total)
    yield wait(future)

    yield gen.sleep(1)
    assert len(a.nbytes) + len(b.nbytes) == 1


@gen_cluster(client=True, nthreads=[("127.0.0.1", 1)] * 20)
def test_gather_many_small(c, s, a, *workers):
    a.total_out_connections = 2
    futures = yield c._scatter(list(range(100)))

    assert all(w.data for w in workers)

    def f(*args):
        return 10

    future = c.submit(f, *futures, workers=a.address)
    yield wait(future)

    types = list(pluck(0, a.log))
    req = [i for i, t in enumerate(types) if t == "request-dep"]
    recv = [i for i, t in enumerate(types) if t == "receive-dep"]
    assert min(recv) > max(req)

    assert a.comm_nbytes == 0


@gen_cluster(client=True, nthreads=[("127.0.0.1", 1)] * 3)
def test_multiple_transfers(c, s, w1, w2, w3):
    x = c.submit(inc, 1, workers=w1.address)
    y = c.submit(inc, 2, workers=w2.address)
    z = c.submit(add, x, y, workers=w3.address)

    yield wait(z)

    r = w3.startstops[z.key]
    transfers = [t for t in r if t["action"] == "transfer"]
    assert len(transfers) == 2


@gen_cluster(client=True, nthreads=[("127.0.0.1", 1)] * 3)
def test_share_communication(c, s, w1, w2, w3):
    x = c.submit(mul, b"1", int(w3.target_message_size + 1), workers=w1.address)
    y = c.submit(mul, b"2", int(w3.target_message_size + 1), workers=w2.address)
    yield wait([x, y])
    yield c._replicate([x, y], workers=[w1.address, w2.address])
    z = c.submit(add, x, y, workers=w3.address)
    yield wait(z)
    assert len(w3.incoming_transfer_log) == 2
    assert w1.outgoing_transfer_log
    assert w2.outgoing_transfer_log


@gen_cluster(client=True)
def test_dont_overlap_communications_to_same_worker(c, s, a, b):
    x = c.submit(mul, b"1", int(b.target_message_size + 1), workers=a.address)
    y = c.submit(mul, b"2", int(b.target_message_size + 1), workers=a.address)
    yield wait([x, y])
    z = c.submit(add, x, y, workers=b.address)
    yield wait(z)
    assert len(b.incoming_transfer_log) == 2
    l1, l2 = b.incoming_transfer_log

    assert l1["stop"] < l2["start"]


@pytest.mark.avoid_travis
@gen_cluster(client=True)
def test_log_exception_on_failed_task(c, s, a, b):
    with tmpfile() as fn:
        fh = logging.FileHandler(fn)
        try:
            from distributed.worker import logger

            logger.addHandler(fh)

            future = c.submit(div, 1, 0)
            yield wait(future)

            yield gen.sleep(0.1)
            fh.flush()
            with open(fn) as f:
                text = f.read()

            assert "ZeroDivisionError" in text
            assert "Exception" in text
        finally:
            logger.removeHandler(fh)


@gen_cluster(client=True)
def test_clean_up_dependencies(c, s, a, b):
    x = delayed(inc)(1)
    y = delayed(inc)(2)
    xx = delayed(inc)(x)
    yy = delayed(inc)(y)
    z = delayed(add)(xx, yy)

    zz = c.persist(z)
    yield wait(zz)

    start = time()
    while len(a.data) + len(b.data) > 1:
        yield gen.sleep(0.01)
        assert time() < start + 2

    assert set(a.data) | set(b.data) == {zz.key}


@gen_cluster(client=True)
def test_hold_onto_dependents(c, s, a, b):
    x = c.submit(inc, 1, workers=a.address)
    y = c.submit(inc, x, workers=b.address)
    yield wait(y)

    assert x.key in b.data

    yield c._cancel(y)
    yield gen.sleep(0.1)

    assert x.key in b.data


@pytest.mark.slow
@gen_cluster(client=False, nthreads=[])
def test_worker_death_timeout(s):
    with dask.config.set({"distributed.comm.timeouts.connect": "1s"}):
        yield s.close()
        w = Worker(s.address, death_timeout=1)

    with pytest.raises(TimeoutError) as info:
        yield w

    assert "Worker" in str(info.value)
    assert "timed out" in str(info.value) or "failed to start" in str(info.value)

    assert w.status == "closed"


