summaryrefslogtreecommitdiff
path: root/src/mongo/s/catalog/replset/sharding_catalog_manager_impl.cpp
blob: 2b982106dc5565533b56365e8dfd22c1a0c2afe6 (plain)
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
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
/**
 *    Copyright (C) 2015 MongoDB Inc.
 *
 *    This program is free software: you can redistribute it and/or  modify
 *    it under the terms of the GNU Affero General Public License, version 3,
 *    as published by the Free Software Foundation.
 *
 *    This program is distributed in the hope that it will be useful,
 *    but WITHOUT ANY WARRANTY; without even the implied warranty of
 *    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 *    GNU Affero General Public License for more details.
 *
 *    You should have received a copy of the GNU Affero General Public License
 *    along with this program.  If not, see <http://www.gnu.org/licenses/>.
 *
 *    As a special exception, the copyright holders give permission to link the
 *    code of portions of this program with the OpenSSL library under certain
 *    conditions as described in each individual source file and distribute
 *    linked combinations including the program with the OpenSSL library. You
 *    must comply with the GNU Affero General Public License in all respects for
 *    all of the code used other than as permitted herein. If you modify file(s)
 *    with this exception, you may extend this exception to your version of the
 *    file(s), but you are not obligated to do so. If you do not wish to do so,
 *    delete this exception statement from your version. If you delete this
 *    exception statement from all source files in the program, then also delete
 *    it in the license file.
 */

#define MONGO_LOG_DEFAULT_COMPONENT ::mongo::logger::LogComponent::kSharding

#include "mongo/platform/basic.h"

#include "mongo/s/catalog/replset/sharding_catalog_manager_impl.h"

#include <iomanip>

#include "mongo/base/status.h"
#include "mongo/base/status_with.h"
#include "mongo/bson/bsonobjbuilder.h"
#include "mongo/bson/util/bson_extract.h"
#include "mongo/client/connection_string.h"
#include "mongo/client/read_preference.h"
#include "mongo/client/remote_command_targeter.h"
#include "mongo/client/replica_set_monitor.h"
#include "mongo/db/client.h"
#include "mongo/db/commands.h"
#include "mongo/db/commands/feature_compatibility_version.h"
#include "mongo/db/db_raii.h"
#include "mongo/db/dbdirectclient.h"
#include "mongo/db/namespace_string.h"
#include "mongo/db/operation_context.h"
#include "mongo/db/s/sharding_state.h"
#include "mongo/db/s/type_shard_identity.h"
#include "mongo/db/wire_version.h"
#include "mongo/executor/network_interface.h"
#include "mongo/executor/task_executor.h"
#include "mongo/rpc/get_status_from_command_result.h"
#include "mongo/s/balancer/balancer_policy.h"
#include "mongo/s/balancer/type_migration.h"
#include "mongo/s/catalog/config_server_version.h"
#include "mongo/s/catalog/sharding_catalog_client.h"
#include "mongo/s/catalog/type_collection.h"
#include "mongo/s/catalog/type_config_version.h"
#include "mongo/s/catalog/type_database.h"
#include "mongo/s/catalog/type_lockpings.h"
#include "mongo/s/catalog/type_locks.h"
#include "mongo/s/catalog/type_shard.h"
#include "mongo/s/catalog/type_tags.h"
#include "mongo/s/client/shard.h"
#include "mongo/s/client/shard_registry.h"
#include "mongo/s/cluster_identity_loader.h"
#include "mongo/s/grid.h"
#include "mongo/s/set_shard_version_request.h"
#include "mongo/s/shard_key_pattern.h"
#include "mongo/s/write_ops/batched_command_request.h"
#include "mongo/s/write_ops/batched_command_response.h"
#include "mongo/stdx/memory.h"
#include "mongo/util/log.h"
#include "mongo/util/mongoutils/str.h"
#include "mongo/util/net/hostandport.h"
#include "mongo/util/scopeguard.h"

namespace mongo {

using std::string;
using std::vector;
using str::stream;

namespace {

using CallbackHandle = executor::TaskExecutor::CallbackHandle;
using CallbackArgs = executor::TaskExecutor::CallbackArgs;
using RemoteCommandCallbackArgs = executor::TaskExecutor::RemoteCommandCallbackArgs;
using RemoteCommandCallbackFn = executor::TaskExecutor::RemoteCommandCallbackFn;

const Seconds kDefaultFindHostMaxWaitTime(20);

const ReadPreferenceSetting kConfigReadSelector(ReadPreference::Nearest, TagSet{});
const ReadPreferenceSetting kConfigPrimarySelector(ReadPreference::PrimaryOnly);
const WriteConcernOptions kNoWaitWriteConcern(1, WriteConcernOptions::SyncMode::UNSET, Seconds(0));

/**
 * Append min, max and version information from chunk to the buffer for logChange purposes.
*/
void appendShortVersion(BufBuilder* b, const ChunkType& chunk) {
    BSONObjBuilder bb(*b);
    bb.append(ChunkType::min(), chunk.getMin());
    bb.append(ChunkType::max(), chunk.getMax());
    if (chunk.isVersionSet())
        chunk.getVersion().addToBSON(bb, ChunkType::DEPRECATED_lastmod());
    bb.done();
}

/**
 * Checks if the given key range for the given namespace conflicts with an existing key range.
 * Note: range should have the full shard key.
 * Returns ErrorCodes::RangeOverlapConflict is an overlap is detected.
 */
Status checkForOveralappedZonedKeyRange(OperationContext* txn,
                                        Shard* configServer,
                                        const NamespaceString& ns,
                                        const ChunkRange& range,
                                        const string& zoneName,
                                        const KeyPattern& shardKeyPattern) {
    DistributionStatus chunkDist(ns, ShardToChunksMap{});

    auto tagStatus = configServer->exhaustiveFindOnConfig(txn,
                                                          kConfigPrimarySelector,
                                                          repl::ReadConcernLevel::kLocalReadConcern,
                                                          NamespaceString(TagsType::ConfigNS),
                                                          BSON(TagsType::ns(ns.ns())),
                                                          BSONObj(),
                                                          0);
    if (!tagStatus.isOK()) {
        return tagStatus.getStatus();
    }

    const auto& tagDocList = tagStatus.getValue().docs;
    for (const auto& tagDoc : tagDocList) {
        auto tagParseStatus = TagsType::fromBSON(tagDoc);
        if (!tagParseStatus.isOK()) {
            return tagParseStatus.getStatus();
        }

        // Always extend ranges to full shard key to be compatible with tags created before
        // the zone commands were implemented.
        const auto& parsedTagDoc = tagParseStatus.getValue();
        auto overlapStatus = chunkDist.addRangeToZone(
            ZoneRange(shardKeyPattern.extendRangeBound(parsedTagDoc.getMinKey(), false),
                      shardKeyPattern.extendRangeBound(parsedTagDoc.getMaxKey(), false),
                      parsedTagDoc.getTag()));
        if (!overlapStatus.isOK()) {
            return overlapStatus;
        }
    }

    auto overlapStatus =
        chunkDist.addRangeToZone(ZoneRange(range.getMin(), range.getMax(), zoneName));
    if (!overlapStatus.isOK()) {
        return overlapStatus;
    }

    return Status::OK();
}

/**
 * Returns a new range based on the given range with the full shard key.
 * Returns:
 * - ErrorCodes::NamespaceNotSharded if ns is not sharded.
 * - ErrorCodes::ShardKeyNotFound if range is not compatible (for example, not a prefix of shard
 * key) with the shard key of ns.
 */
StatusWith<ChunkRange> includeFullShardKey(OperationContext* txn,
                                           Shard* configServer,
                                           const NamespaceString& ns,
                                           const ChunkRange& range,
                                           KeyPattern* shardKeyPatternOut) {
    auto findCollStatus =
        configServer->exhaustiveFindOnConfig(txn,
                                             kConfigPrimarySelector,
                                             repl::ReadConcernLevel::kLocalReadConcern,
                                             NamespaceString(CollectionType::ConfigNS),
                                             BSON(CollectionType::fullNs(ns.ns())),
                                             BSONObj(),
                                             1);

    if (!findCollStatus.isOK()) {
        return findCollStatus.getStatus();
    }

    const auto& findCollResult = findCollStatus.getValue().docs;

    if (findCollResult.size() < 1) {
        return {ErrorCodes::NamespaceNotSharded, str::stream() << ns.ns() << " is not sharded"};
    }

    auto parseStatus = CollectionType::fromBSON(findCollResult.front());
    if (!parseStatus.isOK()) {
        return parseStatus.getStatus();
    }

    auto collDoc = parseStatus.getValue();
    if (collDoc.getDropped()) {
        return {ErrorCodes::NamespaceNotSharded, str::stream() << ns.ns() << " is not sharded"};
    }

    const auto& shardKeyPattern = collDoc.getKeyPattern();
    const auto& shardKeyBSON = shardKeyPattern.toBSON();
    *shardKeyPatternOut = shardKeyPattern;

    if (!range.getMin().isFieldNamePrefixOf(shardKeyBSON)) {
        return {ErrorCodes::ShardKeyNotFound,
                str::stream() << "min: " << range.getMin() << " is not a prefix of the shard key "
                              << shardKeyBSON
                              << " of ns: "
                              << ns.ns()};
    }

    if (!range.getMax().isFieldNamePrefixOf(shardKeyBSON)) {
        return {ErrorCodes::ShardKeyNotFound,
                str::stream() << "max: " << range.getMax() << " is not a prefix of the shard key "
                              << shardKeyBSON
                              << " of ns: "
                              << ns.ns()};
    }

    return ChunkRange(shardKeyPattern.extendRangeBound(range.getMin(), false),
                      shardKeyPattern.extendRangeBound(range.getMax(), false));
}

