mongodb全库备份与恢复

一次服务器异常事件导致mongodb服务挂掉无法重启，报错信息：

2017-10-10T14:35:16.078+0800 I NETWORK  [initandlisten] connection accepted from 192.168.1.52:55362 #4 (2 connections now open)
2017-10-10T14:35:16.111+0800 I QUERY    [conn4] query chelun.cw_user_position query: { $query: { uid: "22014082" }, $comment: "This is a query comment", $orderby: {} } planSummary: IXSCAN { uid: 1.0 } ntoreturn:999999 ntoskip:0 nscanned:1 nscannedObjects:1 keyUpdates:0 writeConflicts:0 numYields:1 nreturned:1 reslen:468 locks:{} 31ms
2017-10-10T14:35:16.166+0800 F -        [conn4] Invalid access at address: 0x7fe6e3645038
2017-10-10T14:35:16.195+0800 F -        [conn4] Got signal: 7 (Bus error).

 0xf662c9 0xf65942 0xf65c9e 0x7ff2f79787e0 0x7ff2f64636d5 0x7ff2f6ca5886 0xd2981c 0xd4902c 0xd50ad6 0xd49976 0x9046cc 0xc3f001 0xc3df09 0xa32f4c 0xa33705 0xbd54a4 0xbd5854 0xbd5e8d 0x9a7e29 0x9a919c 0x9a96db 0x9ac11d 0x9d0a04...skipping...

查询了资料，这种情况是mongodb异常关闭导致出错，尝试使用repair修复没有成功，找寻各种启动修复方法也均没有成功。好在服务不是核心业务，利用从库的备份（或者定期的备份数据）来恢复到一个新的库；

1、新建mongodb实例

假定服务端口定位27027，配置文件如下：

bind_ip=192.168.xx.xx
#指定服务器监听的端口，默认是27017
port=27027

#以守护进程的方式运行MongoDB
fork=true

objcheck=true

#2G
#oplogSize=2000

#pidfile
pidfilepath=/home/LBSUser/pidfile/mongo_27027.pid

#指定数据目录，默认是/data/db/。每个mongod进程都需要独立的目录，
#启动mongod时就会在数据目录中创建mongod.lock文件，防止其他mongod进程使用该数据目录。
dbpath=/data/LBSdb

#指定日志输出路径，如果不指定则会在终端输出。每次启动都会覆盖原来的日志，如果不想覆盖就要用--logappend选项
logpath=/data/log/mongodb_27027.log

logappend=true

master=true

#auth=true

profile=2

slowms=1

#only：默认为空，用于从选项，指定一个数据库进行复制
#only=test


#.禁止HTTP状态接口
nohttpinterface=true
#.禁止REST接口-在生产环境下建议不要启用MongoDB的REST接口
rest=false

#keyFile=/home/LBSUser/etc/keyfile

指定配置文件启动mongodb服务：

mongod -f /home/LBSUser/LBSmongo/mongodb.conf

（这里从库或者集群搭建略）

搭建好后，准备数据

二、导出数据

从原实例的从库节点中导出最新的数据备份（原端口是27017）

mongodump -h 192.168.xx.xx --port 27017 --oplog -o /data/backup/tmp27107/

导出记录：

2017-10-10T15:11:59.215+0800 [#######################.]  xx.cw_user_position  19578438/20169757  (97.1%) 2017-10-10T15:12:02.215+0800 [#######################.]  xx.cw_user_position  19834340/20169757  (98.3%) 2017-10-10T15:12:05.215+0800 [#######################.]  xx.cw_user_position  20113312/20169757  (99.7%) 2017-10-10T15:12:05.645+0800 writing xx.cw_user_position metadata to /data/backup/tmp27107/xx/cw_user_position.metadata.json 2017-10-10T15:12:05.653+0800 done dumping xx.cw_user_position 2017-10-10T15:12:05.654+0800 writing captured oplog to /data/backup/tmp27107/oplog.bson