@gen_cluster(client=True)
def test_stop_doing_unnecessary_work(c, s, a, b):
    futures = c.map(slowinc, range(1000), delay=0.01)
    yield gen.sleep(0.1)

    del futures

    start = time()
    while a.executing:
        yield gen.sleep(0.01)
        assert time() - start < 0.5


@gen_cluster(client=True, nthreads=[("127.0.0.1", 1)])
def test_priorities(c, s, w):
    values = []
    for i in range(10):
        a = delayed(slowinc)(i, dask_key_name="a-%d" % i, delay=0.01)
        a1 = delayed(inc)(a, dask_key_name="a1-%d" % i)
        a2 = delayed(inc)(a1, dask_key_name="a2-%d" % i)
        b1 = delayed(dec)(a, dask_key_name="b1-%d" % i)  # <<-- least favored

        values.append(a2)
        values.append(b1)

    futures = c.compute(values)
    yield wait(futures)

    log = [
        t[0]
        for t in w.log
        if t[1] == "executing" and t[2] == "memory" and not t[0].startswith("finalize")
    ]

    assert any(key.startswith("b1") for key in log[: len(log) // 2])


@gen_cluster(client=True)
def test_heartbeats(c, s, a, b):
    x = s.workers[a.address].last_seen
    start = time()
    yield gen.sleep(a.periodic_callbacks["heartbeat"].callback_time / 1000 + 0.1)
    while s.workers[a.address].last_seen == x:
        yield gen.sleep(0.01)
        assert time() < start + 2
    assert a.periodic_callbacks["heartbeat"].callback_time < 1000


@pytest.mark.parametrize("worker", [Worker, Nanny])
def test_worker_dir(worker):
    with tmpfile() as fn:

        @gen_cluster(client=True, worker_kwargs={"local_directory": fn})
        def test_worker_dir(c, s, a, b):
            directories = [w.local_directory for w in s.workers.values()]
            assert all(d.startswith(fn) for d in directories)
            assert len(set(directories)) == 2  # distinct

        test_worker_dir()


@gen_cluster(client=True)
def test_dataframe_attribute_error(c, s, a, b):
    class BadSize(object):
        def __init__(self, data):
            self.data = data

        def __sizeof__(self):
            raise TypeError("Hello")

    future = c.submit(BadSize, 123)
    result = yield future
    assert result.data == 123


@gen_cluster(client=True)
def test_fail_write_to_disk(c, s, a, b):
    class Bad(object):
        def __getstate__(self):
            raise TypeError()

        def __sizeof__(self):
            return int(100e9)

    future = c.submit(Bad)
    yield wait(future)

    assert future.status == "error"

    with pytest.raises(TypeError):
        yield future

    futures = c.map(inc, range(10))
    results = yield c._gather(futures)
    assert results == list(map(inc, range(10)))


@pytest.mark.skip(reason="Our logic here is faulty")
@gen_cluster(
    nthreads=[("127.0.0.1", 2)], client=True, worker_kwargs={"memory_limit": 10e9}
)
def test_fail_write_many_to_disk(c, s, a):
    a.validate = False
    yield gen.sleep(0.1)
    assert not a.paused

    class Bad(object):
        def __init__(self, x):
            pass

        def __getstate__(self):
            raise TypeError()

        def __sizeof__(self):
            return int(2e9)

    futures = c.map(Bad, range(11))
    future = c.submit(lambda *args: 123, *futures)

    yield wait(future)

    with pytest.raises(Exception) as info:
        yield future

    # workers still operational
    result = yield c.submit(inc, 1, workers=a.address)
    assert result == 2


@gen_cluster()
def test_pid(s, a, b):
    assert s.workers[a.address].pid == os.getpid()