}  // namespace

ShardingCatalogManagerImpl::ShardingCatalogManagerImpl(
    ShardingCatalogClient* catalogClient, std::unique_ptr<executor::TaskExecutor> addShardExecutor)
    : _catalogClient(catalogClient), _executorForAddShard(std::move(addShardExecutor)) {}

ShardingCatalogManagerImpl::~ShardingCatalogManagerImpl() = default;

Status ShardingCatalogManagerImpl::startup() {
    stdx::lock_guard<stdx::mutex> lk(_mutex);
    if (_started) {
        return Status::OK();
    }
    _started = true;
    _executorForAddShard->startup();
    return Status::OK();
}

void ShardingCatalogManagerImpl::shutDown(OperationContext* txn) {
    LOG(1) << "ShardingCatalogManagerImpl::shutDown() called.";
    {
        stdx::lock_guard<stdx::mutex> lk(_mutex);
        _inShutdown = true;
    }

    _executorForAddShard->shutdown();
    _executorForAddShard->join();
}

StatusWith<Shard::CommandResponse> ShardingCatalogManagerImpl::_runCommandForAddShard(
    OperationContext* txn,
    RemoteCommandTargeter* targeter,
    const std::string& dbName,
    const BSONObj& cmdObj) {
    auto host = targeter->findHost(ReadPreferenceSetting{ReadPreference::PrimaryOnly},
                                   RemoteCommandTargeter::selectFindHostMaxWaitTime(txn));
    if (!host.isOK()) {
        return host.getStatus();
    }

    executor::RemoteCommandRequest request(
        host.getValue(), dbName, cmdObj, rpc::makeEmptyMetadata(), nullptr, Seconds(30));
    executor::RemoteCommandResponse swResponse =
        Status(ErrorCodes::InternalError, "Internal error running command");

    auto callStatus = _executorForAddShard->scheduleRemoteCommand(
        request, [&swResponse](const executor::TaskExecutor::RemoteCommandCallbackArgs& args) {
            swResponse = args.response;
        });
    if (!callStatus.isOK()) {
        return callStatus.getStatus();
    }

    // Block until the command is carried out
    _executorForAddShard->wait(callStatus.getValue());

    if (!swResponse.isOK()) {
        if (swResponse.status.compareCode(ErrorCodes::ExceededTimeLimit)) {
            LOG(0) << "Operation for addShard timed out with status " << swResponse.status;
        }
        return swResponse.status;
    }

    BSONObj responseObj = swResponse.data.getOwned();
    BSONObj responseMetadata = swResponse.metadata.getOwned();
    Status commandStatus = getStatusFromCommandResult(responseObj);
    Status writeConcernStatus = getWriteConcernStatusFromCommandResult(responseObj);

    return Shard::CommandResponse(std::move(responseObj),
                                  std::move(responseMetadata),
                                  std::move(commandStatus),
                                  std::move(writeConcernStatus));
}

StatusWith<boost::optional<ShardType>> ShardingCatalogManagerImpl::_checkIfShardExists(
    OperationContext* txn,
    const ConnectionString& proposedShardConnectionString,
    const std::string* proposedShardName,
    long long proposedShardMaxSize) {
    // Check whether any host in the connection is already part of the cluster.
    const auto existingShards =
        _catalogClient->getAllShards(txn, repl::ReadConcernLevel::kLocalReadConcern);
    if (!existingShards.isOK()) {
        return Status(existingShards.getStatus().code(),
                      str::stream() << "Failed to load existing shards during addShard"
                                    << causedBy(existingShards.getStatus().reason()));
    }

    // Now check if this shard already exists - if it already exists *with the same options* then
    // the addShard request can return success early without doing anything more.
    for (const auto& existingShard : existingShards.getValue().value) {
        auto swExistingShardConnStr = ConnectionString::parse(existingShard.getHost());
        if (!swExistingShardConnStr.isOK()) {
            return swExistingShardConnStr.getStatus();
        }
        auto existingShardConnStr = std::move(swExistingShardConnStr.getValue());
        // Function for determining if the options for the shard that is being added match the
        // options of an existing shard that conflicts with it.
        auto shardsAreEquivalent = [&]() {
            if (proposedShardName && *proposedShardName != existingShard.getName()) {
                return false;
            }
            if (proposedShardConnectionString.type() != existingShardConnStr.type()) {
                return false;
            }
            if (proposedShardConnectionString.type() == ConnectionString::SET &&
                proposedShardConnectionString.getSetName() != existingShardConnStr.getSetName()) {
                return false;
            }
            if (proposedShardMaxSize != existingShard.getMaxSizeMB()) {
                return false;
            }
            return true;
        };

        if (existingShardConnStr.type() == ConnectionString::SET &&
            proposedShardConnectionString.type() == ConnectionString::SET &&
            existingShardConnStr.getSetName() == proposedShardConnectionString.getSetName()) {
            // An existing shard has the same replica set name as the shard being added.
            // If the options aren't the same, then this is an error,
            // but if the options match then the addShard operation should be immediately
            // considered a success and terminated.
            if (shardsAreEquivalent()) {
                return {existingShard};
            } else {
                return {ErrorCodes::IllegalOperation,
                        str::stream() << "A shard already exists containing the replica set '"
                                      << existingShardConnStr.getSetName()
                                      << "'"};
            }
        }

        for (const auto& existingHost : existingShardConnStr.getServers()) {
            // Look if any of the hosts in the existing shard are present within the shard trying
            // to be added.
            for (const auto& addingHost : proposedShardConnectionString.getServers()) {
                if (existingHost == addingHost) {
                    // At least one of the hosts in the shard being added already exists in an
                    // existing shard.  If the options aren't the same, then this is an error,
                    // but if the options match then the addShard operation should be immediately
                    // considered a success and terminated.
                    if (shardsAreEquivalent()) {
                        return {existingShard};
                    } else {
                        return {ErrorCodes::IllegalOperation,
                                str::stream() << "'" << addingHost.toString() << "' "
                                              << "is already a member of the existing shard '"
                                              << existingShard.getHost()
                                              << "' ("
                                              << existingShard.getName()
                                              << ")."};
                    }
                }
            }
        }
        if (proposedShardName && *proposedShardName == existingShard.getName()) {
            // If we get here then we're trying to add a shard with the same name as an existing
            // shard, but there was no overlap in the hosts between the existing shard and the
            // proposed connection string for the new shard.
            return Status(ErrorCodes::IllegalOperation,
                          str::stream() << "A shard named " << *proposedShardName
                                        << " already exists");
        }
    }
    return {boost::none};
}

StatusWith<ShardType> ShardingCatalogManagerImpl::_validateHostAsShard(
    OperationContext* txn,
    std::shared_ptr<RemoteCommandTargeter> targeter,
    const std::string* shardProposedName,
    const ConnectionString& connectionString) {

    // Check if the node being added is a mongos or a version of mongod too old to speak the current
    // communication protocol.
    auto swCommandResponse =
        _runCommandForAddShard(txn, targeter.get(), "admin", BSON("isMaster" << 1));
    if (!swCommandResponse.isOK()) {
        if (swCommandResponse.getStatus() == ErrorCodes::RPCProtocolNegotiationFailed) {
            // Mongos to mongos commands are no longer supported in the wire protocol
            // (because mongos does not support OP_COMMAND), similarly for a new mongos
            // and an old mongod. So the call will fail in such cases.
            // TODO: If/When mongos ever supports opCommands, this logic will break because
            // cmdStatus will be OK.
            return {ErrorCodes::RPCProtocolNegotiationFailed,
                    str::stream() << targeter->connectionString().toString()
                                  << " does not recognize the RPC protocol being used. This is"
                                  << " likely because it contains a node that is a mongos or an old"
                                  << " version of mongod."};
        } else {
            return swCommandResponse.getStatus();
        }
    }

    // Check for a command response error
    auto resIsMasterStatus = std::move(swCommandResponse.getValue().commandStatus);
    if (!resIsMasterStatus.isOK()) {
        return {resIsMasterStatus.code(),
                str::stream() << "Error running isMaster against "
                              << targeter->connectionString().toString()
                              << ": "
                              << causedBy(resIsMasterStatus)};
    }

    auto resIsMaster = std::move(swCommandResponse.getValue().response);

    // Check that the node being added is a new enough version.
    // If we're running this code, that means the mongos that the addShard request originated from
    // must be at least version 3.4 (since 3.2 mongoses don't know about the _configsvrAddShard
    // command).  Since it is illegal to have v3.4 mongoses with v3.2 shards, we should reject
    // adding any shards that are not v3.4.  We can determine this by checking that the
    // maxWireVersion reported in isMaster is at least COMMANDS_ACCEPT_WRITE_CONCERN.
    // TODO(SERVER-25623): This approach won't work to prevent v3.6 mongoses from adding v3.4
    // shards, so we'll have to rethink this during the 3.5 development cycle.

    long long maxWireVersion;
    Status status = bsonExtractIntegerField(resIsMaster, "maxWireVersion", &maxWireVersion);
    if (!status.isOK()) {
        return Status(status.code(),
                      str::stream() << "isMaster returned invalid 'maxWireVersion' "
                                    << "field when attempting to add "
                                    << connectionString.toString()
                                    << " as a shard: "
                                    << status.reason());
    }
    if (maxWireVersion < WireVersion::COMMANDS_ACCEPT_WRITE_CONCERN) {
        return Status(ErrorCodes::IncompatibleServerVersion,
                      str::stream() << "Cannot add " << connectionString.toString()
                                    << " as a shard because we detected a mongod with server "
                                       "version older than 3.4.0.  It is invalid to add v3.2 and "
                                       "older shards through a v3.4 mongos.");
    }


