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
|
/**
* Verify that causal consistency is respected if a tenant migration commits with an earlier optime
* timestamp than the latest optime associated with cloning on the recipient.
*
* @tags: [requires_fcv_47, requires_majority_read_concern, incompatible_with_eft,
* incompatible_with_windows_tls, incompatible_with_macos, requires_persistence,
* featureFlagTenantMigrations]
*/
(function() {
"use strict";
load("jstests/libs/fail_point_util.js");
load("jstests/libs/uuid_util.js");
load("jstests/replsets/libs/tenant_migration_test.js");
load("jstests/replsets/rslib.js");
function assertCanFindWithReadConcern(conn, dbName, collName, expectedDoc, readConcern) {
const res = assert.commandWorked(conn.getDB(dbName).runCommand({find: collName, readConcern}));
assert(res.cursor.firstBatch, tojson(res));
assert.eq(1, res.cursor.firstBatch.length, tojson(res));
assert.eq(expectedDoc, res.cursor.firstBatch[0], tojson(res));
}
let counter = 0;
let makeTenantId = function() {
return "tenant_" + counter++;
};
// Local read concern case.
(() => {
const migrationX509Options = TenantMigrationUtil.makeX509OptionsForTest();
// Simulate a lagged node by setting secondaryDelaySecs on one recipient secondary. Verify this
// does not prevent reading all the tenant's data after the migration commits.
const recipientRst = new ReplSetTest(
{name: "recipient_local_case", nodes: 3, nodeOptions: migrationX509Options.recipient});
recipientRst.startSet();
let config = recipientRst.getReplSetConfig();
config.members[2].priority = 0;
config.members[2][selectDelayFieldName(recipientRst)] = 5;
recipientRst.initiate(config);
const normalSecondary = recipientRst.getSecondaries()[0];
const laggedSecondary = recipientRst.getSecondaries()[1];
const tmt = new TenantMigrationTest({name: jsTestName(), recipientRst});
const tenantId = makeTenantId();
const migrationOpts = {
migrationIdString: extractUUIDFromObject(UUID()),
tenantId: tenantId,
};
const dbName = tenantId + "_test";
const collName = "foo";
// Insert tenant data to be copied. Save the operationTime to use for afterClusterTime reads
// later, simulating a client in a causally consistent session.
const insertRes = assert.commandWorked(tmt.getDonorPrimary().getDB(dbName).runCommand(
"insert", {insert: collName, documents: [{_id: 0, x: 0}]}));
assert(insertRes.operationTime, tojson(insertRes));
// Start a migration and pause the recipient before it copies documents from the donor.
const hangAfterCreateCollectionFp = configureFailPoint(
tmt.getRecipientRst().getPrimary(), "tenantCollectionClonerHangAfterCreateCollection");
assert.commandWorked(tmt.startMigration(migrationOpts));
hangAfterCreateCollectionFp.wait();
// Do writes on the recipient to advance its cluster time past the donor's.
let bulk = tmt.getRecipientPrimary().getDB("unrelatedDB").bar.initializeUnorderedBulkOp();
for (let i = 0; i < 100; i++) {
bulk.insert({x: i});
}
assert.commandWorked(bulk.execute());
// Allow the migration to complete. The cloned op should be written with a later opTime on the
// recipient than the migration commits with on the donor.
hangAfterCreateCollectionFp.off();
TenantMigrationTest.assertCommitted(tmt.waitForMigrationToComplete(migrationOpts));
// Local reads should always see all the tenant's data, with or without afterClusterTime.
assertCanFindWithReadConcern(
laggedSecondary, dbName, collName, {_id: 0, x: 0}, {level: "local"});
assertCanFindWithReadConcern(laggedSecondary,
dbName,
collName,
{_id: 0, x: 0},
{level: "local", afterClusterTime: insertRes.operationTime});
assertCanFindWithReadConcern(
normalSecondary, dbName, collName, {_id: 0, x: 0}, {level: "local"});
assertCanFindWithReadConcern(normalSecondary,
dbName,
collName,
{_id: 0, x: 0},
{level: "local", afterClusterTime: insertRes.operationTime});
recipientRst.stopSet();
tmt.stop();
})();
// Majority read concern case.
