MongoDB数据量较大时如何构建索引--减少业务最少影响
在数据量较大或请求量较大,直接建立索引对性能有显著影响时,可以利用复制集(数据量较大时一般为线上环境,使用复制集为必然选择或者使用分片.)中部分机器宕机不影响复制集工作的特性,继而建立索引。备注：添加索引的表使用WT引擎，数据量有1.5亿左右。
1. 副本集配置参数
节点1：
$ more shard1.conf
dbpath=/data/users/mgousr01/mongodb/dbdata/shard1_1
logpath=/data/users/mgousr01/mongodb/logs/shard1_1.log
pidfilepath=/data/users/mgousr01/mongodb/dbdata/shard1_1/shard1-1.pid
directoryperdb=true
logappend=true
replSet=shard1
shardsvr=true
bind_ip=127.0.0.1,x.x.x.x
port=37017
oplogSize=9024
fork=true
#noprealloc=true
#auth=true
journal=true
profile=1
slowms=10
maxConns=12000
storageEngine = wiredTiger
wiredTigerCacheSizeGB=96
#clusterAuthMode=keyFile
keyFile=/data/users/mgousr01/mongodb/etc/keyFilers0.key
wiredTigerDirectoryForIndexes=on
wiredTigerCollectionBlockCompressor=zlib
wiredTigerJournalCompressor=zlib
节点2：
$ more shard2.conf
dbpath=/data/users/mgousr01/mongodb/dbdata/shard2_1
logpath=/data/users/mgousr01/mongodb/logs/shard2_1.log
pidfilepath=/data/users/mgousr01/mongodb/dbdata/shard2_1/shard2-1.pid
directoryperdb=true
logappend=true
replSet=shard1
shardsvr=true
bind_ip=127.0.0.1,x.x.x.x
port=37017
oplogSize=9024
fork=true
#noprealloc=true
#auth=true
journal=true
profile=1
slowms=10
maxConns=12000
storageEngine = wiredTiger
wiredTigerCacheSizeGB=96
#clusterAuthMode=keyFile
keyFile=/data/users/mgousr01/mongodb/etc/keyFilers0.key
wiredTigerDirectoryForIndexes=on
wiredTigerCollectionBlockCompressor=zlib
wiredTigerJournalCompressor=zlib
节点3：
$ more shard3.conf
dbpath=/data/users/mgousr01/mongodb/dbdata/shard3_1
logpath=/data/users/mgousr01/mongodb/logs/shard3_1.log
pidfilepath=/data/users/mgousr01/mongodb/dbdata/shard3_1/shard3-1.pid
directoryperdb=true
logappend=true
replSet=shard1
shardsvr=true
bind_ip=127.0.0.1,x.x.x.x
port=37017
oplogSize=9024
fork=true
#noprealloc=true
#auth=true
journal=true
profile=1
slowms=10
maxConns=12000
storageEngine = wiredTiger
wiredTigerCacheSizeGB=96
#clusterAuthMode=keyFile
keyFile=/data/users/mgousr01/mongodb/etc/keyFilers0.key
wiredTigerDirectoryForIndexes=on
wiredTigerCollectionBlockCompressor=zlib
wiredTigerJournalCompressor=zlib
2. 启动mongodb
mongod -f 启动
3. 配置副本集命令(登陆任意一台主机)
config={_id:'shard1',members:[{_id:0,host:'x.x.x.x:37017',priority:1,tags:{'use':'xxx'}},{_id:1,host:'x.x.x.x:37017',priority:1,tags:{'use':'xxx'}},{_id:2,host:'x.x.x.x:37017',priority:1,tags:{'use':'xxx'}}]}
rs.initiate(config)
4. 在primary库进行写操作模拟线上
for(i=0;i use chicago
Build the Index
> db.users.createIndex({username:1,created:1},{unique:true},{name:"username_created_unique"},{background:true})
【建议以后创建索引 要有命名规范】
(5)查看索引信息
> db.users.getIndexes()
[
{
"v" : 1,
"key" : {
"_id" : 1
},
"name" : "_id_",
"ns" : "chicago.users"
},
{
"v" : 1,
"unique" : true,
"key" : {
"username" : 1,
"created" : 1
},
"name" : "username_1_created_1",
"ns" : "chicago.users"
}
]
(6)再次停掉副本集的mongod进程
$ pwd
/data/users/mgousr01/mongodb/etc
$ mongod -f shard3.conf --shutdown
(7)启动mongod进程
将shard3.conf配置文件的replSet=shard1注释去掉；然后启动
mongod -f shard3.conf
mongo ip:37017/admin
启动后，会将节点加入到副本集中，然后同步primary数据，secondary上的索引不会对主造成影响导致主从不一致状况发生。
7.对第二个secondary副本操作--节点2
重复第6步即可。
8.构建所有secondarys索引大致步骤
For each secondary in the set, build an index according to the following steps:
(1)Stop One Secondary
(2)Build the Index
(3)Restart the Program mongod
9.构建primary节点索引
(1)登陆到主节点
mongo ip:37017/admin
(2) 将主节点降级
shard1:PRIMARY> rs.stepDown(30)
2016-04-19T12:49:44.423+0800 I NETWORKDBClientCursor::init call() failed
2016-04-19T12:49:44.426+0800 E QUERY Error: error doing query: failed
at DBQuery._exec (src/mongo/shell/query.js:83:36)
at DBQuery.hasNext (src/mongo/shell/query.js:240:10)
at DBCollection.findOne (src/mongo/shell/collection.js:187:19)
at DB.runCommand (src/mongo/shell/db.js:58:41)
at DB.adminCommand (src/mongo/shell/db.js:66:41)
at Function.rs.stepDown (src/mongo/shell/utils.js:1006:15)
at (shell):1:4 at src/mongo/shell/query.js:83
2016-04-19T12:49:44.427+0800 I NETWORKtrying reconnect to xxxx failed
2016-04-19T12:49:44.428+0800 I NETWORKreconnect xxxx ok
shard1:SECONDARY>
降级命令执行后，会主动变成secondary节点。上述两个secondary节点会有一个节点选举成为primary节点。
(3)后续构建索引的方法和第6步一样。
说明：
让primary降级：rs.stepDown(downseconds=60)，primary降级期间，它不会参与选举，如果降级时限过后，副本集还是没有primary，它会参与选举。
Preventing Electoins
让secondaries保持状态：rs.freeze(seconds)，这样就可以在这个时间内对primary做处理，而不用担心除非选举。
解除状态锁定：rs.freeze(0)。
页:
[1]