    // Check whether there is a master. If there isn't, the replica set may not have been
    // initiated. If the connection is a standalone, it will return true for isMaster.
    bool isMaster;
    status = bsonExtractBooleanField(resIsMaster, "ismaster", &isMaster);
    if (!status.isOK()) {
        return Status(status.code(),
                      str::stream() << "isMaster returned invalid 'ismaster' "
                                    << "field when attempting to add "
                                    << connectionString.toString()
                                    << " as a shard: "
                                    << status.reason());
    }
    if (!isMaster) {
        return {ErrorCodes::NotMaster,
                str::stream()
                    << connectionString.toString()
                    << " does not have a master. If this is a replica set, ensure that it has a"
                    << " healthy primary and that the set has been properly initiated."};
    }

    const string providedSetName = connectionString.getSetName();
    const string foundSetName = resIsMaster["setName"].str();

    // Make sure the specified replica set name (if any) matches the actual shard's replica set
    if (providedSetName.empty() && !foundSetName.empty()) {
        return {ErrorCodes::OperationFailed,
                str::stream() << "host is part of set " << foundSetName << "; "
                              << "use replica set url format "
                              << "<setname>/<server1>,<server2>, ..."};
    }

    if (!providedSetName.empty() && foundSetName.empty()) {
        return {ErrorCodes::OperationFailed,
                str::stream() << "host did not return a set name; "
                              << "is the replica set still initializing? "
                              << resIsMaster};
    }

    // Make sure the set name specified in the connection string matches the one where its hosts
    // belong into
    if (!providedSetName.empty() && (providedSetName != foundSetName)) {
        return {ErrorCodes::OperationFailed,
                str::stream() << "the provided connection string (" << connectionString.toString()
                              << ") does not match the actual set name "
                              << foundSetName};
    }

    // Is it a config server?
    if (resIsMaster.hasField("configsvr")) {
        return {ErrorCodes::OperationFailed,
                str::stream() << "Cannot add " << connectionString.toString()
                              << " as a shard since it is a config server"};
    }

    // If the shard is part of a replica set, make sure all the hosts mentioned in the connection
    // string are part of the set. It is fine if not all members of the set are mentioned in the
    // connection string, though.
    if (!providedSetName.empty()) {
        std::set<string> hostSet;

        BSONObjIterator iter(resIsMaster["hosts"].Obj());
        while (iter.more()) {
            hostSet.insert(iter.next().String());  // host:port
        }

        if (resIsMaster["passives"].isABSONObj()) {
            BSONObjIterator piter(resIsMaster["passives"].Obj());
            while (piter.more()) {
                hostSet.insert(piter.next().String());  // host:port
            }
        }

        if (resIsMaster["arbiters"].isABSONObj()) {
            BSONObjIterator piter(resIsMaster["arbiters"].Obj());
            while (piter.more()) {
                hostSet.insert(piter.next().String());  // host:port
            }
        }

        vector<HostAndPort> hosts = connectionString.getServers();
        for (size_t i = 0; i < hosts.size(); i++) {
            const string host = hosts[i].toString();  // host:port
            if (hostSet.find(host) == hostSet.end()) {
                return {ErrorCodes::OperationFailed,
                        str::stream() << "in seed list " << connectionString.toString() << ", host "
                                      << host
                                      << " does not belong to replica set "
                                      << foundSetName
                                      << "; found "
                                      << resIsMaster.toString()};
            }
        }
    }

    string actualShardName;

    if (shardProposedName) {
        actualShardName = *shardProposedName;
    } else if (!foundSetName.empty()) {
        // Default it to the name of the replica set
        actualShardName = foundSetName;
    }

    // Disallow adding shard replica set with name 'config'
    if (actualShardName == "config") {
        return {ErrorCodes::BadValue, "use of shard replica set with name 'config' is not allowed"};
    }

    // Retrieve the most up to date connection string that we know from the replica set monitor (if
    // this is a replica set shard, otherwise it will be the same value as connectionString).
    ConnectionString actualShardConnStr = targeter->connectionString();

    ShardType shard;
    shard.setName(actualShardName);
    shard.setHost(actualShardConnStr.toString());
    shard.setState(ShardType::ShardState::kShardAware);

    return shard;
}

StatusWith<std::vector<std::string>> ShardingCatalogManagerImpl::_getDBNamesListFromShard(
    OperationContext* txn, std::shared_ptr<RemoteCommandTargeter> targeter) {

    auto swCommandResponse =
        _runCommandForAddShard(txn, targeter.get(), "admin", BSON("listDatabases" << 1));
    if (!swCommandResponse.isOK()) {
        return swCommandResponse.getStatus();
    }

    auto cmdStatus = std::move(swCommandResponse.getValue().commandStatus);
    if (!cmdStatus.isOK()) {
        return cmdStatus;
    }

    auto cmdResult = std::move(swCommandResponse.getValue().response);

    vector<string> dbNames;

    for (const auto& dbEntry : cmdResult["databases"].Obj()) {
        const string& dbName = dbEntry["name"].String();

        if (!(dbName == "local" || dbName == "admin")) {
            dbNames.push_back(dbName);
        }
    }

    return dbNames;
}

StatusWith<std::string> ShardingCatalogManagerImpl::_generateNewShardName(OperationContext* txn) {
    BSONObjBuilder shardNameRegex;
    shardNameRegex.appendRegex(ShardType::name(), "^shard");

    auto findStatus = Grid::get(txn)->shardRegistry()->getConfigShard()->exhaustiveFindOnConfig(
        txn,
        kConfigReadSelector,
        repl::ReadConcernLevel::kMajorityReadConcern,
        NamespaceString(ShardType::ConfigNS),
        shardNameRegex.obj(),
        BSON(ShardType::name() << -1),
        1);
    if (!findStatus.isOK()) {
        return findStatus.getStatus();
    }

    const auto& docs = findStatus.getValue().docs;

    int count = 0;
    if (!docs.empty()) {
        const auto shardStatus = ShardType::fromBSON(docs.front());
        if (!shardStatus.isOK()) {
            return shardStatus.getStatus();
        }

        std::istringstream is(shardStatus.getValue().getName().substr(5));
        is >> count;
        count++;
    }

    // TODO fix so that we can have more than 10000 automatically generated shard names
    if (count < 9999) {
        std::stringstream ss;
        ss << "shard" << std::setfill('0') << std::setw(4) << count;
        return ss.str();
    }

    return Status(ErrorCodes::OperationFailed, "unable to generate new shard name");
}

StatusWith<string> ShardingCatalogManagerImpl::addShard(
    OperationContext* txn,
    const std::string* shardProposedName,
    const ConnectionString& shardConnectionString,
    const long long maxSize) {
    if (shardConnectionString.type() == ConnectionString::INVALID) {
        return {ErrorCodes::BadValue, "Invalid connection string"};
    }

    if (shardProposedName && shardProposedName->empty()) {
        return {ErrorCodes::BadValue, "shard name cannot be empty"};
    }

    // Only one addShard operation can be in progress at a time.
    Lock::ExclusiveLock lk(txn->lockState(), _kShardMembershipLock);

    // TODO: Don't create a detached Shard object, create a detached RemoteCommandTargeter instead.
    const std::shared_ptr<Shard> shard{
        Grid::get(txn)->shardRegistry()->createConnection(shardConnectionString)};
    invariant(shard);
    auto targeter = shard->getTargeter();

    auto stopMonitoringGuard = MakeGuard([&] {
        if (shardConnectionString.type() == ConnectionString::SET) {
            // This is a workaround for the case were we could have some bad shard being
            // requested to be added and we put that bad connection string on the global replica set
            // monitor registry. It needs to be cleaned up so that when a correct replica set is
            // added, it will be recreated.
            ReplicaSetMonitor::remove(shardConnectionString.getSetName());
        }
    });

    // Validate the specified connection string may serve as shard at all
    auto shardStatus =
        _validateHostAsShard(txn, targeter, shardProposedName, shardConnectionString);
    if (!shardStatus.isOK()) {
        return shardStatus.getStatus();
    }
    ShardType& shardType = shardStatus.getValue();


    // Check if this shard has already been added (can happen in the case of a retry after a network
    // error, for example) and thus this addShard request should be considered a no-op.
    auto existingShard =
        _checkIfShardExists(txn, shardConnectionString, shardProposedName, maxSize);
    if (!existingShard.isOK()) {
        return existingShard.getStatus();
    }
    if (existingShard.getValue()) {
        // These hosts already belong to an existing shard, so report success and terminate the
        // addShard request.
        return existingShard.getValue()->getName();
    }


    // Check that none of the existing shard candidate's dbs exist already
    auto dbNamesStatus = _getDBNamesListFromShard(txn, targeter);
    if (!dbNamesStatus.isOK()) {
        return dbNamesStatus.getStatus();
    }

    for (const string& dbName : dbNamesStatus.getValue()) {
        auto dbt = _catalogClient->getDatabase(txn, dbName);
        if (dbt.isOK()) {
            const auto& dbDoc = dbt.getValue().value;
            return Status(ErrorCodes::OperationFailed,
                          str::stream() << "can't add shard "
                                        << "'"
                                        << shardConnectionString.toString()
                                        << "'"
                                        << " because a local database '"
                                        << dbName
                                        << "' exists in another "
                                        << dbDoc.getPrimary());
        } else if (dbt != ErrorCodes::NamespaceNotFound) {
            return dbt.getStatus();
        }
    }

