欢迎您访问程序员文章站本站旨在为大家提供分享程序员计算机编程知识!
您现在的位置是: 首页  >  IT编程

MySQLNDBCluster的安装教程

程序员文章站 2022-06-24 20:49:59
MySQL集群是一个无共享的(share-nothing)、分布式节点架构的存储方案,目的是提供容错性和高性能。 MySQL NDB Cluster架构如图: MySQL...

MySQL集群是一个无共享的(share-nothing)、分布式节点架构的存储方案,目的是提供容错性和高性能。

MySQL NDB Cluster架构如图:

MySQLNDBCluster的安装教程

MySQL NDB Cluster架构按照节点类型分为三部分: 管理节点:管理节点通过对配置文件conifg.ini的维护来对其他节点进行管理。该文件可以用来配置有多少副本需要维护、在数据节点上为数据和索引分配多少内存、数据节点的位置、数据节点上保存数据的磁盘的位置,SQL节点的位置信息等,管理节点只能有一个。SQL节点:SQL节点可以理解为应用程序和数据节点的一个桥梁,应用程序不能直接访问数据节点,只能先访问SQL节点,然后SQL节点再去访问数据节点来返回数据,Cluster中可以有多个SQL节点,通过每个SQL节点查询到的数据都是一致的,一般来说,SQL节点越多,分配到每个SQL节点的负载就越小,系统的整体性能就越好;数据节点:数据节点用来存放数据,可有多个数据节点。

下面演示MySQL NDB Cluster的安装:

1、环境信息:

管理节点Mgmd:10.24.32.180
SQL节点Mysqld1:10.24.32.181
SQL节点Mysqld2:10.24.32.182
数据节点Ndbd1:10.24.32.183
数据节点Ndbd2:10.24.32.184

2、下载MySQL NDB Cluster软件;

[root@strong /]# ll /u01/
total 873104
-rw-r--r--. 1 root root 894037512 Mar  6 13:26 mysql-cluster-gpl-7.5.9-linux-glibc2.12-x86_64.tar.gz
[root@strong /]# 

3、安装SQL节点,在每个SQL节点都执行如下操作; 1)创建mysql用户和组,如果该用户和组不存在:

[root@strong ~]# grep mysql /etc/passwd
[root@strong ~]# grep mysql /etc/group
[root@strong ~]# groupadd mysql
[root@strong ~]# useradd -g mysql -s /bin/false mysql
[root@strong ~]# 

2)解压缩 mysql-cluster软件;

[root@strong local]# tar zxvf /u01/mysql-cluster-gpl-7.5.9-linux-glibc2.12-x86_64.tar.gz
[root@strong local]# ln -s mysql-cluster-gpl-7.5.9-linux-glibc2.12-x86_64/ mysql
[root@strong local]# 

3)设置数据库系统

[root@strong local]# cd mysql
[root@strong mysql]# ./bin/mysqld --initialize --user=mysql
2018-03-06T11:04:27.469235Z 0 [Warning] TIMESTAMP with implicit DEFAULT value is deprecated. Please use --explicit_defaults_for_timestamp server option (see documentation for more details).
2018-03-06T11:04:28.202258Z 0 [Warning] InnoDB: New log files created, LSN=45790
2018-03-06T11:04:28.291271Z 0 [Warning] InnoDB: Creating foreign key constraint system tables.
2018-03-06T11:04:28.360238Z 0 [Warning] No existing UUID has been found, so we assume that this is the first time that this server has been started. Generating a new UUID: 238f3d44-212e-11e8-a946-000c298da433.
2018-03-06T11:04:28.365955Z 0 [Warning] Gtid table is not ready to be used. Table 'mysql.gtid_executed' cannot be opened.
2018-03-06T11:04:28.371758Z 1 [Note] A temporary password is generated for root@localhost: 9gidWnDuEo_:

4)拷贝MySQL启动脚本;

[root@strong mysql]# cp support-files/mysql.server /etc/init.d/mysql.server

5)修改Root密码;

[root@strong mysql]# service mysql.server start
Starting MySQL.Logging to '/usr/local/mysql/data/strong.mysql.com.err'.
..... SUCCESS! 
[root@strong mysql]# ./bin/mysql -uroot -p
Enter password: 
Welcome to the MySQL monitor.  Commands end with ; or \g.
Your MySQL connection id is 5
Server version: 5.7.21-ndb-7.5.9-cluster-gpl

Copyright (c) 2000, 2018, Oracle and/or its affiliates. All rights reserved.

