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
|
/**
* Tests that a donor shard durably records a migration's state, inserts pending entries into its
* own and the recipient's config.rangeDeletions, and informs itself and the recipient of the
* migration's outcome by updating or deleting its own and the recipient's config.rangeDeletions
* entries for the migration.
*
* This test expects migrations to use the FCV 4.4 protocol.
* @tags: [requires_fcv_44]
*/
(function() {
'use strict';
load("jstests/libs/fail_point_util.js");
load('jstests/libs/parallel_shell_helpers.js');
function getNewNs(dbName) {
if (typeof getNewNs.counter == 'undefined') {
getNewNs.counter = 0;
}
getNewNs.counter++;
const collName = "ns" + getNewNs.counter;
return [collName, dbName + "." + collName];
}
const dbName = "test";
var st = new ShardingTest({shards: 2});
assert.commandWorked(st.s.adminCommand({enableSharding: dbName}));
assert.commandWorked(st.s.adminCommand({movePrimary: dbName, to: st.shard0.shardName}));
function getCollectionUuidAndEpoch(ns) {
const collectionDoc = st.s.getDB("config").getCollection("collections").findOne({_id: ns});
assert.neq(null, collectionDoc);
assert.neq(null, collectionDoc.uuid);
assert.neq(null, collectionDoc.lastmodEpoch);
return [collectionDoc.uuid, collectionDoc.lastmodEpoch];
}
function assertHasMigrationCoordinatorDoc({conn, ns, uuid, epoch}) {
const query = {
nss: ns,
collectionUuid: uuid,
donorShardId: st.shard0.shardName,
recipientShardId: st.shard1.shardName,
"range.min._id": MinKey,
"range.max._id": MaxKey,
"preMigrationChunkVersion.0": Timestamp(1, 0),
"preMigrationChunkVersion.1": epoch
};
assert.neq(
null,
conn.getDB("config").getCollection("migrationCoordinators").findOne(query),
"did not find document matching query " + tojson(query) +
", contents of config.migrationCoordinators on " + conn + ": " +
tojson(conn.getDB("config").getCollection("migrationCoordinators").find().toArray()));
}
function assertEventuallyDoesNotHaveMigrationCoordinatorDoc(conn) {
assert.soon(() => {
return 0 == conn.getDB("config").getCollection("migrationCoordinators").find().itcount();
});
}
function assertHasRangeDeletionDoc({conn, pending, whenToClean, ns, uuid}) {
const query = {
nss: ns,
collectionUuid: uuid,
donorShardId: st.shard0.shardName,
"range.min._id": MinKey,
"range.max._id": MaxKey,
pending: (pending ? true : {$exists: false}),
whenToClean: whenToClean
};
assert.neq(null,
conn.getDB("config").getCollection("rangeDeletions").findOne(query),
"did not find document matching query " + tojson(query) +
", contents of config.rangeDeletions on " + conn + ": " +
tojson(conn.getDB("config").getCollection("rangeDeletions").find().toArray()));
}
function assertEventuallyDoesNotHaveRangeDeletionDoc(conn) {
assert.soon(() => {
return 0 == conn.getDB("config").getCollection("rangeDeletions").find().itcount();
});
}
(() => {
const [collName, ns] = getNewNs(dbName);
jsTest.log("Test end-to-end migration when migration commit succeeds, ns is " + ns);
// Insert some docs into the collection.
const numDocs = 1000;
var bulk = st.s.getDB(dbName).getCollection(collName).initializeUnorderedBulkOp();
for (var i = 0; i < numDocs; i++) {
bulk.insert({_id: i});
}
assert.commandWorked(bulk.execute());
// Shard the collection.
assert.commandWorked(st.s.adminCommand({shardCollection: ns, key: {_id: 1}}));
const [uuid, epoch] = getCollectionUuidAndEpoch(ns);
// Run the moveChunk asynchronously, pausing during cloning to allow the test to make
// assertions.
let step4Failpoint = configureFailPoint(st.shard0, "moveChunkHangAtStep4");
const awaitResult = startParallelShell(
funWithArgs(function(ns, toShardName) {
assert.commandWorked(db.adminCommand({moveChunk: ns, find: {_id: 0}, to: toShardName}));
}, ns, st.shard1.shardName), st.s.port);
// Assert that the durable state for coordinating the migration was written correctly.
