基于 Delayed Secondary 实现时间点恢复

环境准备

初始化一个有 Delayed Secondary 节点的分片集群

1. 创建一个分片集群，其中单分片中 mongod 的数量为 4
 mlaunch --config 1 --mongos 1 --replicaset --nodes 4 --sharded sh_1
2. 连接到分片主节点，配置一个 Delayed Secondary 节点，设定延迟的时间为 3 分钟
mongo --port 27018

cfg = rs.conf()
cfg.members[3].priority = 0
cfg.members[3].hidden = true
cfg.members[3].slaveDelay = 180
rs.reconfig(cfg)

向集群里插入数据

1. 连接到 mongos, 插入数据
mongo

var doc = {
   "name": "Kylin Soong",
   "balance": 99.99
}
for (var i = 0; i < 20000; i++) {
   doc.accountNo = i
   db.account.insertOne( doc )
}
2. 等待 3 分钟后，在Delayed Secondary 节点查看数据已经存在
mongo --port 27021

db.account.count()
20000

db.account.find({accountNo: 10001})
{ "_id" : ObjectId("5df98ad3ba9be09784784b20"), "name" : "Kylin Soong", "balance" : 99.99, "accountNo" : 10001 }

人为制造误删除操作

1. 登录 mongos, 删除 accountNo 为 10000 的记录
db.account.remove({accountNo: 10001})
WriteResult({ "nRemoved" : 1 })
2. mongs 上查询，确保数据已经删除
db.account.find({accountNo: 10001})

记录恢复时间点，复制 DS 节点的数据文件

1. 连接到 DS 节点，确保数据存在，记录时间点
db.account.find({accountNo: 10001})
{ "_id" : ObjectId("5df98ad3ba9be09784784b21"), "name" : "Kylin Soong", "balance" : 99.99, "accountNo" : 10001 }

use local
db.oplog.rs.find().sort({$natural: -1}).limit(1).next().ts
Timestamp(1576645449, 1)
2. 停止 DS 节点**
use admin
db.shutdownServer({force: true})
3. 拷贝 DS 节点数据文件到新位置
mkdir duplicateDS

cp -r data/sh_1/rs4/* duplicateDS/
4. 单节点启动，指向新数据文件
mongod --dbpath duplicateDS/db --logpath duplicateDS/mongod.log --port 28001 --fork  --wiredTigerCacheSizeGB 1
5. 删除 local 数据库
mongo --port 28001

use local
db.dropDatabase()
6. 以复制集启动
use admin
db.shutdownServer()

mongod --replSet sh_restore  --dbpath duplicateDS/db --logpath duplicateDS/mongod.log --port 28001 --fork  --wiredTigerCacheSizeGB 1
7. 初始化新复制集
rs.initiate()

获取主节点上增量 Ops Log

1. 使用 mongodump 在分片集群的主节点上获取增量 Ops Log
mongodump --host 127.0.0.1:27018 -d local -c oplog.rs -q '{"ts": {"$gt": {"$timestamp": {"t": 1576645449, "i": 1}}}}'
Note
 时间点是 DS 上所获取到的时间。