Oracle is a registered trademark of Oracle Corporation and/or its
affiliates. Other names may be trademarks of their respective
owners.

Type 'help;' or '\h' for help. Type '\c' to clear the current input statement.

mysql> alter user 'root'@'localhost' identified by 'root';
Query OK, 0 rows affected (0.00 sec)

mysql> 

4、安装数据节点,在每个SQL节点都执行如下操作; 将ndbd、ndbmtd拷贝至/usr/local/bin/目录中:

[root@strong u01]# tar zxvf mysql-cluster-gpl-7.5.9-linux-glibc2.12-x86_64.tar.gz
[root@strong u01]# cp mysql-cluster-gpl-7.5.9-linux-glibc2.12-x86_64/bin/ndbd /usr/local/bin/ndbd
[root@strong u01]# cp mysql-cluster-gpl-7.5.9-linux-glibc2.12-x86_64/bin/ndbmtd /usr/local/bin/ndbmtd
[root@strong u01]# ll /usr/local/bin/
total 38280
-rwxr-xr-x. 1 root root 39197765 Mar  6 19:38 ndbd
-rwxr-xr-x. 1 root root        0 Mar  6 19:39 ndbmtd
[root@strong u01]# 

5、安装管理节点; 将ndb_mgm、ndb_mgmd拷贝至/usr/local/bin/目录中:

[root@strong local]# scp mysql-cluster-gpl-7.5.9-linux-glibc2.12-x86_64/bin/ndb_mgm* root@10.24.32.180:/usr/local/bin/
The authenticity of host '10.24.32.180 (10.24.32.180)' can't be established.
ECDSA key fingerprint is 5b:97:d1:f5:8b:84:a8:ac:9b:bf:b2:b4:0d:71:92:a5.
Are you sure you want to continue connecting (yes/no)? yes
Warning: Permanently added '10.24.32.180' (ECDSA) to the list of known hosts.
root@10.24.32.180's password: 
ndb_mgm                                                                                                             100% 7940KB   7.8MB/s   00:00    
ndb_mgmd                                                                                                            100%   17MB  17.3MB/s   00:00    
[root@strong local]# 

6、配置数据节点和SQL节点; 每个数据节点和SQL节点都需要一个my.cnf配置文件,该文件包括 连接串ndb-connectstring:告诉节点到哪里查找到管理节点信息;存储引擎信息:ndbcluster。配置信息:

[root@strong ~]# vi /etc/my.cnf 
[mysqld]

ndbcluster #run NDB storage engine

[mysql_cluster]

ndb-connectstring=10.24.32.180  #location of management server

7、配置管理节点; 创建目录,放置配置文件:

[root@strong local]# mkdir mysql
[root@strong local]# cd mysql/
[root@strong mysql]# mkdir mysql-cluster
[root@strong mysql-cluster]# vi config.ini 
[ndbd default]

#Opitions affecting ndbd processes on all nodes:

NoOfReplicas=1  #Number of replicas


[ndb_mgmd]

HostName=10.24.32.180              #Hostname or IP address of MGM node
DataDir=/var/lib/mysql-cluster     #Dir for MGM node log files

[ndbd]

HostName=10.24.32.183
NodeId=2
DataDir=/u01

[ndbd]

HostName=10.24.32.184
NodeId=3
DataDir=/u01

[mysqld]

HostName=10.24.32.181

[mysqld]

HostName=10.24.32.182

8、启动Cluster; Cluster需要各个节点都进行启动后才可以运行,节点的启动顺序为管理节点——>数据节点——>SQL节点。 启动管理节点:

