程序博客网 > linux下载命令sz

mongodb 分片底层的主备切换是否会影响分片集群对外服务的详细剖析

来源：互联网 发布：linux下载命令sz 编辑：程序博客网 时间：2024/06/01 10:29



1、先在shard1上准备数据

库：ysdb1

集合：c1

数据：db.c1.save({x:1});……db.c1.save({x:4});

 

[mongodb@hch_test_dbm1_121_62 mongodb]$ /usr/local/mongodb-linux-x86_64-3.0.3/bin/mongo mongodb1:37027/admin

MongoDB shell version: 3.0.3

connecting to: mongodb1:37027/admin

Server has startup warnings:

2016-07-11T14:38:24.351+0800 I CONTROL  [initandlisten]

2016-07-11T14:38:24.351+0800 I CONTROL  [initandlisten] ** WARNING: You are running on a NUMA machine.

2016-07-11T14:38:24.351+0800 I CONTROL  [initandlisten] **          We suggest launching mongod like this to avoid performance problems:

2016-07-11T14:38:24.351+0800 I CONTROL  [initandlisten] **              numactl --interleave=all mongod [other options]

2016-07-11T14:38:24.351+0800 I CONTROL  [initandlisten]

2016-07-11T14:38:24.351+0800 I CONTROL  [initandlisten] ** WARNING: /sys/kernel/mm/transparent_hugepage/enabled is 'always'.

2016-07-11T14:38:24.351+0800 I CONTROL  [initandlisten] **        We suggest setting it to 'never'

2016-07-11T14:38:24.351+0800 I CONTROL  [initandlisten]

2016-07-11T14:38:24.352+0800 I CONTROL  [initandlisten] ** WARNING: /sys/kernel/mm/transparent_hugepage/defrag is 'always'.

2016-07-11T14:38:24.352+0800 I CONTROL  [initandlisten] **        We suggest setting it to 'never'

2016-07-11T14:38:24.352+0800 I CONTROL  [initandlisten]

shard1:PRIMARY>

shard1:PRIMARY>

shard1:PRIMARY> use ysdb1;

switched to db ysdb1

shard1:PRIMARY> db.c1.save({x:1});

WriteResult({ "nInserted" : 1 })

shard1:PRIMARY> db.c1.save({x:2});

WriteResult({ "nInserted" : 1 })

shard1:PRIMARY> db.c1.save({x:3});

WriteResult({ "nInserted" : 1 })

shard1:PRIMARY> db.c1.save({x:4});

WriteResult({ "nInserted" : 1 })

shard1:PRIMARY>

 

 

2、去mongos上check新建的库集合

 

[mongodb@hch_test_dbm1_121_63 ~]$ /usr/local/mongodb-linux-x86_64-3.0.3/bin/mongo mongodb1:40000

MongoDB shell version: 3.0.3

connecting to: mongodb1:40000/test

mongos> show dbs;

admin   (empty)

bg      0.078GB

config  0.016GB

ysdb1   0.078GB  # 库存在

mongos> use ysdb1;

switched to db ysdb1

mongos> db.c1.find();

mongos> show tables; # 表不存在，也就意味着mongos severs默认连接的不是shard1而是shard2

mongos>

 

PS：看来需要在mongodb上mongos上面录入测试数据，然后在shard2上面做主备切换，看mongos客户端上是否能正常连接使用。

 

3、在mongos servers上录入数据

[mongodb@hch_test_dbm1_121_63 ~]$ /usr/local/mongodb-linux-x86_64-3.0.3/bin/mongo mongodb1:40000

MongoDB shell version: 3.0.3

connecting to: mongodb1:40000/test

mongos> use ysdb1;

switched to db ysdb1

mongos> show dbs;

admin   (empty)

bg      0.078GB

config  0.016GB

ysdb1   0.078GB

mongos> show tables;

mongos> db.c1.save({x:5});

WriteResult({ "nInserted" : 1 })

mongos> db.c1.save({x:6});

WriteResult({ "nInserted" : 1 })

mongos> db.c1.save({x:7});

WriteResult({ "nInserted" : 1 })

mongos>

mongos> db.c1.find();

{ "_id" : ObjectId("5783542d549dd66d4ac66d52"), "x" : 5 }

{ "_id" : ObjectId("57835430549dd66d4ac66d53"), "x" : 6 }

{ "_id" : ObjectId("57835432549dd66d4ac66d54"), "x" : 7 }

mongos>

 

 

4、然后去shard2上进行主备切换操作

[mongodb@hch_test_dbm1_121_62 mongodb]$ /usr/local/mongodb-linux-x86_64-3.0.3/bin/mongo mongodb1:37028/admin

MongoDB shell version: 3.0.3

connecting to: mongodb1:37028/admin

Server has startup warnings:

2016-07-09T19:08:58.250+0800 I CONTROL  [initandlisten]

2016-07-09T19:08:58.250+0800 I CONTROL  [initandlisten] ** WARNING: You are running on a NUMA machine.