找出恢复时间点

1. 使用 bsondump 查看Ops Log
bsondump dump/local/oplog.rs.bson
...
{"ts":{"$timestamp":{"t":1576646969,"i":1}},"t":{"$numberLong":"1"},"h":{"$numberLong":"0"},"v":{"$numberInt":"2"},"op":"n","ns":"","wall":{"$date":{"$numberLong":"1576646969858"}},"o":{"msg":"periodic noop"}}
{"ts":{"$timestamp":{"t":1576646979,"i":1}},"t":{"$numberLong":"1"},"h":{"$numberLong":"0"},"v":{"$numberInt":"2"},"op":"d","ns":"test.account","ui":{"$binary":{"base64":"80MbK3/1QvGwfEvch7y6zQ==","subType":"04"}},"wall":{"$date":{"$numberLong":"1576646979722"}},"o":{"_id":{"$oid":"5df9b2f7bceee8cb722b53b5"}}}
{"ts":{"$timestamp":{"t":1576646979,"i":2}},"t":{"$numberLong":"1"},"h":{"$numberLong":"0"},"v":{"$numberInt":"2"},"op":"d","ns":"test.account","ui":{"$binary":{"base64":"80MbK3/1QvGwfEvch7y6zQ==","subType":"04"}},"wall":{"$date":{"$numberLong":"1576646979722"}},"o":{"_id":{"$oid":"5df9b2f7bceee8cb722b53b6"}}}
{"ts":{"$timestamp":{"t":1576646979,"i":3}},"t":{"$numberLong":"1"},"h":{"$numberLong":"0"},"v":{"$numberInt":"2"},"op":"d","ns":"test.account","ui":{"$binary":{"base64":"80MbK3/1QvGwfEvch7y6zQ==","subType":"04"}},"wall":{"$date":{"$numberLong":"1576646979722"}},"o":{"_id":{"$oid":"5df9b2f7bceee8cb722b53b7"}}}
{"ts":{"$timestamp":{"t":1576646979,"i":4}},"t":{"$numberLong":"1"},"h":{"$numberLong":"0"},"v":{"$numberInt":"2"},"op":"d","ns":"test.account","ui":{"$binary":{"base64":"80MbK3/1QvGwfEvch7y6zQ==","subType":"04"}},"wall":{"$date":{"$numberLong":"1576646979722"}},"o":{"_id":{"$oid":"5df9b2f7bceee8cb722b53b8"}}}
{"ts":{"$timestamp":{"t":1576646979,"i":5}},"t":{"$numberLong":"1"},"h":{"$numberLong":"0"},"v":{"$numberInt":"2"},"op":"d","ns":"test.account","ui":{"$binary":{"base64":"80MbK3/1QvGwfEvch7y6zQ==","subType":"04"}},"wall":{"$date":{"$numberLong":"1576646979722"}},"o":{"_id":{"$oid":"5df9b2f7bceee8cb722b53b9"}}}
{"ts":{"$timestamp":{"t":1576646979,"i":6}},"t":{"$numberLong":"1"},"h":{"$numberLong":"0"},"v":{"$numberInt":"2"},"op":"d","ns":"test.account","ui":{"$binary":{"base64":"80MbK3/1QvGwfEvch7y6zQ==","subType":"04"}},"wall":{"$date":{"$numberLong":"1576646979722"}},"o":{"_id":{"$oid":"5df9b2f7bceee8cb722b53ba"}}}
{"ts":{"$timestamp":{"t":1576646979,"i":7}},"t":{"$numberLong":"1"},"h":{"$numberLong":"0"},"v":{"$numberInt":"2"},"op":"d","ns":"test.account","ui":{"$binary":{"base64":"80MbK3/1QvGwfEvch7y6zQ==","subType":"04"}},"wall":{"$date":{"$numberLong":"1576646979722"}},"o":{"_id":{"$oid":"5df9b2f7bceee8cb722b53bb"}}}
{"ts":{"$timestamp":{"t":1576646979,"i":8}},"t":{"$numberLong":"1"},"h":{"$numberLong":"0"},"v":{"$numberInt":"2"},"op":"d","ns":"test.account","ui":{"$binary":{"base64":"80MbK3/1QvGwfEvch7y6zQ==","subType":"04"}},"wall":{"$date":{"$numberLong":"1576646979722"}},"o":{"_id":{"$oid":"5df9b2f7bceee8cb722b53bc"}}}
{"ts":{"$timestamp":{"t":1576646979,"i":9}},"t":{"$numberLong":"1"},"h":{"$numberLong":"0"},"v":{"$numberInt":"2"},"op":"d","ns":"test.account","ui":{"$binary":{"base64":"80MbK3/1QvGwfEvch7y6zQ==","subType":"04"}},"wall":{"$date":{"$numberLong":"1576646979722"}},"o":{"_id":{"$oid":"5df9b2f7bceee8cb722b53bd"}}}
{"ts":{"$timestamp":{"t":1576646979,"i":10}},"t":{"$numberLong":"1"},"h":{"$numberLong":"0"},"v":{"$numberInt":"2"},"op":"d","ns":"test.account","ui":{"$binary":{"base64":"80MbK3/1QvGwfEvch7y6zQ==","subType":"04"}},"wall":{"$date":{"$numberLong":"1576646979723"}},"o":{"_id":{"$oid":"5df9b2f7bceee8cb722b53be"}}}
Note
 如上恢复时间点为 1576646979。

执行恢复

1. 创建一个空目录
mkdir empty
2. 执行恢复
mongorestore -h 127.0.0.1:28001 --oplogReplay --oplogFile dump/local/oplog.rs.bson --oplogLimit '1576646969:1' empty/

results matching ""

    No results matching ""