    // If a name for a shard wasn't provided, generate one
    if (shardType.getName().empty()) {
        StatusWith<string> result = _generateNewShardName(txn);
        if (!result.isOK()) {
            return result.getStatus();
        }
        shardType.setName(result.getValue());
    }

    if (maxSize > 0) {
        shardType.setMaxSizeMB(maxSize);
    }

    // If the minimum allowed version for the cluster is 3.4, set the featureCompatibilityVersion to
    // 3.4 on the shard.
    if (serverGlobalParams.featureCompatibilityVersion.load() ==
        ServerGlobalParams::FeatureCompatibilityVersion_34) {
        auto versionResponse =
            _runCommandForAddShard(txn,
                                   targeter.get(),
                                   "admin",
                                   BSON(FeatureCompatibilityVersion::kCommandName
                                        << FeatureCompatibilityVersion::kVersion34));
        if (!versionResponse.isOK()) {
            return versionResponse.getStatus();
        }

        if (!versionResponse.getValue().commandStatus.isOK()) {
            if (versionResponse.getStatus().code() == ErrorCodes::CommandNotFound) {
                return Status(ErrorCodes::OperationFailed,
                              "featureCompatibilityVersion for cluster is 3.4, cannot add a shard "
                              "with version below 3.4");
            }
            return versionResponse.getValue().commandStatus;
        }
    }

    auto commandRequest = createShardIdentityUpsertForAddShard(txn, shardType.getName());

    LOG(2) << "going to insert shardIdentity document into shard: " << shardType;

    auto swCommandResponse = _runCommandForAddShard(txn, targeter.get(), "admin", commandRequest);
    if (!swCommandResponse.isOK()) {
        return swCommandResponse.getStatus();
    }

    auto commandResponse = std::move(swCommandResponse.getValue());

    BatchedCommandResponse batchResponse;
    auto batchResponseStatus =
        Shard::CommandResponse::processBatchWriteResponse(commandResponse, &batchResponse);
    if (!batchResponseStatus.isOK()) {
        return batchResponseStatus;
    }

    log() << "going to insert new entry for shard into config.shards: " << shardType.toString();

    Status result = _catalogClient->insertConfigDocument(
        txn, ShardType::ConfigNS, shardType.toBSON(), ShardingCatalogClient::kMajorityWriteConcern);
    if (!result.isOK()) {
        log() << "error adding shard: " << shardType.toBSON() << " err: " << result.reason();
        return result;
    }

    // Add all databases which were discovered on the new shard
    for (const string& dbName : dbNamesStatus.getValue()) {
        DatabaseType dbt;
        dbt.setName(dbName);
        dbt.setPrimary(shardType.getName());
        dbt.setSharded(false);

        Status status = _catalogClient->updateDatabase(txn, dbName, dbt);
        if (!status.isOK()) {
            log() << "adding shard " << shardConnectionString.toString()
                  << " even though could not add database " << dbName;
        }
    }

    // Record in changelog
    BSONObjBuilder shardDetails;
    shardDetails.append("name", shardType.getName());
    shardDetails.append("host", shardConnectionString.toString());

    _catalogClient->logChange(
        txn, "addShard", "", shardDetails.obj(), ShardingCatalogClient::kMajorityWriteConcern);

    // Ensure the added shard is visible to this process.
    auto shardRegistry = Grid::get(txn)->shardRegistry();
    if (!shardRegistry->getShard(txn, shardType.getName()).isOK()) {
        return {ErrorCodes::OperationFailed,
                "Could not find shard metadata for shard after adding it. This most likely "
                "indicates that the shard was removed immediately after it was added."};
    }
    stopMonitoringGuard.Dismiss();

    return shardType.getName();
}

Status ShardingCatalogManagerImpl::addShardToZone(OperationContext* txn,
                                                  const std::string& shardName,
                                                  const std::string& zoneName) {
    Lock::ExclusiveLock lk(txn->lockState(), _kZoneOpLock);

    auto updateStatus = _catalogClient->updateConfigDocument(
        txn,
        ShardType::ConfigNS,
        BSON(ShardType::name(shardName)),
        BSON("$addToSet" << BSON(ShardType::tags() << zoneName)),
        false,
        kNoWaitWriteConcern);

    if (!updateStatus.isOK()) {
        return updateStatus.getStatus();
    }

    if (!updateStatus.getValue()) {
        return {ErrorCodes::ShardNotFound,
                str::stream() << "shard " << shardName << " does not exist"};
    }

    return Status::OK();
}

Status ShardingCatalogManagerImpl::removeShardFromZone(OperationContext* txn,
                                                       const std::string& shardName,
                                                       const std::string& zoneName) {
    Lock::ExclusiveLock lk(txn->lockState(), _kZoneOpLock);

    auto configShard = Grid::get(txn)->shardRegistry()->getConfigShard();
    const NamespaceString shardNS(ShardType::ConfigNS);

    //
    // Check whether the shard even exist in the first place.
    //

    auto findShardExistsStatus =
        configShard->exhaustiveFindOnConfig(txn,
                                            kConfigPrimarySelector,
                                            repl::ReadConcernLevel::kLocalReadConcern,
                                            shardNS,
                                            BSON(ShardType::name() << shardName),
                                            BSONObj(),
                                            1);

    if (!findShardExistsStatus.isOK()) {
        return findShardExistsStatus.getStatus();
    }

    if (findShardExistsStatus.getValue().docs.size() == 0) {
        return {ErrorCodes::ShardNotFound,
                str::stream() << "shard " << shardName << " does not exist"};
    }

    //
    // Check how many shards belongs to this zone.
    //

    auto findShardStatus =
        configShard->exhaustiveFindOnConfig(txn,
                                            kConfigPrimarySelector,
                                            repl::ReadConcernLevel::kLocalReadConcern,
                                            shardNS,
                                            BSON(ShardType::tags() << zoneName),
                                            BSONObj(),
                                            2);

    if (!findShardStatus.isOK()) {
        return findShardStatus.getStatus();
    }

    const auto shardDocs = findShardStatus.getValue().docs;

    if (shardDocs.size() == 0) {
        // The zone doesn't exists, this could be a retry.
        return Status::OK();
    }

    if (shardDocs.size() == 1) {
        auto shardDocStatus = ShardType::fromBSON(shardDocs.front());
        if (!shardDocStatus.isOK()) {
            return shardDocStatus.getStatus();
        }

        auto shardDoc = shardDocStatus.getValue();
        if (shardDoc.getName() != shardName) {
            // The last shard that belongs to this zone is a different shard.
            // This could be a retry, so return OK.
            return Status::OK();
        }

        auto findChunkRangeStatus =
            configShard->exhaustiveFindOnConfig(txn,
                                                kConfigPrimarySelector,
                                                repl::ReadConcernLevel::kLocalReadConcern,
                                                NamespaceString(TagsType::ConfigNS),
                                                BSON(TagsType::tag() << zoneName),
                                                BSONObj(),
                                                1);

        if (!findChunkRangeStatus.isOK()) {
            return findChunkRangeStatus.getStatus();
        }

        if (findChunkRangeStatus.getValue().docs.size() > 0) {
            return {ErrorCodes::ZoneStillInUse,
                    "cannot remove a shard from zone if a chunk range is associated with it"};
        }
    }

    //
    // Perform update.
    //

    auto updateStatus =
        _catalogClient->updateConfigDocument(txn,
                                             ShardType::ConfigNS,
                                             BSON(ShardType::name(shardName)),
                                             BSON("$pull" << BSON(ShardType::tags() << zoneName)),
                                             false,
                                             kNoWaitWriteConcern);

    if (!updateStatus.isOK()) {
        return updateStatus.getStatus();
    }

    // The update did not match a document, another thread could have removed it.
    if (!updateStatus.getValue()) {
        return {ErrorCodes::ShardNotFound,
                str::stream() << "shard " << shardName << " no longer exist"};
    }

    return Status::OK();
}


Status ShardingCatalogManagerImpl::assignKeyRangeToZone(OperationContext* txn,
                                                        const NamespaceString& ns,
                                                        const ChunkRange& givenRange,
                                                        const string& zoneName) {
    Lock::ExclusiveLock lk(txn->lockState(), _kZoneOpLock);

    auto configServer = Grid::get(txn)->shardRegistry()->getConfigShard();

    KeyPattern shardKeyPattern{BSONObj()};
    auto fullShardKeyStatus =
        includeFullShardKey(txn, configServer.get(), ns, givenRange, &shardKeyPattern);
    if (!fullShardKeyStatus.isOK()) {
        return fullShardKeyStatus.getStatus();
    }

    const auto& fullShardKeyRange = fullShardKeyStatus.getValue();

    auto zoneExistStatus =
        configServer->exhaustiveFindOnConfig(txn,
                                             kConfigPrimarySelector,
                                             repl::ReadConcernLevel::kLocalReadConcern,
                                             NamespaceString(ShardType::ConfigNS),
                                             BSON(ShardType::tags() << zoneName),
                                             BSONObj(),
                                             1);

    if (!zoneExistStatus.isOK()) {
        return zoneExistStatus.getStatus();
    }

    auto zoneExist = zoneExistStatus.getValue().docs.size() > 0;
    if (!zoneExist) {
        return {ErrorCodes::ZoneNotFound,
                str::stream() << "zone " << zoneName << " does not exist"};
    }

    auto overlapStatus = checkForOveralappedZonedKeyRange(
        txn, configServer.get(), ns, fullShardKeyRange, zoneName, shardKeyPattern);
    if (!overlapStatus.isOK()) {
        return overlapStatus;
    }