[root@strong mysql-cluster]# ndb_mgmd -f ./config.ini 
MySQL Cluster Management Server mysql-5.7.21 ndb-7.5.9
[root@strong mysql-cluster]# ps aux |grep ndb_mgmd
root      8155  2.5  0.6 504480  3380 ?        Ssl  20:48   0:00 ndb_mgmd -f ./config.ini
root      8166 13.0  0.1 112632   968 pts/0    S+   20:48   0:00 grep --color=auto ndb_mgmd
[root@strong mysql-cluster]# 

启动数据节点:

[root@strong local]# ndbd
2018-03-06 20:51:40 [ndbd] INFO     -- Angel connected to '10.24.32.180:1186'
2018-03-06 20:51:40 [ndbd] INFO     -- Angel allocated nodeid: 2

启动SQL节点:

[root@strong ~]# service mysql.server start
Starting MySQL........................................................................................................................................... SUCCESS! 

9、节点全部成功启动后,在管理节点查看集群状态;

[root@strong mysql-cluster]# ndb_mgm
-- NDB Cluster -- Management Client --
ndb_mgm> show
Connected to Management Server at: localhost:1186
Cluster Configuration
---------------------
[ndbd(NDB)]	2 node(s)
id=2	@10.24.32.183  (mysql-5.7.21 ndb-7.5.9, Nodegroup: 0, *)
id=3	@10.24.32.184  (mysql-5.7.21 ndb-7.5.9, Nodegroup: 1)

[ndb_mgmd(MGM)]	1 node(s)
id=1	@10.24.32.180  (mysql-5.7.21 ndb-7.5.9)

[mysqld(API)]	2 node(s)
id=4	@10.24.32.181  (mysql-5.7.21 ndb-7.5.9)
id=5	@10.24.32.182  (mysql-5.7.21 ndb-7.5.9)

ndb_mgm> 

ndb_mgm是ndb_mgmd(MySQL Cluster Server)的客户端管理工具,通过它可以方便的检查Cluster的状态、启动备份、关闭Cluster等功能。 通过上面的状态信息,可以看到:

1)集群的管理服务端口是1186;
Connected to Management Server at: localhost:1186
2)集群的数据节点有2个;
[ndbd(NDB)]	2 node(s)
id=2	@10.24.32.183  (mysql-5.7.21 ndb-7.5.9, Nodegroup: 0, *)
id=3	@10.24.32.184  (mysql-5.7.21 ndb-7.5.9, Nodegroup: 1)
3)管理节点有一个;
[ndb_mgmd(MGM)]	1 node(s)
id=1	@10.24.32.180  (mysql-5.7.21 ndb-7.5.9)
4)SQL节点有2个,且处于连接状态;
[mysqld(API)]	2 node(s)
id=4	@10.24.32.181  (mysql-5.7.21 ndb-7.5.9)
id=5	@10.24.32.182  (mysql-5.7.21 ndb-7.5.9)

10、测试Cluster; 使用Cluster,则表的存储引擎必须为NDB,其他类型存储引擎的数据不会保存到数据节点中,对于CLuster的一个重要功能就是防止单点故障,下面将分别进行测试: 1)NDB存储引擎测试

--在任一SQL节点(本例用10.24.32.181)的test库创建测试表t_cluster,设置存储引擎为NDB:
mysql> create table t_cluster(id int,name varchar(100),cdate datetime default now()) engine=NDB;
Query OK, 0 rows affected (0.42 sec)

mysql> insert into t_cluster(id,name) values(1,'Alen'),(2,'Cluster Test');
Query OK, 2 rows affected (0.10 sec)
Records: 2  Duplicates: 0  Warnings: 0

mysql> select *from t_cluster;
+------+--------------+---------------------+
| id   | name         | cdate               |
+------+--------------+---------------------+
|    1 | Alen         | 2018-03-06 23:07:50 |
|    2 | Cluster Test | 2018-03-06 23:07:50 |
+------+--------------+---------------------+
2 rows in set (0.03 sec)