step4Failpoint.wait();
assertHasMigrationCoordinatorDoc({conn: st.shard0, ns, uuid, epoch});
assertHasRangeDeletionDoc({conn: st.shard0, pending: true, whenToClean: "delayed", ns, uuid});
assertHasRangeDeletionDoc({conn: st.shard1, pending: true, whenToClean: "now", ns, uuid});
step4Failpoint.off();
// Allow the moveChunk to finish.
awaitResult();
// Donor shard eventually cleans up the orphans.
assert.soon(function() {
return st.shard0.getDB(dbName).getCollection(collName).count() === 0;
});
assert.eq(numDocs, st.s.getDB(dbName).getCollection(collName).find().itcount());
// The durable state for coordinating the migration is eventually cleaned up.
assertEventuallyDoesNotHaveMigrationCoordinatorDoc(st.shard0);
// TODO (SERVER-44159): Delete document from config.rangeDeletions when CollectionRangeDeleter
// finishes deleting a range
// assertEventuallyDoesNotHaveRangeDeletionDoc(st.shard0);
assertEventuallyDoesNotHaveRangeDeletionDoc(st.shard1);
})();
(() => {
const [collName, ns] = getNewNs(dbName);
jsTest.log("Test end-to-end migration when migration commit fails, ns is " + ns);
// Insert some docs into the collection.
const numDocs = 1000;
var bulk = st.s.getDB(dbName).getCollection(collName).initializeUnorderedBulkOp();
for (var i = 0; i < numDocs; i++) {
bulk.insert({_id: i});
}
assert.commandWorked(bulk.execute());
// Shard the collection.
assert.commandWorked(st.s.adminCommand({shardCollection: ns, key: {_id: 1}}));
const [uuid, epoch] = getCollectionUuidAndEpoch(ns);
// Turn on a failpoint to make the migration commit fail on the config server.
let migrationCommitVersionErrorFailpoint =
configureFailPoint(st.configRS.getPrimary(), "migrationCommitVersionError");
// Run the moveChunk asynchronously, pausing during cloning to allow the test to make
// assertions.
let step4Failpoint = configureFailPoint(st.shard0, "moveChunkHangAtStep4");
let step5Failpoint = configureFailPoint(st.shard0, "moveChunkHangAtStep5");
const awaitResult = startParallelShell(
funWithArgs(function(ns, toShardName) {
// Expect StaleEpoch because of the failpoint that will make the migration commit fail.
assert.commandFailedWithCode(
db.adminCommand({moveChunk: ns, find: {_id: 0}, to: toShardName}),
ErrorCodes.StaleEpoch);
}, ns, st.shard1.shardName), st.s.port);
// Assert that the durable state for coordinating the migration was written correctly.
step4Failpoint.wait();
assertHasMigrationCoordinatorDoc({conn: st.shard0, ns, uuid, epoch});
assertHasRangeDeletionDoc({conn: st.shard0, pending: true, whenToClean: "delayed", ns, uuid});
assertHasRangeDeletionDoc({conn: st.shard1, pending: true, whenToClean: "now", ns, uuid});
step4Failpoint.off();
// Assert that the recipient has 'numDocs' orphans.
step5Failpoint.wait();
assert.eq(numDocs, st.shard1.getDB(dbName).getCollection(collName).count());
step5Failpoint.off();
// Allow the moveChunk to finish.
awaitResult();
// Recipient shard eventually cleans up the orphans.
assert.soon(function() {
return st.shard1.getDB(dbName).getCollection(collName).count() === 0;
});
assert.eq(numDocs, st.s.getDB(dbName).getCollection(collName).find().itcount());
// The durable state for coordinating the migration is eventually cleaned up.
assertEventuallyDoesNotHaveMigrationCoordinatorDoc(st.shard0);
// TODO (SERVER-44159): Delete document from config.rangeDeletions when CollectionRangeDeleter
// finishes deleting a range
// assertEventuallyDoesNotHaveRangeDeletionDoc(st.shard0);
assertEventuallyDoesNotHaveRangeDeletionDoc(st.shard1);
migrationCommitVersionErrorFailpoint.off();
})();
st.stop();
})();
|