    BSONObj updateQuery(
        BSON("_id" << BSON(TagsType::ns(ns.ns()) << TagsType::min(fullShardKeyRange.getMin()))));

    BSONObjBuilder updateBuilder;
    updateBuilder.append("_id",
                         BSON(TagsType::ns(ns.ns()) << TagsType::min(fullShardKeyRange.getMin())));
    updateBuilder.append(TagsType::ns(), ns.ns());
    updateBuilder.append(TagsType::min(), fullShardKeyRange.getMin());
    updateBuilder.append(TagsType::max(), fullShardKeyRange.getMax());
    updateBuilder.append(TagsType::tag(), zoneName);

    auto updateStatus = _catalogClient->updateConfigDocument(
        txn, TagsType::ConfigNS, updateQuery, updateBuilder.obj(), true, kNoWaitWriteConcern);

    if (!updateStatus.isOK()) {
        return updateStatus.getStatus();
    }

    return Status::OK();
}

Status ShardingCatalogManagerImpl::removeKeyRangeFromZone(OperationContext* txn,
                                                          const NamespaceString& ns,
                                                          const ChunkRange& range) {
    Lock::ExclusiveLock lk(txn->lockState(), _kZoneOpLock);

    auto configServer = Grid::get(txn)->shardRegistry()->getConfigShard();

    KeyPattern shardKeyPattern{BSONObj()};
    auto fullShardKeyStatus =
        includeFullShardKey(txn, configServer.get(), ns, range, &shardKeyPattern);
    if (!fullShardKeyStatus.isOK()) {
        return fullShardKeyStatus.getStatus();
    }

    BSONObjBuilder removeBuilder;
    removeBuilder.append("_id", BSON(TagsType::ns(ns.ns()) << TagsType::min(range.getMin())));
    removeBuilder.append(TagsType::max(), range.getMax());

    return _catalogClient->removeConfigDocuments(
        txn, TagsType::ConfigNS, removeBuilder.obj(), kNoWaitWriteConcern);
}

Status ShardingCatalogManagerImpl::commitChunkSplit(OperationContext* txn,
                                                    const NamespaceString& ns,
                                                    const OID& requestEpoch,
                                                    const ChunkRange& range,
                                                    const std::vector<BSONObj>& splitPoints,
                                                    const std::string& shardName) {
    // Take _kChunkOpLock in exclusive mode to prevent concurrent chunk splits, merges, and
    // migrations
    // TODO(SERVER-25359): Replace with a collection-specific lock map to allow splits/merges/
    // move chunks on different collections to proceed in parallel
    Lock::ExclusiveLock lk(txn->lockState(), _kChunkOpLock);

    // Acquire GlobalLock in MODE_X twice to prevent yielding.
    // GlobalLock and the following lock on config.chunks are only needed to support
    // mixed-mode operation with mongoses from 3.2
    // TODO(SERVER-25337): Remove GlobalLock and config.chunks lock after 3.4
    Lock::GlobalLock firstGlobalLock(txn->lockState(), MODE_X, UINT_MAX);
    Lock::GlobalLock secondGlobalLock(txn->lockState(), MODE_X, UINT_MAX);

    // Acquire lock on config.chunks in MODE_X
    AutoGetCollection autoColl(txn, NamespaceString(ChunkType::ConfigNS), MODE_X);

    // Get the chunk with highest version for this namespace
    auto findStatus = grid.shardRegistry()->getConfigShard()->exhaustiveFindOnConfig(
        txn,
        ReadPreferenceSetting{ReadPreference::PrimaryOnly},
        repl::ReadConcernLevel::kLocalReadConcern,
        NamespaceString(ChunkType::ConfigNS),
        BSON("ns" << ns.ns()),
        BSON(ChunkType::DEPRECATED_lastmod << -1),
        1);

    if (!findStatus.isOK()) {
        return findStatus.getStatus();
    }

    const auto& chunksVector = findStatus.getValue().docs;
    if (chunksVector.empty())
        return {ErrorCodes::IllegalOperation,
                "collection does not exist, isn't sharded, or has no chunks"};

    ChunkVersion collVersion =
        ChunkVersion::fromBSON(chunksVector.front(), ChunkType::DEPRECATED_lastmod());

    // Return an error if epoch of chunk does not match epoch of request
    if (collVersion.epoch() != requestEpoch) {
        return {ErrorCodes::StaleEpoch,
                "epoch of chunk does not match epoch of request. This most likely means "
                "that the collection was dropped and re-created."};
    }

    std::vector<ChunkType> newChunks;

    auto newChunkBounds(splitPoints);
    ChunkVersion currentMaxVersion = collVersion;
    auto startKey = range.getMin();
    newChunkBounds.push_back(
        range.getMax());  // makes it easier to have 'max' in the next loop. remove later.

    BSONArrayBuilder updates;

    for (const auto& endKey : newChunkBounds) {
        // verify that splitPoints are non-equivalent
        if (endKey.woCompare(startKey) == 0) {
            return {ErrorCodes::InvalidOptions,
                    str::stream() << "split on lower bound "
                                     " of chunk "
                                  << "["
                                  << startKey
                                  << ","
                                  << endKey
                                  << ")"
                                  << "is not allowed"};
        }

        // verify that splits don't create too-big shard keys
        Status shardKeyStatus = ShardKeyPattern::checkShardKeySize(endKey);
        if (!shardKeyStatus.isOK()) {
            return shardKeyStatus;
        }

        // splits only update the 'minor' portion of version
        currentMaxVersion.incMinor();

        // build an update operation against the chunks collection of the config database
        // with upsert true
        BSONObjBuilder op;
        op.append("op", "u");
        op.appendBool("b", true);
        op.append("ns", ChunkType::ConfigNS);

        // add the modified (new) chunk information as the update object
        BSONObjBuilder n(op.subobjStart("o"));
        n.append(ChunkType::name(), ChunkType::genID(ns.ns(), startKey));
        currentMaxVersion.addToBSON(n, ChunkType::DEPRECATED_lastmod());
        n.append(ChunkType::ns(), ns.ns());
        n.append(ChunkType::min(), startKey);
        n.append(ChunkType::max(), endKey);
        n.append(ChunkType::shard(), shardName);
        n.done();

        // add the chunk's _id as the query part of the update statement
        BSONObjBuilder q(op.subobjStart("o2"));
        q.append(ChunkType::name(), ChunkType::genID(ns.ns(), startKey));
        q.done();

        updates.append(op.obj());

        // remember this chunk info for logging later
        ChunkType chunk;
        chunk.setMin(startKey);
        chunk.setMax(endKey);
        chunk.setVersion(currentMaxVersion);

        newChunks.push_back(std::move(chunk));

        startKey = endKey;
    }

    newChunkBounds.pop_back();  // 'max' was used as sentinel

    BSONArrayBuilder preCond;
    {
        BSONObjBuilder b;
        b.append("ns", ChunkType::ConfigNS);
        b.append("q",
                 BSON("query" << BSON(ChunkType::ns(ns.ns()) << ChunkType::min() << range.getMin()
                                                             << ChunkType::max()
                                                             << range.getMax())
                              << "orderby"
                              << BSON(ChunkType::DEPRECATED_lastmod() << -1)));
        {
            BSONObjBuilder bb(b.subobjStart("res"));
            bb.append(ChunkType::DEPRECATED_epoch(), requestEpoch);
            bb.append(ChunkType::shard(), shardName);
        }
        preCond.append(b.obj());
    }

    // apply the batch of updates to remote and local metadata
    Status applyOpsStatus =
        grid.catalogClient(txn)->applyChunkOpsDeprecated(txn,
                                                         updates.arr(),
                                                         preCond.arr(),
                                                         ns.ns(),
                                                         currentMaxVersion,
                                                         WriteConcernOptions(),
                                                         repl::ReadConcernLevel::kLocalReadConcern);
    if (!applyOpsStatus.isOK()) {
        return applyOpsStatus;
    }

    // log changes
    BSONObjBuilder logDetail;
    {
        BSONObjBuilder b(logDetail.subobjStart("before"));
        b.append(ChunkType::min(), range.getMin());
        b.append(ChunkType::max(), range.getMax());
        collVersion.addToBSON(b, ChunkType::DEPRECATED_lastmod());
    }

    if (newChunks.size() == 2) {
        _appendShortVersion(logDetail.subobjStart("left"), newChunks[0]);
        _appendShortVersion(logDetail.subobjStart("right"), newChunks[1]);

        grid.catalogClient(txn)->logChange(
            txn, "split", ns.ns(), logDetail.obj(), WriteConcernOptions());
    } else {
        BSONObj beforeDetailObj = logDetail.obj();
        BSONObj firstDetailObj = beforeDetailObj.getOwned();
        const int newChunksSize = newChunks.size();

        for (int i = 0; i < newChunksSize; i++) {
            BSONObjBuilder chunkDetail;
            chunkDetail.appendElements(beforeDetailObj);
            chunkDetail.append("number", i + 1);
            chunkDetail.append("of", newChunksSize);
            _appendShortVersion(chunkDetail.subobjStart("chunk"), newChunks[i]);

            grid.catalogClient(txn)->logChange(
                txn, "multi-split", ns.ns(), chunkDetail.obj(), WriteConcernOptions());
        }
    }

    return applyOpsStatus;
}

Status ShardingCatalogManagerImpl::commitChunkMerge(OperationContext* txn,
                                                    const NamespaceString& ns,
                                                    const OID& requestEpoch,
                                                    const std::vector<BSONObj>& chunkBoundaries,
                                                    const std::string& shardName) {
    // Take _kChunkOpLock in exclusive mode to prevent concurrent chunk splits, merges, and
    // migrations
    // TODO(SERVER-25359): Replace with a collection-specific lock map to allow splits/merges/
    // move chunks on different collections to proceed in parallel
    Lock::ExclusiveLock lk(txn->lockState(), _kChunkOpLock);

