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
|
/**
* This tests that slaveOk'd queries in sharded setups get correctly routed when a slave goes into
* RECOVERING state, and don't break
*/
// Shard secondaries are restarted, which may cause that shard's primary to stepdown while it does
// not see the secondaries. Either the primary connection gets reset, or the primary could change.
TestData.skipCheckingUUIDsConsistentAcrossCluster = true;
(function() {
'use strict';
load("jstests/replsets/rslib.js");
var shardTest =
new ShardingTest({name: "recovering_slaveok", shards: 2, mongos: 2, other: {rs: true}});
var mongos = shardTest.s0;
var mongosSOK = shardTest.s1;
mongosSOK.setSlaveOk();
var admin = mongos.getDB("admin");
var config = mongos.getDB("config");
const dbName = "test";
var dbase = mongos.getDB(dbName);
var coll = dbase.getCollection("foo");
var dbaseSOk = mongosSOK.getDB("" + dbase);
var collSOk = mongosSOK.getCollection("" + coll);
var rsA = shardTest.rs0;
var rsB = shardTest.rs1;
assert.commandWorked(rsA.getPrimary().getDB("test_a").dummy.insert({x: 1}));
assert.commandWorked(rsB.getPrimary().getDB("test_b").dummy.insert({x: 1}));
rsA.awaitReplication();
rsB.awaitReplication();
print("1: initial insert");
assert.commandWorked(coll.save({_id: -1, a: "a", date: new Date()}));
assert.commandWorked(coll.save({_id: 1, b: "b", date: new Date()}));
print("2: shard collection");
shardTest.shardColl(coll,
/* shardBy */ {_id: 1},
/* splitAt */ {_id: 0},
/* move chunk */ {_id: 0},
/* dbname */ null,
/* waitForDelete */ true);
print("3: test normal and slaveOk queries");
// Make shardA and rsA the same
var shardA = shardTest.getShard(coll, {_id: -1});
var shardAColl = shardA.getCollection("" + coll);
var shardB = shardTest.getShard(coll, {_id: 1});
if (shardA.name == rsB.getURL()) {
var swap = rsB;
rsB = rsA;
rsA = swap;
}
rsA.awaitReplication();
rsB.awaitReplication();
// Because of async migration cleanup, we need to wait for this condition to be true
assert.soon(function() {
return coll.find().itcount() == collSOk.find().itcount();
});
assert.eq(shardAColl.find().itcount(), 1);
assert.eq(shardAColl.findOne()._id, -1);
print("5: make one of the secondaries RECOVERING");
var secs = rsA.getSecondaries();
var goodSec = secs[0];
var badSec = secs[1];
assert.commandWorked(badSec.adminCommand("replSetMaintenance"));
rsA.waitForState(badSec, ReplSetTest.State.RECOVERING);
print("6: stop non-RECOVERING secondary");
rsA.stop(goodSec);
print("7: check our regular and slaveOk query");
assert.eq(2, coll.find().itcount());
assert.eq(2, collSOk.find().itcount());
print("8: restart both our secondaries clean");
rsA.restart(rsA.getSecondaries(), {remember: true, startClean: true}, undefined, 5 * 60 * 1000);
print("9: wait for recovery");
rsA.waitForState(rsA.getSecondaries(), ReplSetTest.State.SECONDARY, 5 * 60 * 1000);
print("10: check our regular and slaveOk query");
// We need to make sure our nodes are considered accessible from mongos - otherwise we fail
// See SERVER-7274
awaitRSClientHosts(coll.getMongo(), rsA.nodes, {ok: true});
awaitRSClientHosts(coll.getMongo(), rsB.nodes, {ok: true});
// We need to make sure at least one secondary is accessible from mongos - otherwise we fail
// See SERVER-7699
awaitRSClientHosts(collSOk.getMongo(), [rsA.getSecondaries()[0]], {secondary: true, ok: true});
awaitRSClientHosts(collSOk.getMongo(), [rsB.getSecondaries()[0]], {secondary: true, ok: true});
print("SlaveOK Query...");
var sOKCount = collSOk.find().itcount();
var collCount = null;
try {
print("Normal query...");
collCount = coll.find().itcount();
} catch (e) {
printjson(e);
// There may have been a stepdown caused by step 8, so we run this twice in a row. The first
// time can error out.
print("Error may have been caused by stepdown, try again.");
collCount = coll.find().itcount();
}
assert.eq(collCount, sOKCount);
shardTest.stop();
})();
|