@gen_cluster(client=True)
def test_get_client(c, s, a, b):
    def f(x):
        cc = get_client()
        future = cc.submit(inc, x)
        return future.result()

    assert default_client() is c

    future = c.submit(f, 10, workers=a.address)
    result = yield future
    assert result == 11

    assert a._client
    assert not b._client

    assert a._client is c
    assert default_client() is c

    a_client = a._client

    for i in range(10):
        yield wait(c.submit(f, i))

    assert a._client is a_client


def test_get_client_sync(client):
    def f(x):
        cc = get_client()
        future = cc.submit(inc, x)
        return future.result()

    future = client.submit(f, 10)
    assert future.result() == 11


@gen_cluster(client=True)
def test_get_client_coroutine(c, s, a, b):
    @gen.coroutine
    def f():
        client = yield get_client()
        future = client.submit(inc, 10)
        result = yield future
        raise gen.Return(result)

    results = yield c.run(f)
    assert results == {a.address: 11, b.address: 11}


def test_get_client_coroutine_sync(client, s, a, b):
    @gen.coroutine
    def f():
        client = yield get_client()
        future = client.submit(inc, 10)
        result = yield future
        raise gen.Return(result)

    results = client.run(f)
    assert results == {a["address"]: 11, b["address"]: 11}


@gen_cluster()
def test_global_workers(s, a, b):
    n = len(Worker._instances)
    w = first(Worker._instances)
    assert w is a or w is b


@pytest.mark.skipif(WINDOWS, reason="file descriptors")
@gen_cluster(nthreads=[])
def test_worker_fds(s):
    psutil = pytest.importorskip("psutil")
    yield gen.sleep(0.05)
    start = psutil.Process().num_fds()

    worker = yield Worker(s.address, loop=s.loop)
    yield gen.sleep(0.1)
    middle = psutil.Process().num_fds()
    start = time()
    while middle > start:
        yield gen.sleep(0.01)
        assert time() < start + 1

    yield worker.close()

    start = time()
    while psutil.Process().num_fds() > start:
        yield gen.sleep(0.01)
        assert time() < start + 0.5


@gen_cluster(nthreads=[])
async def test_service_hosts_match_worker(s):
    pytest.importorskip("bokeh")
    from distributed.dashboard import BokehWorker

    async with Worker(
        s.address, services={("dashboard", ":0"): BokehWorker}, host="tcp://0.0.0.0"
    ) as w:
        sock = first(w.services["dashboard"].server._http._sockets.values())
        assert sock.getsockname()[0] in ("::", "0.0.0.0")

    async with Worker(
        s.address, services={("dashboard", ":0"): BokehWorker}, host="tcp://127.0.0.1"
    ) as w:
        sock = first(w.services["dashboard"].server._http._sockets.values())
        assert sock.getsockname()[0] in ("::", "0.0.0.0")

    async with Worker(
        s.address, services={("dashboard", 0): BokehWorker}, host="tcp://127.0.0.1"
    ) as w:
        sock = first(w.services["dashboard"].server._http._sockets.values())
        assert sock.getsockname()[0] == "127.0.0.1"


@gen_cluster(nthreads=[])
def test_start_services(s):
    pytest.importorskip("bokeh")
    from distributed.dashboard import BokehWorker

    services = {("dashboard", ":1234"): BokehWorker}

    w = yield Worker(s.address, services=services)

    assert w.services["dashboard"].server.port == 1234
    yield w.close()


@gen_test()
def test_scheduler_file():
    with tmpfile() as fn:
        s = yield Scheduler(scheduler_file=fn, port=8009)
        w = yield Worker(scheduler_file=fn)
        assert set(s.workers) == {w.address}
        yield w.close()
        s.stop()


@gen_cluster(client=True)
def test_scheduler_delay(c, s, a, b):
    old = a.scheduler_delay
    assert abs(a.scheduler_delay) < 0.3
    assert abs(b.scheduler_delay) < 0.3
    yield gen.sleep(a.periodic_callbacks["heartbeat"].callback_time / 1000 + 0.3)
    assert a.scheduler_delay != old


@gen_cluster(client=True)
def test_statistical_profiling(c, s, a, b):
    futures = c.map(slowinc, range(10), delay=0.1)
    yield wait(futures)

    profile = a.profile_keys["slowinc"]
    assert profile["count"]