(() => {
const tmt = new TenantMigrationTest({name: jsTestName(), sharedOptions: {nodes: 3}});
const tenantId = makeTenantId();
const migrationOpts = {
migrationIdString: extractUUIDFromObject(UUID()),
tenantId: tenantId,
};
const dbName = tenantId + "_test";
const collName = "foo";
// Insert tenant data to be copied.
const insertRes = assert.commandWorked(tmt.getDonorPrimary().getDB(dbName).runCommand(
"insert", {insert: collName, documents: [{_id: 0, x: 0}]}));
assert(insertRes.operationTime, tojson(insertRes));
// Simulate a lagged node by disabling snapshotting on a secondary so it cannot advance its
// majority committed snapshot to include the block timestamp. Verify this prevents reading
// from it until the majority snapshot advances.
const laggedSecondary = tmt.getRecipientRst().getSecondaries()[0];
const normalSecondary = tmt.getRecipientRst().getSecondaries()[1];
// Start a migration and pause the recipient before it copies documents from the donor. Disable
// snapshotting after waiting for the last op to become committed, so a last committed snapshot
// exists but does not contain any documents from the donor.
const hangAfterCreateCollectionFp = configureFailPoint(
tmt.getRecipientRst().getPrimary(), "tenantCollectionClonerHangAfterCreateCollection");
assert.commandWorked(tmt.startMigration(migrationOpts));
hangAfterCreateCollectionFp.wait();
tmt.getRecipientRst().awaitLastOpCommitted();
const snapshotFp = configureFailPoint(laggedSecondary, "disableSnapshotting");
// Do writes on the recipient to advance its cluster time past the donor's and advance the
// majority committed snapshot on the non-lagged nodes.
let bulk = tmt.getRecipientPrimary().getDB("unrelatedDB").bar.initializeUnorderedBulkOp();
for (let i = 0; i < 100; i++) {
bulk.insert({x: i});
}
assert.commandWorked(bulk.execute());
// Allow the migration to complete. The cloned op should commit with a later opTime on the
// recipient than the migration commits with on the donor.
hangAfterCreateCollectionFp.off();
TenantMigrationTest.assertCommitted(tmt.waitForMigrationToComplete(migrationOpts));
// Verify majority reads cannot be served on the lagged recipient secondary with or without
// afterClusterTime because its majority committed snapshot is behind the block timestamp.
assert.commandFailedWithCode(
laggedSecondary.getDB(dbName).runCommand(
{find: collName, readConcern: {level: "majority"}, maxTimeMS: 2000}),
ErrorCodes.MaxTimeMSExpired);
assert.commandFailedWithCode(laggedSecondary.getDB(dbName).runCommand({
find: collName,
readConcern: {level: "majority", afterClusterTime: insertRes.operationTime},
maxTimeMS: 2000
}),
ErrorCodes.MaxTimeMSExpired);
// Reads on the normal secondary should succeed.
assertCanFindWithReadConcern(
normalSecondary, dbName, collName, {_id: 0, x: 0}, {level: "majority"});
assertCanFindWithReadConcern(normalSecondary,
dbName,
collName,
{_id: 0, x: 0},
{level: "majority", afterClusterTime: insertRes.operationTime});
// When the snapshot is allowed to advance, verify majority reads can now be served on all
// nodes. Insert to trigger advancing the committed snapshot.
snapshotFp.off();
assert.commandWorked(tmt.getRecipientPrimary().getDB(collName).bar.insert({x: "unrelated"}));
assertCanFindWithReadConcern(
laggedSecondary, dbName, collName, {_id: 0, x: 0}, {level: "majority"});
assertCanFindWithReadConcern(laggedSecondary,
dbName,
collName,
{_id: 0, x: 0},
{level: "majority", afterClusterTime: insertRes.operationTime});
assertCanFindWithReadConcern(
normalSecondary, dbName, collName, {_id: 0, x: 0}, {level: "majority"});
assertCanFindWithReadConcern(normalSecondary,
dbName,
collName,
{_id: 0, x: 0},
{level: "majority", afterClusterTime: insertRes.operationTime});
tmt.stop();
})();
// Snapshot read concern is tested in replsets/tenant_migration_concurrent_reads_on_recipient.js
})();
|