2016-07-09T19:08:58.250+0800 I CONTROL  [initandlisten] **          We suggest launching mongod like this to avoid performance problems:

2016-07-09T19:08:58.250+0800 I CONTROL  [initandlisten] **              numactl --interleave=all mongod [other options]

2016-07-09T19:08:58.251+0800 I CONTROL  [initandlisten]

2016-07-09T19:08:58.251+0800 I CONTROL  [initandlisten] ** WARNING: /sys/kernel/mm/transparent_hugepage/enabled is 'always'.

2016-07-09T19:08:58.251+0800 I CONTROL  [initandlisten] **        We suggest setting it to 'never'

2016-07-09T19:08:58.251+0800 I CONTROL  [initandlisten]

2016-07-09T19:08:58.251+0800 I CONTROL  [initandlisten] ** WARNING: /sys/kernel/mm/transparent_hugepage/defrag is 'always'.

2016-07-09T19:08:58.251+0800 I CONTROL  [initandlisten] **        We suggest setting it to 'never'

2016-07-09T19:08:58.251+0800 I CONTROL  [initandlisten]

shard2:PRIMARY> use ysdb1;

switched to db ysdb1

shard2:PRIMARY> db.c1.find();

{ "_id" : ObjectId("5783542d549dd66d4ac66d52"), "x" : 5 }

{ "_id" : ObjectId("57835430549dd66d4ac66d53"), "x" : 6 }

{ "_id" : ObjectId("57835432549dd66d4ac66d54"), "x" : 7 }

shard2:PRIMARY>

shard2:PRIMARY> rs.stepDown();

2016-07-11T16:11:30.274+0800 I NETWORK  DBClientCursor::init call() failed

2016-07-11T16:11:30.281+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

2016-07-11T16:11:30.283+0800 I NETWORK  trying reconnect to mongodb1:37028 (192.168.121.62) failed

2016-07-11T16:11:30.283+0800 I NETWORK  reconnect mongodb1:37028 (192.168.121.62) ok

shard2:SECONDARY>

 

 

5、然后再去mongos客户端上进行数据变更

在原来打开的mongos上进行操作，录入一条新的数据db.c1.save({x:8});：然后回车执行，发现执行成功，也就是意味着shard2上的主备切换基本不影响客户端对集群的正常操作。

mongos> db.c1.save({x:8});

WriteResult({ "nInserted" : 1 })

mongos> db.c1.find();

{ "_id" : ObjectId("5783542d549dd66d4ac66d52"), "x" : 5 }

{ "_id" : ObjectId("57835430549dd66d4ac66d53"), "x" : 6 }

{ "_id" : ObjectId("57835432549dd66d4ac66d54"), "x" : 7 }

{ "_id" : ObjectId("578354a1549dd66d4ac66d55"), "x" : 8 }

mongos>

 

 

6、去shar1和shard2上验证新录入的数据

然后去验证mongos上的操作是否已经准确录入到了shard2分片上。

[mongodb@hch_test_dbm1_121_62 mongodb]$ /usr/local/mongodb-linux-x86_64-3.0.3/bin/mongo mongodb1:37028/ysdb1

MongoDB shell version: 3.0.3

connecting to: mongodb1:37028/ysdb1

Server has startup warnings:

2016-07-09T19:08:58.250+0800 I CONTROL  [initandlisten]

2016-07-09T19:08:58.250+0800 I CONTROL  [initandlisten] ** WARNING: You are running on a NUMA machine.

2016-07-09T19:08:58.250+0800 I CONTROL  [initandlisten] **          We suggest launching mongod like this to avoid performance problems:

2016-07-09T19:08:58.250+0800 I CONTROL  [initandlisten] **              numactl --interleave=all mongod [other options]

2016-07-09T19:08:58.251+0800 I CONTROL  [initandlisten]

2016-07-09T19:08:58.251+0800 I CONTROL  [initandlisten] ** WARNING: /sys/kernel/mm/transparent_hugepage/enabled is 'always'.

2016-07-09T19:08:58.251+0800 I CONTROL  [initandlisten] **        We suggest setting it to 'never'

2016-07-09T19:08:58.251+0800 I CONTROL  [initandlisten]

2016-07-09T19:08:58.251+0800 I CONTROL  [initandlisten] ** WARNING: /sys/kernel/mm/transparent_hugepage/defrag is 'always'.