    // Acquire GlobalLock in MODE_X twice to prevent yielding.
    // GLobalLock and the following lock on config.chunks are only needed to support
    // mixed-mode operation with mongoses from 3.2
    // TODO(SERVER-25337): Remove GlobalLock and config.chunks lock after 3.4
    Lock::GlobalLock firstGlobalLock(txn->lockState(), MODE_X, UINT_MAX);
    Lock::GlobalLock secondGlobalLock(txn->lockState(), MODE_X, UINT_MAX);

    // Acquire lock on config.chunks in MODE_X
    AutoGetCollection autoColl(txn, NamespaceString(ChunkType::ConfigNS), MODE_X);

    // Get the chunk with the highest version for this namespace
    auto findStatus = grid.shardRegistry()->getConfigShard()->exhaustiveFindOnConfig(
        txn,
        ReadPreferenceSetting{ReadPreference::PrimaryOnly},
        repl::ReadConcernLevel::kLocalReadConcern,
        NamespaceString(ChunkType::ConfigNS),
        BSON("ns" << ns.ns()),
        BSON(ChunkType::DEPRECATED_lastmod << -1),
        1);

    if (!findStatus.isOK()) {
        return findStatus.getStatus();
    }

    const auto& chunksVector = findStatus.getValue().docs;
    if (chunksVector.empty())
        return {ErrorCodes::IllegalOperation,
                "collection does not exist, isn't sharded, or has no chunks"};

    ChunkVersion collVersion =
        ChunkVersion::fromBSON(chunksVector.front(), ChunkType::DEPRECATED_lastmod());

    // Return an error if epoch of chunk does not match epoch of request
    if (collVersion.epoch() != requestEpoch) {
        return {ErrorCodes::StaleEpoch,
                "epoch of chunk does not match epoch of request. This most likely means "
                "that the collection was dropped and re-created."};
    }

    // Build chunks to be merged
    std::vector<ChunkType> chunksToMerge;

    ChunkType itChunk;
    itChunk.setMax(chunkBoundaries.front());
    itChunk.setNS(ns.ns());
    itChunk.setShard(shardName);

    // Do not use the first chunk boundary as a max bound while building chunks
    for (size_t i = 1; i < chunkBoundaries.size(); ++i) {
        itChunk.setMin(itChunk.getMax());
        itChunk.setMax(chunkBoundaries[i]);
        chunksToMerge.push_back(itChunk);
    }

    ChunkVersion mergeVersion = collVersion;
    mergeVersion.incMinor();

    BSONArrayBuilder updates;

    // Build an update operation to expand the first chunk into the newly merged chunk
    {
        BSONObjBuilder op;
        op.append("op", "u");
        op.appendBool("b", false);
        op.append("ns", ChunkType::ConfigNS);

        // expand first chunk into newly merged chunk
        ChunkType mergedChunk(chunksToMerge.front());
        mergedChunk.setMax(chunksToMerge.back().getMax());

        // fill in additional details for sending through applyOps
        mergedChunk.setVersion(mergeVersion);

        // add the new chunk information as the update object
        op.append("o", mergedChunk.toBSON());

        // query object
        op.append("o2", BSON(ChunkType::name(mergedChunk.getName())));

        updates.append(op.obj());
    }

    // Build update operations to delete the rest of the chunks to be merged. Remember not
    // to delete the first chunk we're expanding
    for (size_t i = 1; i < chunksToMerge.size(); ++i) {
        BSONObjBuilder op;
        op.append("op", "d");
        op.append("ns", ChunkType::ConfigNS);

        op.append("o", BSON(ChunkType::name(chunksToMerge[i].getName())));

        updates.append(op.obj());
    }

    BSONArrayBuilder preCond;
    {
        BSONObjBuilder b;
        b.append("ns", ChunkType::ConfigNS);
        b.append("q",
                 BSON("query" << BSON(ChunkType::ns(ns.ns())) << "orderby"
                              << BSON(ChunkType::DEPRECATED_lastmod() << -1)));
        {
            BSONObjBuilder bb(b.subobjStart("res"));
            collVersion.addToBSON(bb, ChunkType::DEPRECATED_lastmod());
        }
        preCond.append(b.obj());
    }

    // apply the batch of updates to remote and local metadata
    Status applyOpsStatus =
        grid.catalogClient(txn)->applyChunkOpsDeprecated(txn,
                                                         updates.arr(),
                                                         preCond.arr(),
                                                         ns.ns(),
                                                         mergeVersion,
                                                         WriteConcernOptions(),
                                                         repl::ReadConcernLevel::kLocalReadConcern);
    if (!applyOpsStatus.isOK()) {
        return applyOpsStatus;
    }

    // log changes
    BSONObjBuilder logDetail;
    {
        BSONArrayBuilder b(logDetail.subarrayStart("merged"));
        for (auto chunkToMerge : chunksToMerge) {
            b.append(chunkToMerge.toBSON());
        }
    }
    collVersion.addToBSON(logDetail, "prevShardVersion");
    mergeVersion.addToBSON(logDetail, "mergedVersion");

    grid.catalogClient(txn)->logChange(
        txn, "merge", ns.ns(), logDetail.obj(), WriteConcernOptions());

    return applyOpsStatus;
}

void ShardingCatalogManagerImpl::_appendShortVersion(BufBuilder& b, const ChunkType& chunk) {
    BSONObjBuilder bb(b);
    bb.append(ChunkType::min(), chunk.getMin());
    bb.append(ChunkType::max(), chunk.getMax());
    if (chunk.isVersionSet())
        chunk.getVersion().addToBSON(bb, ChunkType::DEPRECATED_lastmod());
    bb.done();
}

void ShardingCatalogManagerImpl::appendConnectionStats(executor::ConnectionPoolStats* stats) {
    _executorForAddShard->appendConnectionStats(stats);
}

Status ShardingCatalogManagerImpl::initializeConfigDatabaseIfNeeded(OperationContext* txn) {
    {
        stdx::lock_guard<stdx::mutex> lk(_mutex);
        if (_configInitialized) {
            return {ErrorCodes::AlreadyInitialized,
                    "Config database was previously loaded into memory"};
        }
    }

    Status status = _initConfigIndexes(txn);
    if (!status.isOK()) {
        return status;
    }

    // Make sure to write config.version last since we detect rollbacks of config.version and
    // will re-run initializeConfigDatabaseIfNeeded if that happens, but we don't detect rollback
    // of the index builds.
    status = _initConfigVersion(txn);
    if (!status.isOK()) {
        return status;
    }

    stdx::lock_guard<stdx::mutex> lk(_mutex);
    _configInitialized = true;

    return Status::OK();
}

void ShardingCatalogManagerImpl::discardCachedConfigDatabaseInitializationState() {
    stdx::lock_guard<stdx::mutex> lk(_mutex);
    _configInitialized = false;
}

Status ShardingCatalogManagerImpl::_initConfigVersion(OperationContext* txn) {
    auto versionStatus =
        _catalogClient->getConfigVersion(txn, repl::ReadConcernLevel::kLocalReadConcern);
    if (!versionStatus.isOK()) {
        return versionStatus.getStatus();
    }

    auto versionInfo = versionStatus.getValue();
    if (versionInfo.getMinCompatibleVersion() > CURRENT_CONFIG_VERSION) {
        return {ErrorCodes::IncompatibleShardingConfigVersion,
                str::stream() << "current version v" << CURRENT_CONFIG_VERSION
                              << " is older than the cluster min compatible v"
                              << versionInfo.getMinCompatibleVersion()};
    }

    if (versionInfo.getCurrentVersion() == UpgradeHistory_EmptyVersion) {
        VersionType newVersion;
        newVersion.setClusterId(OID::gen());
        newVersion.setMinCompatibleVersion(MIN_COMPATIBLE_CONFIG_VERSION);
        newVersion.setCurrentVersion(CURRENT_CONFIG_VERSION);

        BSONObj versionObj(newVersion.toBSON());
        auto insertStatus = _catalogClient->insertConfigDocument(
            txn, VersionType::ConfigNS, versionObj, kNoWaitWriteConcern);

        return insertStatus;
    }

    if (versionInfo.getCurrentVersion() == UpgradeHistory_UnreportedVersion) {
        return {ErrorCodes::IncompatibleShardingConfigVersion,
                "Assuming config data is old since the version document cannot be found in the "
                "config server and it contains databases besides 'local' and 'admin'. "
                "Please upgrade if this is the case. Otherwise, make sure that the config "
                "server is clean."};
    }

    if (versionInfo.getCurrentVersion() < CURRENT_CONFIG_VERSION) {
        return {ErrorCodes::IncompatibleShardingConfigVersion,
                str::stream() << "need to upgrade current cluster version to v"
                              << CURRENT_CONFIG_VERSION
                              << "; currently at v"
                              << versionInfo.getCurrentVersion()};
    }

    return Status::OK();
}

Status ShardingCatalogManagerImpl::_initConfigIndexes(OperationContext* txn) {
    const bool unique = true;
    auto configShard = Grid::get(txn)->shardRegistry()->getConfigShard();