看到这个后确认导出成功。

三、使用mongorestore导入数据

mongorestore导入数据可以指定库导入，我这个直接全量导入，不指定库：

mongorestore -h 192.168.xx.xx:27027 -d chelun --dir /data/backup/tmp27107/

注意这里是直接导入到我们新的机器上的新实例27027了。

2017-10-10T15:40:28.345+0800 [#######################.]  xx.cw_user_position  8.0 GB/8.2 GB  (97.6%) 2017-10-10T15:40:31.345+0800 [#######################.]  xx.cw_user_position  8.0 GB/8.2 GB  (97.9%) 2017-10-10T15:40:34.345+0800 [#######################.]  xx.cw_user_position  8.0 GB/8.2 GB  (98.2%) 2017-10-10T15:40:37.345+0800 [#######################.]  xx.cw_user_position  8.1 GB/8.2 GB  (98.5%) 2017-10-10T15:40:40.345+0800 [#######################.]  xx.cw_user_position  8.1 GB/8.2 GB  (98.8%) 2017-10-10T15:40:43.345+0800 [#######################.]  xx.cw_user_position  8.1 GB/8.2 GB  (99.1%) 2017-10-10T15:40:46.345+0800 [#######################.]  xx.cw_user_position  8.1 GB/8.2 GB  (99.3%) 2017-10-10T15:40:49.345+0800 [#######################.]  xx.cw_user_position  8.2 GB/8.2 GB  (99.6%) 2017-10-10T15:40:52.345+0800 [#######################.]  xx.cw_user_position  8.2 GB/8.2 GB  (99.9%) 2017-10-10T15:40:53.621+0800 restoring indexes for collection xx.cw_user_position from metadata 2017-10-10T15:44:46.613+0800 finished restoring xx.cw_user_position 2017-10-10T15:44:46.613+0800 restoring users from /data/backup/tmp27107/admin/system.users.bson 2017-10-10T15:44:47.294+0800 done

我的数据有点大，导入时间稍长。

根据导入的提示，确认导入都成功了；

四、登陆检查

做基础检查和抽样检查

[root@s0055-gz data]# mongo 192.168.xx.xx:27027 MongoDB shell version: 3.0.1 connecting to: 192.168.xx.xx:27027/test Server has startup warnings: 2017-10-10T15:12:13.752+0800 I CONTROL  [initandlisten] 2017-10-10T15:12:13.752+0800 I CONTROL  [initandlisten] ** WARNING: You are running on a NUMA machine. 2017-10-10T15:12:13.752+0800 I CONTROL  [initandlisten] **          We suggest launching mongod like this to avoid performance problems: 2017-10-10T15:12:13.752+0800 I CONTROL  [initandlisten] **              numactl --interleave=all mongod [other options] 2017-10-10T15:12:13.752+0800 I CONTROL  [initandlisten] 2017-10-10T15:12:13.752+0800 I CONTROL  [initandlisten] ** WARNING: /sys/kernel/mm/transparent_hugepage/enabled is 'always'. 2017-10-10T15:12:13.752+0800 I CONTROL  [initandlisten] **        We suggest setting it to 'never' 2017-10-10T15:12:13.752+0800 I CONTROL  [initandlisten] 2017-10-10T15:12:13.752+0800 I CONTROL  [initandlisten] ** WARNING: /sys/kernel/mm/transparent_hugepage/defrag is 'always'. 2017-10-10T15:12:13.752+0800 I CONTROL  [initandlisten] **        We suggest setting it to 'never' 2017-10-10T15:12:13.752+0800 I CONTROL  [initandlisten] > db.system.users.find() > use admin switched to db admin > db.system.users.find() > db.system.users.find() { "_id" : "admin.xx_user", "user" : "xx_user", "db" : "admin", "credentials" : { "SCRAM-SHA-1" : { "iterationCount" : 10000, "salt" : "IJUswtPOEvJ1qffHjACQIQ==", "storedKey" : "FoVPjbzw74bcs+8WCShuvluA7ssrU=", "serverKey" : "bQd4TlULaxd/6BMLQHoQtvInNJOw=" } }, "roles" : [ { "role" : "root", "db" : "admin" } ] } > show dbs; admin      0.078GB xx    13.947GB local      2.077GB xx2   0.078GB