2016-07-09T19:08:58.251+0800 I CONTROL  [initandlisten] **        We suggest setting it to 'never'

2016-07-09T19:08:58.251+0800 I CONTROL  [initandlisten]

shard2:PRIMARY> db.c1.find();

{ "_id" : ObjectId("5783542d549dd66d4ac66d52"), "x" : 5 }

{ "_id" : ObjectId("57835430549dd66d4ac66d53"), "x" : 6 }

{ "_id" : ObjectId("57835432549dd66d4ac66d54"), "x" : 7 }

{ "_id" : ObjectId("578354a1549dd66d4ac66d55"), "x" : 8 }

shard2:PRIMARY>  # 看到了shard2上有新添加的x:8这条数据，表示数据已经录入进来。

 

然后去shard1上查看数据，则没有记录：

[mongodb@hch_test_dbm1_121_62 mongodb]$ /usr/local/mongodb-linux-x86_64-3.0.3/bin/mongo mongodb1:37027/ysdb1

MongoDB shell version: 3.0.3

connecting to: mongodb1:37027/ysdb1

Server has startup warnings:

2016-07-11T14:38:24.351+0800 I CONTROL  [initandlisten]

2016-07-11T14:38:24.351+0800 I CONTROL  [initandlisten] ** WARNING: You are running on a NUMA machine.

2016-07-11T14:38:24.351+0800 I CONTROL  [initandlisten] **          We suggest launching mongod like this to avoid performance problems:

2016-07-11T14:38:24.351+0800 I CONTROL  [initandlisten] **              numactl --interleave=all mongod [other options]

2016-07-11T14:38:24.351+0800 I CONTROL  [initandlisten]

2016-07-11T14:38:24.351+0800 I CONTROL  [initandlisten] ** WARNING: /sys/kernel/mm/transparent_hugepage/enabled is 'always'.

2016-07-11T14:38:24.351+0800 I CONTROL  [initandlisten] **        We suggest setting it to 'never'

2016-07-11T14:38:24.351+0800 I CONTROL  [initandlisten]

2016-07-11T14:38:24.352+0800 I CONTROL  [initandlisten] ** WARNING: /sys/kernel/mm/transparent_hugepage/defrag is 'always'.

2016-07-11T14:38:24.352+0800 I CONTROL  [initandlisten] **        We suggest setting it to 'never'

2016-07-11T14:38:24.352+0800 I CONTROL  [initandlisten]

shard1:PRIMARY> db.c1.find();

{ "_id" : ObjectId("578352598ffbf6368275706d"), "x" : 1 }

{ "_id" : ObjectId("5783525b8ffbf6368275706e"), "x" : 2 }

{ "_id" : ObjectId("5783525d8ffbf6368275706f"), "x" : 3 }

{ "_id" : ObjectId("578352618ffbf63682757070"), "x" : 4 }

shard1:PRIMARY>

 

7、分片主备切换操作记录总结

结论：当分片shard2进行主备切换时，对外面连接集群的mongos客户端来说，是透明的，是无影响的。这也就意味着，mongodb分片集群底层的主备故障切换不影响业务应用程序的正常使用，这对于高可用是非常有效的。

3 0
linux下载命令sz linux下载命令sz
原创粉丝点击
热门IT博客
梦想海贼王转盘数据梦想海贼王转盘数据
备忘录java应用场景备忘录java应用场景
ppt数据图表制作
辛普森悖论 知乎
python字符串内置函数
学汉字软件
仿瓜子二手车网站源码
激战2n卡优化
好看的淘宝店铺头像
新网域名转入阿里云
王安忆长恨歌知乎
阿里云创业创新基地
西门子plc编程线
17年机械行业数据采集
执业药师网络兼职
软考程序员考试时间
最新伤感歌曲 网络
查看应用程序端口
京东客户端下载mac版
淘宝助手官网下载
热门问题 老师的惩罚 人脸识别 我在镇武司摸鱼那些年 重生之率土为王 我在大康的咸鱼生活 盘龙之生命进化 天生仙种 凡人之先天五行 春回大明朝 姑娘不必设防，我是瞎子 优乐酷 酷噻网 酷贷网 潮衣酷 酷锋 酷云tv 酷米动画下载 雪弗莱创酷多少钱 酷贝 酷开 42k1 帅酷m10 酷跑x11 酷开电视升级 酷比 手机 韩酷平板电脑 创维酷开怎么样 酷佩 酷美图 名车酷 酷开40u1 酷开42k1 四酷 酷6 酷6美股 6酷网 酷6网页版 ku6 酷六网 雅迪酷k 酷开43k2 酷开k50j 酷开k55j 创维酷k 酷k 酷歌k 酷开42k1电视怎么样 酷开k55价格 创维酷开4k电视怎么样 酷开电视42k1 创维酷开k50 酷开tv42k1y

程序博客网，程序员的互联网技术博客家园。csdn论坛精品 msdn技术资料都在这里