@pytest.mark.slow
@nodebug
@gen_cluster(
    client=True,
    timeout=30,
    config={
        "distributed.worker.profile.interval": "1ms",
        "distributed.worker.profile.cycle": "100ms",
    },
)
def test_statistical_profiling_2(c, s, a, b):
    da = pytest.importorskip("dask.array")
    while True:
        x = da.random.random(1000000, chunks=(10000,))
        y = (x + x * 2) - x.sum().persist()
        yield wait(y)

        profile = yield a.get_profile()
        text = str(profile)
        if profile["count"] and "sum" in text and "random" in text:
            break


@gen_cluster(
    nthreads=[("127.0.0.1", 1)],
    client=True,
    worker_kwargs={"memory_monitor_interval": 10},
)
def test_robust_to_bad_sizeof_estimates(c, s, a):
    np = pytest.importorskip("numpy")
    memory = psutil.Process().memory_info().rss
    a.memory_limit = memory / 0.7 + 400e6

    class BadAccounting(object):
        def __init__(self, data):
            self.data = data

        def __sizeof__(self):
            return 10

    def f(n):
        x = np.ones(int(n), dtype="u1")
        result = BadAccounting(x)
        return result

    futures = c.map(f, [100e6] * 8, pure=False)

    start = time()
    while not a.data.disk:
        yield gen.sleep(0.1)
        assert time() < start + 5


@pytest.mark.slow
@gen_cluster(
    nthreads=[("127.0.0.1", 2)],
    client=True,
    worker_kwargs={
        "memory_monitor_interval": 10,
        "memory_spill_fraction": False,  # don't spill
        "memory_target_fraction": False,
        "memory_pause_fraction": 0.5,
    },
    timeout=20,
)
def test_pause_executor(c, s, a):
    memory = psutil.Process().memory_info().rss
    a.memory_limit = memory / 0.5 + 200e6
    np = pytest.importorskip("numpy")

    def f():
        x = np.ones(int(400e6), dtype="u1")
        sleep(1)

    with captured_logger(logging.getLogger("distributed.worker")) as logger:
        future = c.submit(f)
        futures = c.map(slowinc, range(30), delay=0.1)

        start = time()
        while not a.paused:
            yield gen.sleep(0.01)
            assert time() < start + 4, (
                format_bytes(psutil.Process().memory_info().rss),
                format_bytes(a.memory_limit),
                len(a.data),
            )
        out = logger.getvalue()
        assert "memory" in out.lower()
        assert "pausing" in out.lower()

    assert sum(f.status == "finished" for f in futures) < 4

    yield wait(futures)


@gen_cluster(client=True, worker_kwargs={"profile_cycle_interval": "50 ms"})
def test_statistical_profiling_cycle(c, s, a, b):
    futures = c.map(slowinc, range(20), delay=0.05)
    yield wait(futures)
    yield gen.sleep(0.01)
    end = time()
    assert len(a.profile_history) > 3

    x = yield a.get_profile(start=time() + 10, stop=time() + 20)
    assert not x["count"]

    x = yield a.get_profile(start=0, stop=time() + 10)
    recent = a.profile_recent["count"]
    actual = sum(p["count"] for _, p in a.profile_history) + a.profile_recent["count"]
    x2 = yield a.get_profile(start=0, stop=time() + 10)
    assert x["count"] <= actual <= x2["count"]

    y = yield a.get_profile(start=end - 0.300, stop=time())
    assert 0 < y["count"] <= x["count"]


@gen_cluster(client=True)
def test_get_current_task(c, s, a, b):
    def some_name():
        return get_worker().get_current_task()

    result = yield c.submit(some_name)
    assert result.startswith("some_name")


@gen_cluster(client=True, nthreads=[("127.0.0.1", 1)] * 2)
def test_reschedule(c, s, a, b):
    s.extensions["stealing"]._pc.stop()
    a_address = a.address

    def f(x):
        sleep(0.1)
        if get_worker().address == a_address:
            raise Reschedule()

    futures = c.map(f, range(4))
    futures2 = c.map(slowinc, range(10), delay=0.1, workers=a.address)
    yield wait(futures)

    assert all(f.key in b.data for f in futures)


def test_deque_handler():
    from distributed.worker import logger

    w = Worker("127.0.0.1", 8019)
    deque_handler = w._deque_handler
    logger.info("foo456")
    assert deque_handler.deque
    msg = deque_handler.deque[-1]
    assert "distributed.worker" in deque_handler.format(msg)
    assert any(msg.msg == "foo456" for msg in deque_handler.deque)