    Status result =
        configShard->createIndexOnConfig(txn,
                                         NamespaceString(ChunkType::ConfigNS),
                                         BSON(ChunkType::ns() << 1 << ChunkType::min() << 1),
                                         unique);
    if (!result.isOK()) {
        return Status(result.code(),
                      str::stream() << "couldn't create ns_1_min_1 index on config db"
                                    << causedBy(result));
    }

    result = configShard->createIndexOnConfig(
        txn,
        NamespaceString(ChunkType::ConfigNS),
        BSON(ChunkType::ns() << 1 << ChunkType::shard() << 1 << ChunkType::min() << 1),
        unique);
    if (!result.isOK()) {
        return Status(result.code(),
                      str::stream() << "couldn't create ns_1_shard_1_min_1 index on config db"
                                    << causedBy(result));
    }

    result = configShard->createIndexOnConfig(
        txn,
        NamespaceString(ChunkType::ConfigNS),
        BSON(ChunkType::ns() << 1 << ChunkType::DEPRECATED_lastmod() << 1),
        unique);
    if (!result.isOK()) {
        return Status(result.code(),
                      str::stream() << "couldn't create ns_1_lastmod_1 index on config db"
                                    << causedBy(result));
    }

    result = configShard->createIndexOnConfig(
        txn,
        NamespaceString(MigrationType::ConfigNS),
        BSON(MigrationType::ns() << 1 << MigrationType::min() << 1),
        unique);
    if (!result.isOK()) {
        return Status(result.code(),
                      str::stream() << "couldn't create ns_1_min_1 index on config.migrations"
                                    << causedBy(result));
    }

    result = configShard->createIndexOnConfig(
        txn, NamespaceString(ShardType::ConfigNS), BSON(ShardType::host() << 1), unique);
    if (!result.isOK()) {
        return Status(result.code(),
                      str::stream() << "couldn't create host_1 index on config db"
                                    << causedBy(result));
    }

    result = configShard->createIndexOnConfig(
        txn, NamespaceString(LocksType::ConfigNS), BSON(LocksType::lockID() << 1), !unique);
    if (!result.isOK()) {
        return Status(result.code(),
                      str::stream() << "couldn't create lock id index on config db"
                                    << causedBy(result));
    }

    result =
        configShard->createIndexOnConfig(txn,
                                         NamespaceString(LocksType::ConfigNS),
                                         BSON(LocksType::state() << 1 << LocksType::process() << 1),
                                         !unique);
    if (!result.isOK()) {
        return Status(result.code(),
                      str::stream() << "couldn't create state and process id index on config db"
                                    << causedBy(result));
    }

    result = configShard->createIndexOnConfig(
        txn, NamespaceString(LockpingsType::ConfigNS), BSON(LockpingsType::ping() << 1), !unique);
    if (!result.isOK()) {
        return Status(result.code(),
                      str::stream() << "couldn't create lockping ping time index on config db"
                                    << causedBy(result));
    }

    result = configShard->createIndexOnConfig(txn,
                                              NamespaceString(TagsType::ConfigNS),
                                              BSON(TagsType::ns() << 1 << TagsType::min() << 1),
                                              unique);
    if (!result.isOK()) {
        return Status(result.code(),
                      str::stream() << "couldn't create ns_1_min_1 index on config db"
                                    << causedBy(result));
    }

    result = configShard->createIndexOnConfig(txn,
                                              NamespaceString(TagsType::ConfigNS),
                                              BSON(TagsType::ns() << 1 << TagsType::tag() << 1),
                                              !unique);
    if (!result.isOK()) {
        return Status(result.code(),
                      str::stream() << "couldn't create ns_1_tag_1 index on config db"
                                    << causedBy(result));
    }

    return Status::OK();
}

Status ShardingCatalogManagerImpl::initializeShardingAwarenessOnUnawareShards(
    OperationContext* txn) {
    auto swShards = _getAllShardingUnawareShards(txn);
    if (!swShards.isOK()) {
        return swShards.getStatus();
    } else {
        auto shards = std::move(swShards.getValue());
        for (const auto& shard : shards) {
            auto status = upsertShardIdentityOnShard(txn, shard);
            if (!status.isOK()) {
                return status;
            }
        }
    }

    // Note: this OK status means only that tasks to initialize sharding awareness on the shards
    // were scheduled against the task executor, not that the tasks actually succeeded.
    return Status::OK();
}

StatusWith<std::vector<ShardType>> ShardingCatalogManagerImpl::_getAllShardingUnawareShards(
    OperationContext* txn) {
    std::vector<ShardType> shards;
    auto findStatus = Grid::get(txn)->shardRegistry()->getConfigShard()->exhaustiveFindOnConfig(
        txn,
        ReadPreferenceSetting{ReadPreference::PrimaryOnly},
        repl::ReadConcernLevel::kLocalReadConcern,
        NamespaceString(ShardType::ConfigNS),
        BSON(
            "state" << BSON("$ne" << static_cast<std::underlying_type<ShardType::ShardState>::type>(
                                ShardType::ShardState::kShardAware))),  // shard is sharding unaware
        BSONObj(),                                                      // no sort
        boost::none);                                                   // no limit
    if (!findStatus.isOK()) {
        return findStatus.getStatus();
    }

    for (const BSONObj& doc : findStatus.getValue().docs) {
        auto shardRes = ShardType::fromBSON(doc);
        if (!shardRes.isOK()) {
            return {ErrorCodes::FailedToParse,
                    stream() << "Failed to parse shard " << causedBy(shardRes.getStatus()) << doc};
        }

        Status validateStatus = shardRes.getValue().validate();
        if (!validateStatus.isOK()) {
            return {validateStatus.code(),
                    stream() << "Failed to validate shard " << causedBy(validateStatus) << doc};
        }

        shards.push_back(shardRes.getValue());
    }

    return shards;
}

Status ShardingCatalogManagerImpl::upsertShardIdentityOnShard(OperationContext* txn,
                                                              ShardType shardType) {

    auto commandRequest = createShardIdentityUpsertForAddShard(txn, shardType.getName());

    auto swConnString = ConnectionString::parse(shardType.getHost());
    if (!swConnString.isOK()) {
        return swConnString.getStatus();
    }

    // TODO: Don't create a detached Shard object, create a detached RemoteCommandTargeter
    // instead.
    const std::shared_ptr<Shard> shard{
        Grid::get(txn)->shardRegistry()->createConnection(swConnString.getValue())};
    invariant(shard);
    auto targeter = shard->getTargeter();

    _scheduleAddShardTask(
        std::move(shardType), std::move(targeter), std::move(commandRequest), false);

    return Status::OK();
}

void ShardingCatalogManagerImpl::cancelAddShardTaskIfNeeded(const ShardId& shardId) {
    stdx::lock_guard<stdx::mutex> lk(_addShardHandlesMutex);
    if (_hasAddShardHandle_inlock(shardId)) {
        auto cbHandle = _getAddShardHandle_inlock(shardId);
        _executorForAddShard->cancel(cbHandle);
        // Untrack the handle here so that if this shard is re-added before the CallbackCanceled
        // status is delivered to the callback, a new addShard task for the shard will be
        // created.
        _untrackAddShardHandle_inlock(shardId);
    }
}

void ShardingCatalogManagerImpl::_scheduleAddShardTaskUnlessCanceled(
    const CallbackArgs& cbArgs,
    const ShardType shardType,
    std::shared_ptr<RemoteCommandTargeter> targeter,
    const BSONObj commandRequest) {
    if (cbArgs.status == ErrorCodes::CallbackCanceled) {
        return;
    }
    _scheduleAddShardTask(
        std::move(shardType), std::move(targeter), std::move(commandRequest), true);
}

void ShardingCatalogManagerImpl::_scheduleAddShardTask(
    const ShardType shardType,
    std::shared_ptr<RemoteCommandTargeter> targeter,
    const BSONObj commandRequest,
    const bool isRetry) {
    stdx::lock_guard<stdx::mutex> lk(_addShardHandlesMutex);

    if (isRetry) {
        // Untrack the handle from scheduleWorkAt, and schedule a new addShard task.
        _untrackAddShardHandle_inlock(shardType.getName());
    } else {
        // We should never be able to schedule an addShard task while one is running, because
        // there is a unique index on the _id field in config.shards.
        invariant(!_hasAddShardHandle_inlock(shardType.getName()));
    }

    // Schedule the shardIdentity upsert request to run immediately, and track the handle.

    auto swHost = targeter->findHost(ReadPreferenceSetting{ReadPreference::PrimaryOnly},
                                     Milliseconds(kDefaultFindHostMaxWaitTime));
    if (!swHost.isOK()) {
        // A 3.2 mongos must have previously successfully communicated with hosts in this shard,
        // so a failure to find a host here is probably transient, and it is safe to retry.
        warning() << "Failed to find host for shard " << shardType
                  << " when trying to upsert a shardIdentity document, "
                  << causedBy(swHost.getStatus());
        const Date_t now = _executorForAddShard->now();
        const Date_t when = now + getAddShardTaskRetryInterval();
        _trackAddShardHandle_inlock(
            shardType.getName(),
            _executorForAddShard->scheduleWorkAt(
                when,
                stdx::bind(&ShardingCatalogManagerImpl::_scheduleAddShardTaskUnlessCanceled,
                           this,
                           stdx::placeholders::_1,
                           shardType,
                           std::move(targeter),
                           std::move(commandRequest))));
        return;
    }

    executor::RemoteCommandRequest request(
        swHost.getValue(), "admin", commandRequest, rpc::makeEmptyMetadata(), nullptr, Seconds(30));

    const RemoteCommandCallbackFn callback =
        stdx::bind(&ShardingCatalogManagerImpl::_handleAddShardTaskResponse,
                   this,
                   stdx::placeholders::_1,
                   shardType,
                   std::move(targeter));

