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
|
//
//
// Tests cleanupOrphaned concurrent with moveChunk with a hashed shard key.
// Inserts orphan documents to the donor and recipient shards during the moveChunk and
// verifies that cleanupOrphaned removes orphans.
//
load('./jstests/libs/chunk_manipulation_util.js');
load('./jstests/libs/cleanup_orphaned_util.js');
(function() {
"use strict";
var staticMongod = MongoRunner.runMongod({}); // For startParallelOps.
var st = new ShardingTest({shards: 2, other: {separateConfig: true}});
var mongos = st.s0, admin = mongos.getDB('admin'), dbName = 'foo', ns = dbName + '.bar',
coll = mongos.getCollection(ns);
assert.commandWorked(admin.runCommand({enableSharding: dbName}));
printjson(admin.runCommand({movePrimary: dbName, to: st.shard0.shardName}));
assert.commandWorked(admin.runCommand({shardCollection: ns, key: {key: 'hashed'}}));
// Makes four chunks by default, two on each shard.
var chunks = st.config.chunks.find().sort({min: 1}).toArray();
assert.eq(4, chunks.length);
var chunkWithDoc = chunks[1];
print('Trying to make doc that hashes to this chunk: ' + tojson(chunkWithDoc));
var found = false;
for (var i = 0; i < 10000; i++) {
var doc = {key: ObjectId()}, hash = mongos.adminCommand({_hashBSONElement: doc.key}).out;
print('doc.key ' + doc.key + ' hashes to ' + hash);
if (mongos.getCollection('config.chunks')
.findOne(
{_id: chunkWithDoc._id, 'min.key': {$lte: hash}, 'max.key': {$gt: hash}})) {
found = true;
break;
}
}
assert(found, "Couldn't make doc that belongs to chunk 1.");
print('Doc: ' + tojson(doc));
coll.insert(doc);
assert.eq(null, coll.getDB().getLastError());
//
// Start a moveChunk in the background from shard 0 to shard 1. Pause it at
// some points in the donor's and recipient's work flows, and test
// cleanupOrphaned.
//
var donor, recip;
if (chunkWithDoc.shard == st.shard0.shardName) {
donor = st.shard0;
recip = st.shard1;
} else {
recip = st.shard0;
donor = st.shard1;
}
jsTest.log('setting failpoint startedMoveChunk (donor) and cloned (recipient)');
pauseMoveChunkAtStep(donor, moveChunkStepNames.startedMoveChunk);
pauseMigrateAtStep(recip, migrateStepNames.cloned);
var joinMoveChunk = moveChunkParallel(staticMongod,
st.s0.host,
null,
[chunkWithDoc.min, chunkWithDoc.max], // bounds
coll.getFullName(),
recip.shardName);
waitForMoveChunkStep(donor, moveChunkStepNames.startedMoveChunk);
waitForMigrateStep(recip, migrateStepNames.cloned);
proceedToMigrateStep(recip, migrateStepNames.catchup);
// recipient has run _recvChunkStart and begun its migration thread;
// 'doc' has been cloned and chunkWithDoc is noted as 'pending' on recipient.
var donorColl = donor.getCollection(ns), recipColl = recip.getCollection(ns);
assert.eq(1, donorColl.count());
assert.eq(1, recipColl.count());
// cleanupOrphaned should go through two iterations, since the default chunk
// setup leaves two unowned ranges on each shard.
cleanupOrphaned(donor, ns, 2);
cleanupOrphaned(recip, ns, 2);
assert.eq(1, donorColl.count());
assert.eq(1, recipColl.count());
// recip has been waiting for donor to call _recvChunkCommit.
pauseMoveChunkAtStep(donor, moveChunkStepNames.committed);
unpauseMoveChunkAtStep(donor, moveChunkStepNames.startedMoveChunk);
proceedToMigrateStep(recip, migrateStepNames.steady);
proceedToMigrateStep(recip, migrateStepNames.done);
cleanupOrphaned(donor, ns, 2);
assert.eq(1, donorColl.count());
cleanupOrphaned(recip, ns, 2);
assert.eq(1, recipColl.count());
// Let recip side of the migration finish so that the donor proceeds with the commit.
unpauseMigrateAtStep(recip, migrateStepNames.done);
waitForMoveChunkStep(donor, moveChunkStepNames.committed);
// Donor is paused after the migration chunk commit, but before it finishes the cleanup that
// includes running the range deleter. Thus it technically has orphaned data -- commit is
// complete, but moved data is still present. cleanupOrphaned can remove the data the donor
// would otherwise clean up itself in its post-move delete phase.
cleanupOrphaned(donor, ns, 2);
assert.eq(0, donorColl.count());
// Let migration thread complete.
unpauseMoveChunkAtStep(donor, moveChunkStepNames.committed);
joinMoveChunk();
// donor has finished post-move delete, which had nothing to remove with the range deleter
// because of the preemptive cleanupOrphaned call.
assert.eq(0, donorColl.count());
assert.eq(1, recipColl.count());
assert.eq(1, coll.count());
st.stop();
})();
|