@gen_cluster(nthreads=[], client=True)
def test_avoid_memory_monitor_if_zero_limit(c, s):
    worker = yield Worker(
        s.address, loop=s.loop, memory_limit=0, memory_monitor_interval=10
    )
    assert type(worker.data) is dict
    assert "memory" not in worker.periodic_callbacks

    future = c.submit(inc, 1)
    assert (yield future) == 2
    yield gen.sleep(worker.memory_monitor_interval / 1000)

    yield c.submit(inc, 2)  # worker doesn't pause

    yield worker.close()


@gen_cluster(
    nthreads=[("127.0.0.1", 1)],
    config={
        "distributed.worker.memory.spill": False,
        "distributed.worker.memory.target": False,
    },
)
def test_dict_data_if_no_spill_to_disk(s, w):
    assert type(w.data) is dict


def test_get_worker_name(client):
    def f():
        get_client().submit(inc, 1).result()

    client.run(f)

    def func(dask_scheduler):
        return list(dask_scheduler.clients)

    start = time()
    while not any("worker" in n for n in client.run_on_scheduler(func)):
        sleep(0.1)
        assert time() < start + 10


@gen_cluster(nthreads=[("127.0.0.1", 1)], worker_kwargs={"memory_limit": "2e3 MB"})
def test_parse_memory_limit(s, w):
    assert w.memory_limit == 2e9


@gen_cluster(nthreads=[], client=True)
def test_scheduler_address_config(c, s):
    with dask.config.set({"scheduler-address": s.address}):
        worker = yield Worker(loop=s.loop)
        assert worker.scheduler.address == s.address
    yield worker.close()


@pytest.mark.slow
@gen_cluster(client=True)
def test_wait_for_outgoing(c, s, a, b):
    np = pytest.importorskip("numpy")
    x = np.random.random(10000000)
    future = yield c.scatter(x, workers=a.address)

    y = c.submit(inc, future, workers=b.address)
    yield wait(y)

    assert len(b.incoming_transfer_log) == len(a.outgoing_transfer_log) == 1
    bb = b.incoming_transfer_log[0]["duration"]
    aa = a.outgoing_transfer_log[0]["duration"]
    ratio = aa / bb

    assert 1 / 3 < ratio < 3


@pytest.mark.skipif(
    not sys.platform.startswith("linux"), reason="Need 127.0.0.2 to mean localhost"
)
@gen_cluster(
    nthreads=[("127.0.0.1", 1), ("127.0.0.1", 1), ("127.0.0.2", 1)], client=True
)
def test_prefer_gather_from_local_address(c, s, w1, w2, w3):
    x = yield c.scatter(123, workers=[w1.address, w3.address], broadcast=True)

    y = c.submit(inc, x, workers=[w2.address])
    yield wait(y)

    assert any(d["who"] == w2.address for d in w1.outgoing_transfer_log)
    assert not any(d["who"] == w2.address for d in w3.outgoing_transfer_log)


@gen_cluster(
    client=True,
    nthreads=[("127.0.0.1", 1)] * 20,
    timeout=30,
    config={"distributed.worker.connections.incoming": 1},
)
def test_avoid_oversubscription(c, s, *workers):
    np = pytest.importorskip("numpy")
    x = c.submit(np.random.random, 1000000, workers=[workers[0].address])
    yield wait(x)

    futures = [c.submit(len, x, pure=False, workers=[w.address]) for w in workers[1:]]

    yield wait(futures)

    # Original worker not responsible for all transfers
    assert len(workers[0].outgoing_transfer_log) < len(workers) - 2

    # Some other workers did some work
    assert len([w for w in workers if len(w.outgoing_transfer_log) > 0]) >= 3


@gen_cluster(client=True, worker_kwargs={"metrics": {"my_port": lambda w: w.port}})
def test_custom_metrics(c, s, a, b):
    assert s.workers[a.address].metrics["my_port"] == a.port
    assert s.workers[b.address].metrics["my_port"] == b.port


@gen_cluster(client=True)
def test_register_worker_callbacks(c, s, a, b):
    # preload function to run
    def mystartup(dask_worker):
        dask_worker.init_variable = 1

    def mystartup2():
        import os

        os.environ["MY_ENV_VALUE"] = "WORKER_ENV_VALUE"
        return "Env set."