    _trackAddShardHandle_inlock(shardType.getName(),
                                _executorForAddShard->scheduleRemoteCommand(request, callback));
}

void ShardingCatalogManagerImpl::_handleAddShardTaskResponse(
    const RemoteCommandCallbackArgs& cbArgs,
    ShardType shardType,
    std::shared_ptr<RemoteCommandTargeter> targeter) {
    stdx::unique_lock<stdx::mutex> lk(_addShardHandlesMutex);

    Status responseStatus = cbArgs.response.status;
    if (responseStatus == ErrorCodes::CallbackCanceled) {
        return;
    }

    // Untrack the handle from scheduleRemoteCommand regardless of whether the command
    // succeeded. If it failed, we will track the handle for the rescheduled task before
    // releasing the mutex.
    _untrackAddShardHandle_inlock(shardType.getName());

    // Examine the response to determine if the upsert succeeded.

    bool rescheduleTask = false;

    auto swResponse = cbArgs.response;
    if (!swResponse.isOK()) {
        warning() << "Failed to upsert shardIdentity document during addShard into shard "
                  << shardType.getName() << "(" << shardType.getHost()
                  << "). The shardIdentity upsert will continue to be retried. "
                  << causedBy(swResponse.status);
        rescheduleTask = true;
    } else {
        // Create a CommandResponse object in order to use processBatchWriteResponse.
        BSONObj responseObj = swResponse.data.getOwned();
        BSONObj responseMetadata = swResponse.metadata.getOwned();
        Status commandStatus = getStatusFromCommandResult(responseObj);
        Status writeConcernStatus = getWriteConcernStatusFromCommandResult(responseObj);
        Shard::CommandResponse commandResponse(std::move(responseObj),
                                               std::move(responseMetadata),
                                               std::move(commandStatus),
                                               std::move(writeConcernStatus));

        BatchedCommandResponse batchResponse;
        auto batchResponseStatus =
            Shard::CommandResponse::processBatchWriteResponse(commandResponse, &batchResponse);
        if (!batchResponseStatus.isOK()) {
            if (batchResponseStatus == ErrorCodes::DuplicateKey) {
                warning()
                    << "Received duplicate key error when inserting the shardIdentity "
                       "document into "
                    << shardType.getName() << "(" << shardType.getHost()
                    << "). This means the shard has a shardIdentity document with a clusterId "
                       "that differs from this cluster's clusterId. It may still belong to "
                       "or not have been properly removed from another cluster. The "
                       "shardIdentity upsert will continue to be retried.";
            } else {
                warning() << "Failed to upsert shardIdentity document into shard "
                          << shardType.getName() << "(" << shardType.getHost()
                          << ") during addShard. The shardIdentity upsert will continue to be "
                             "retried. "
                          << causedBy(batchResponseStatus);
            }
            rescheduleTask = true;
        }
    }

    if (rescheduleTask) {
        // If the command did not succeed, schedule the upsert shardIdentity task again with a
        // delay.
        const Date_t now = _executorForAddShard->now();
        const Date_t when = now + getAddShardTaskRetryInterval();

        // Track the handle from scheduleWorkAt.
        _trackAddShardHandle_inlock(
            shardType.getName(),
            _executorForAddShard->scheduleWorkAt(
                when,
                stdx::bind(&ShardingCatalogManagerImpl::_scheduleAddShardTaskUnlessCanceled,
                           this,
                           stdx::placeholders::_1,
                           shardType,
                           std::move(targeter),
                           std::move(cbArgs.request.cmdObj))));
        return;
    }

    // If the command succeeded, update config.shards to mark the shard as shardAware.

    // Release the _addShardHandlesMutex before updating config.shards, since it involves disk
    // I/O.
    // At worst, a redundant addShard task will be scheduled by a new primary if the current
    // primary fails during that write.
    lk.unlock();

    // This thread is part of a thread pool owned by the addShard TaskExecutor. Threads in that
    // pool are not created with Client objects associated with them, so a Client is created and
    // attached here to do the local update. The Client is destroyed at the end of the scope,
    // leaving the thread state as it was before.
    Client::initThread(getThreadName().c_str());
    ON_BLOCK_EXIT([&] { Client::destroy(); });

    // Use the thread's Client to create an OperationContext to perform the local write to
    // config.shards. This OperationContext will automatically be destroyed when it goes out of
    // scope at the end of this code block.
    auto txnPtr = cc().makeOperationContext();

    // Use kNoWaitWriteConcern to prevent waiting in this callback, since we don't handle a
    // failed response anyway. If the write is rolled back, the new config primary will attempt to
    // initialize sharding awareness on this shard again, and this update to config.shards will
    // be automatically retried then. If it fails because the shard was removed through the normal
    // removeShard path (so the entry in config.shards was deleted), no new addShard task will
    // get scheduled on the next transition to primary.
    auto updateStatus = _catalogClient->updateConfigDocument(
        txnPtr.get(),
        ShardType::ConfigNS,
        BSON(ShardType::name(shardType.getName())),
        BSON("$set" << BSON(ShardType::state()
                            << static_cast<std::underlying_type<ShardType::ShardState>::type>(
                                ShardType::ShardState::kShardAware))),
        false,
        kNoWaitWriteConcern);

    if (!updateStatus.isOK()) {
        warning() << "Failed to mark shard " << shardType.getName() << "(" << shardType.getHost()
                  << ") as shardAware in config.shards. This will be retried the next time a "
                     "config server transitions to primary. "
                  << causedBy(updateStatus.getStatus());
    }
}

BSONObj ShardingCatalogManagerImpl::createShardIdentityUpsertForAddShard(
    OperationContext* txn, const std::string& shardName) {
    std::unique_ptr<BatchedUpdateDocument> updateDoc(new BatchedUpdateDocument());

    BSONObjBuilder query;
    query.append("_id", "shardIdentity");
    query.append(ShardIdentityType::shardName(), shardName);
    query.append(ShardIdentityType::clusterId(), ClusterIdentityLoader::get(txn)->getClusterId());
    updateDoc->setQuery(query.obj());

    BSONObjBuilder update;
    {
        BSONObjBuilder set(update.subobjStart("$set"));
        set.append(ShardIdentityType::configsvrConnString(),
                   Grid::get(txn)->shardRegistry()->getConfigServerConnectionString().toString());
    }
    updateDoc->setUpdateExpr(update.obj());
    updateDoc->setUpsert(true);

    std::unique_ptr<BatchedUpdateRequest> updateRequest(new BatchedUpdateRequest());
    updateRequest->addToUpdates(updateDoc.release());

    BatchedCommandRequest commandRequest(updateRequest.release());
    commandRequest.setNS(NamespaceString::kConfigCollectionNamespace);
    commandRequest.setWriteConcern(ShardingCatalogClient::kMajorityWriteConcern.toBSON());

    return commandRequest.toBSON();
}


bool ShardingCatalogManagerImpl::_hasAddShardHandle_inlock(const ShardId& shardId) {
    return _addShardHandles.find(shardId) != _addShardHandles.end();
}

const CallbackHandle& ShardingCatalogManagerImpl::_getAddShardHandle_inlock(
    const ShardId& shardId) {
    invariant(_hasAddShardHandle_inlock(shardId));
    return _addShardHandles.find(shardId)->second;
}

void ShardingCatalogManagerImpl::_trackAddShardHandle_inlock(
    const ShardId shardId, const StatusWith<CallbackHandle>& swHandle) {
    if (swHandle.getStatus() == ErrorCodes::ShutdownInProgress) {
        return;
    }
    fassert(40219, swHandle.getStatus());
    _addShardHandles.insert(std::pair<ShardId, CallbackHandle>(shardId, swHandle.getValue()));
}

void ShardingCatalogManagerImpl::_untrackAddShardHandle_inlock(const ShardId& shardId) {
    auto it = _addShardHandles.find(shardId);
    invariant(it != _addShardHandles.end());
    _addShardHandles.erase(shardId);
}

Status ShardingCatalogManagerImpl::setFeatureCompatibilityVersionOnShards(
    OperationContext* txn, const std::string& version) {

    // No shards should be added until we have forwarded featureCompatibilityVersion to all shards.
    Lock::SharedLock lk(txn->lockState(), _kShardMembershipLock);

    std::vector<ShardId> shardIds;
    grid.shardRegistry()->getAllShardIds(&shardIds);
    for (const ShardId& shardId : shardIds) {
        const auto shardStatus = grid.shardRegistry()->getShard(txn, shardId);
        if (!shardStatus.isOK()) {
            continue;
        }
        const auto shard = shardStatus.getValue();

        auto response =
            shard->runCommand(txn,
                              ReadPreferenceSetting{ReadPreference::PrimaryOnly},
                              "admin",
                              BSON(FeatureCompatibilityVersion::kCommandName << version),
                              Shard::RetryPolicy::kIdempotent);
        if (!response.isOK()) {
            return response.getStatus();
        }
        if (!response.getValue().commandStatus.isOK()) {
            return response.getValue().commandStatus;
        }
        if (!response.getValue().writeConcernStatus.isOK()) {
            return response.getValue().writeConcernStatus;
        }
    }

    return Status::OK();
}

}  // namespace mongo