MongoDB修改oplog大小

那海蓝蓝 2022-05-28 我要评论

修改oplog有四种方法：

方法一

步骤如下：

停掉所有secondary节点
主节点删除local目录下文件，副本节点删除数据目录下所有文件
修改所有节点的配置文件，如：oplogSize=1000
重启所有节点，包括主节点和副本节点
重新配置replca set，副本节点会重新同步数据（initial sync）

优点：操作简单。
缺点：需要停服务，若数据量大，数据同步代价高。

方法二

步骤如下：

remove其中一个secondary节点，并关闭mongod服务，删除数据目录下所有文件
修改此节点的参数文件，如：oplogSize=1000，并启动mongod服务
在primary节点执行rs.add()将此节点加入到replica set
循环1-3步骤，将所有副本节点全部改完
primary节点执行rs.stepDown()将主节点降级为副本节点
从新的主节点remove掉此节点，并进行步骤1-3
这样逐个修改每个节点，完成oplog修改。

优点：解决了方法一中的停机问题。
缺点：每个节点都要逐个重新同步，时间代价更高。

方法三

The oplog exists internally as a capped collection, so you cannot modify its size in the course of normal operations.另：改变oplog大小，需要在每个节点上执行维护模式。（官方推荐）

1.关闭mongod

关闭mongod实例，如果是primary节点，执行rs.stepDown() 降级

handong1:PRIMARY> rs.stepDown()
{
	"ok" : 1,
	"$clusterTime" : {
		"clusterTime" : Timestamp(1619693040, 1),
		"signature" : {
			"hash" : BinData(0,"AAAAAAAAAAAAAAAAAAAAAAAAAAA="),
			"keyId" : NumberLong(0)
		}
	},
	"operationTime" : Timestamp(1619693040, 1)
}
handong1:SECONDARY> 
handong1:SECONDARY> 
handong1:SECONDARY> 
handong1:SECONDARY> use admin
switched to db admin
handong1:SECONDARY> db.shutdownServer()
2021-04-29T18:44:33.947+0800 I  NETWORK  [js] DBClientConnection failed to receive message from 127.0.0.1:27017 - HostUnreachable: Connection closed by peer
server should be down...
2021-04-29T18:44:33.950+0800 I  NETWORK  [js] trying reconnect to 127.0.0.1:27017 failed
2021-04-29T18:44:33.967+0800 I  NETWORK  [js] reconnect 127.0.0.1:27017 failed failed

2.修改配置文件

修改配置文件，修改端口，注释掉replSet和认证相关的设置

port=27018
fork=true
journal = true
maxConns=500
logappend=true
directoryperdb=true
dbpath=/mongodb/data
logpath=/mongodb/logs/mongodb.log

3.启动mongod实例，并备份oplog

mongod -f /mongodb/conf/mongodb.conf 
about to fork child process, waiting until server is ready for connections.
forked process: 31553
child process started successfully, parent exiting

mongodump -d local -c oplog.rs --port 27018 -h 172.16.254.134 -o /mongodb/backup
2021-04-29T18:55:18.167+0800	writing local.oplog.rs to /mongodb/backup/local/oplog.rs.bson
2021-04-29T18:55:18.170+0800	done dumping local.oplog.rs (798 documents)

4.重建oplog

保存oplog最新时间点

> use local
switched to db local
> db.temp.save( db.oplog.rs.find( { }, { ts: 1, h: 1 } ).sort( {$natural : -1} ).limit(1).next() )
WriteResult({ "nInserted" : 1 })

> db.temp.find()
{ "_id" : ObjectId("608a914089abaa981f14e888"), "ts" : Timestamp(1619693066, 1), "h" : NumberLong(0) }

删除旧的oplog

db.oplog.rs.drop()

重建新的oplog，大小为2GB

> db.runCommand( { create: "oplog.rs", capped: true, size: (2 * 1024 * 1024 * 1024) } )

5.插入前面保存的oplog时间点记录

> db.oplog.rs.save( db.temp.findOne() )
> db.oplog.rs.find()

6.关闭mongod实例

> use admin
switched to db admin
> db.shutdownServer()
2021-04-29T19:06:53.745+0800 I  NETWORK  [js] DBClientConnection failed to receive message from 127.0.0.1:27018 - HostUnreachable: Connection closed by peer
server should be down...
2021-04-29T19:06:53.749+0800 I  NETWORK  [js] trying reconnect to 127.0.0.1:27018 failed
2021-04-29T19:06:53.749+0800 I  NETWORK  [js] reconnect 127.0.0.1:27018 failed failed

最后恢复mongodb.conf到初始状态，启动

方法四

如果你的MongoDB版本为4.0以后的版本，可以直接使用replSetResizeOplog修改。

1.查看oplog大小

handong1:SECONDARY> db.getReplicationInfo()
{
	"logSizeMB" : 1000,
	"usedMB" : 0.17,
	"timeDiff" : 6736,
	"timeDiffHours" : 1.87,
	"tFirst" : "Thu Apr 29 2021 17:19:14 GMT+0800 (CST)",
	"tLast" : "Thu Apr 29 2021 19:11:30 GMT+0800 (CST)",
	"now" : "Thu Apr 29 2021 19:11:42 GMT+0800 (CST)"
}

2.修改oplog大小

handong1:SECONDARY> db.adminCommand({replSetResizeOplog:1,size:2000})
{
	"ok" : 1,
	"$clusterTime" : {
		"clusterTime" : Timestamp(1619694744, 14),
		"signature" : {
			"hash" : BinData(0,"AAAAAAAAAAAAAAAAAAAAAAAAAAA="),
			"keyId" : NumberLong(0)
		}
	},
	"operationTime" : Timestamp(1619694744, 14)
}

3.验证oplog大小

handong1:SECONDARY> db.getReplicationInfo()
{
	"logSizeMB" : 2000,
	"usedMB" : 0.18,
	"timeDiff" : 6852,
	"timeDiffHours" : 1.9,
	"tFirst" : "Thu Apr 29 2021 17:19:14 GMT+0800 (CST)",
	"tLast" : "Thu Apr 29 2021 19:13:26 GMT+0800 (CST)",
	"now" : "Thu Apr 29 2021 19:13:28 GMT+0800 (CST)"
}

4.整理碎片，回收空间（可选）

handong1:SECONDARY> use local
switched to db local
handong1:SECONDARY> db.runCommand({"compact" : "oplog.rs"})
{
	"ok" : 1,
	"$clusterTime" : {
		"clusterTime" : Timestamp(1619694840, 1),
		"signature" : {
			"hash" : BinData(0,"AAAAAAAAAAAAAAAAAAAAAAAAAAA="),
			"keyId" : NumberLong(0)
		}
	},
	"operationTime" : Timestamp(1619694840, 1)
}