    # Check that preload function has been run
    def test_import(dask_worker):
        return hasattr(dask_worker, "init_variable")
        #       and dask_worker.init_variable == 1

    def test_startup2():
        import os

        return os.getenv("MY_ENV_VALUE", None) == "WORKER_ENV_VALUE"

    # Nothing has been run yet
    result = yield c.run(test_import)
    assert list(result.values()) == [False] * 2
    result = yield c.run(test_startup2)
    assert list(result.values()) == [False] * 2

    # Start a worker and check that startup is not run
    worker = yield Worker(s.address, loop=s.loop)
    result = yield c.run(test_import, workers=[worker.address])
    assert list(result.values()) == [False]
    yield worker.close()

    # Add a preload function
    response = yield c.register_worker_callbacks(setup=mystartup)
    assert len(response) == 2

    # Check it has been ran on existing worker
    result = yield c.run(test_import)
    assert list(result.values()) == [True] * 2

    # Start a worker and check it is ran on it
    worker = yield Worker(s.address, loop=s.loop)
    result = yield c.run(test_import, workers=[worker.address])
    assert list(result.values()) == [True]
    yield worker.close()

    # Register another preload function
    response = yield c.register_worker_callbacks(setup=mystartup2)
    assert len(response) == 2

    # Check it has been run
    result = yield c.run(test_startup2)
    assert list(result.values()) == [True] * 2

    # Start a worker and check it is ran on it
    worker = yield Worker(s.address, loop=s.loop)
    result = yield c.run(test_import, workers=[worker.address])
    assert list(result.values()) == [True]
    result = yield c.run(test_startup2, workers=[worker.address])
    assert list(result.values()) == [True]
    yield worker.close()


@gen_cluster(client=True)
def test_register_worker_callbacks_err(c, s, a, b):
    with pytest.raises(ZeroDivisionError):
        yield c.register_worker_callbacks(setup=lambda: 1 / 0)


@gen_cluster(nthreads=[])
def test_data_types(s):
    w = yield Worker(s.address, data=dict)
    assert isinstance(w.data, dict)
    yield w.close()

    data = dict()
    w = yield Worker(s.address, data=data)
    assert w.data is data
    yield w.close()

    class Data(dict):
        def __init__(self, x, y):
            self.x = x
            self.y = y

    w = yield Worker(s.address, data=(Data, {"x": 123, "y": 456}))
    assert w.data.x == 123
    assert w.data.y == 456
    yield w.close()


@gen_cluster(nthreads=[])
def test_local_directory(s):
    with tmpfile() as fn:
        with dask.config.set(temporary_directory=fn):
            w = yield Worker(s.address)
            assert w.local_directory.startswith(fn)
            assert "dask-worker-space" in w.local_directory


@pytest.mark.skipif(
    not sys.platform.startswith("linux"), reason="Need 127.0.0.2 to mean localhost"
)
@gen_cluster(nthreads=[], client=True)
def test_host_address(c, s):
    w = yield Worker(s.address, host="127.0.0.2")
    assert "127.0.0.2" in w.address
    yield w.close()

    n = yield Nanny(s.address, host="127.0.0.3")
    assert "127.0.0.3" in n.address
    assert "127.0.0.3" in n.worker_address
    yield n.close()


def test_resource_limit(monkeypatch):
    assert parse_memory_limit("250MiB", 1, total_cores=1) == 1024 * 1024 * 250

    new_limit = 1024 * 1024 * 200
    import distributed.worker

    monkeypatch.setattr(distributed.system, "MEMORY_LIMIT", new_limit)
    assert parse_memory_limit("250MiB", 1, total_cores=1) == new_limit


@pytest.mark.asyncio
@pytest.mark.parametrize("Worker", [Worker, Nanny])
async def test_interface_async(loop, Worker):
    from distributed.utils import get_ip_interface

    psutil = pytest.importorskip("psutil")
    if_names = sorted(psutil.net_if_addrs())
    for if_name in if_names:
        try:
            ipv4_addr = get_ip_interface(if_name)
        except ValueError:
            pass
        else:
            if ipv4_addr == "127.0.0.1":
                break
    else:
        pytest.skip(
            "Could not find loopback interface. "
            "Available interfaces are: %s." % (if_names,)
        )

    async with Scheduler(interface=if_name) as s:
        assert s.address.startswith("tcp://127.0.0.1")
        async with Worker(s.address, interface=if_name) as w:
            assert w.address.startswith("tcp://127.0.0.1")
            assert w.ip == "127.0.0.1"
            async with Client(s.address, asynchronous=True) as c:
                info = c.scheduler_info()
                assert "tcp://127.0.0.1" in info["address"]
                assert all("127.0.0.1" == d["host"] for d in info["workers"].values())


@pytest.mark.asyncio
@pytest.mark.parametrize("Worker", [Worker, Nanny])
async def test_protocol_from_scheduler_address(Worker):
    ucp = pytest.importorskip("ucp")

    async with Scheduler(protocol="ucx") as s:
        assert s.address.startswith("ucx://")
        async with Worker(s.address) as w:
            assert w.address.startswith("ucx://")
            async with Client(s.address, asynchronous=True) as c:
                info = c.scheduler_info()
                assert info["address"].startswith("ucx://")


@pytest.mark.asyncio
@pytest.mark.parametrize("Worker", [Worker, Nanny])
async def test_worker_listens_on_same_interface_by_default(Worker):
    async with Scheduler(host="localhost") as s:
        assert s.ip in {"127.0.0.1", "localhost"}
        async with Worker(s.address) as w:
            assert s.ip == w.ip


@gen_cluster(client=True)
async def test_close_gracefully(c, s, a, b):
    futures = c.map(slowinc, range(200), delay=0.1)
    while not b.data:
        await gen.sleep(0.1)

    mem = set(b.data)
    proc = set(b.executing)

    await b.close_gracefully()

    assert b.status == "closed"
    assert b.address not in s.workers
    assert mem.issubset(set(a.data))
    for key in proc:
        assert s.tasks[key].state in ("processing", "memory")


@pytest.mark.slow
@pytest.mark.asyncio
async def test_lifetime(cleanup):
    async with Scheduler() as s:
        async with Worker(s.address) as a, Worker(s.address, lifetime="1 seconds") as b:
            async with Client(s.address, asynchronous=True) as c:
                futures = c.map(slowinc, range(200), delay=0.1)
                await gen.sleep(1.5)
                assert b.status != "running"
                await b.finished()

                assert set(b.data).issubset(a.data)  # successfully moved data over


@gen_cluster(client=True, worker_kwargs={"lifetime": "10s", "lifetime_stagger": "2s"})
async def test_lifetime_stagger(c, s, a, b):
    assert a.lifetime != b.lifetime
    assert 8 <= a.lifetime <= 12
    assert 8 <= b.lifetime <= 12


@gen_cluster()
async def test_gpu_metrics(s, a, b):
    pytest.importorskip("pynvml")
    from distributed.diagnostics.nvml import count

    assert "gpu" in a.metrics
    assert len(s.workers[a.address].metrics["gpu"]["memory-used"]) == count

    assert "gpu" in a.startup_information
    assert len(s.workers[a.address].extra["gpu"]["name"]) == count


@pytest.mark.asyncio
async def test_bad_metrics(cleanup):
    def bad_metric(w):
        raise Exception("Hello")

    async with Scheduler() as s:
        async with Worker(s.address, metrics={"bad": bad_metric}) as w:
            assert "bad" not in s.workers[w.address].metrics


@pytest.mark.asyncio
async def test_bad_startup(cleanup):
    def bad_startup(w):
        raise Exception("Hello")

    async with Scheduler() as s:
        try:
            w = await Worker(s.address, startup_information={"bad": bad_startup})
        except Exception:
            pytest.fail("Startup exception was raised")


@pytest.mark.asyncio
async def test_update_latency(cleanup):
    async with await Scheduler() as s:
        async with await Worker(s.address) as w:
            original = w.latency
            await w.heartbeat()
            assert original != w.latency

            if w.digests is not None:
                assert w.digests["latency"].size() > 0