mysql> 
--在另一SQL节点(本例用10.24.32.182)的test库查询表t_cluster,结果如下:
mysql> select *from t_cluster;
+------+--------------+---------------------+
| id   | name         | cdate               |
+------+--------------+---------------------+
|    1 | Alen         | 2018-03-06 23:07:50 |
|    2 | Cluster Test | 2018-03-06 23:07:50 |
+------+--------------+---------------------+
2 rows in set (0.02 sec)

mysql> 
--如上,两个SQL节点查询到的数据是一致的。
--在SQL节点10.24.32.181将表t_cluster的存储引擎改为MyISAM,再次插入测试:
mysql> alter table t_cluster engine=myisam;
Query OK, 2 rows affected (0.66 sec)
Records: 2  Duplicates: 0  Warnings: 0

mysql> insert into t_cluster(id,name) values(3,'MyISAM Test');
Query OK, 1 row affected (0.03 sec)

mysql> select *from t_cluster;
+------+--------------+---------------------+
| id   | name         | cdate               |
+------+--------------+---------------------+
|    2 | Cluster Test | 2018-03-06 23:07:50 |
|    1 | Alen         | 2018-03-06 23:07:50 |
|    3 | MyISAM Test  | 2018-03-06 23:15:02 |
+------+--------------+---------------------+
3 rows in set (0.00 sec)

mysql> 
--在SQL节点10.24.32.182再次查询表t_cluster,可以发现,表无法查询,结果如下:
mysql> select *from t_cluster;
ERROR 1412 (HY000): Table definition has changed, please retry transaction
mysql> 
--在SQL节点10.24.32.181将表t_cluster的存储引擎改为NDB:
mysql> alter table t_cluster engine=ndb;
Query OK, 3 rows affected (0.32 sec)
Records: 3  Duplicates: 0  Warnings: 0

mysql> 
--在SQL节点10.24.32.182再次查询,结果如下:
mysql> select *from t_cluster;
+------+--------------+---------------------+
| id   | name         | cdate               |
+------+--------------+---------------------+
|    3 | MyISAM Test  | 2018-03-06 23:15:02 |
|    1 | Alen         | 2018-03-06 23:07:50 |
|    2 | Cluster Test | 2018-03-06 23:07:50 |
+------+--------------+---------------------+
3 rows in set (0.01 sec)

mysql> 
--显然,表t_cluster的数据再次同步到了数据节点,所有SQL节点又都可以正常查询数据。
2)单点故障测试 对于任一种节点,都存在单点故障的可能性,在Cluster的设置过程中,应该尽量对每一类节点设置冗余,以防止单点故障发生时造成的应用中断;对于管理节点,一般不需要特殊的设置,只需要将管理工具和配置文件放在多台主机上即可。 SQL节点发生单点故障
--将SQL节点10.24.32.181的MySQL服务停止:
[root@strong mysql]# mysqladmin -uroot -p shutdown 
Enter password: 
--查看CLuster的状态,可以发现SQL节点10.24.32.181已经断开,另一SQL节点10.24.32.182仍然正常连接:
[root@strong mysql-cluster]# ndb_mgm
-- NDB Cluster -- Management Client --
ndb_mgm> show
Connected to Management Server at: localhost:1186
Cluster Configuration
---------------------
[ndbd(NDB)]	2 node(s)
id=2	@10.24.32.183  (mysql-5.7.21 ndb-7.5.9, Nodegroup: 0, *)
id=3	@10.24.32.184  (mysql-5.7.21 ndb-7.5.9, Nodegroup: 1)

[ndb_mgmd(MGM)]	1 node(s)
id=1	@10.24.32.180  (mysql-5.7.21 ndb-7.5.9)

[mysqld(API)]	2 node(s)
id=4 (not connected, accepting connect from 10.24.32.181)
id=5	@10.24.32.182  (mysql-5.7.21 ndb-7.5.9)

ndb_mgm> 
--从节点10.24.32.182查看t_cluster表,结果如下,显然,SQL节点的单点故障并没有引起数据的查询故障,对于应用来说,需要改变的就是将以前对故障节点的访问改为对非故障节点的访问:
mysql> select *from t_cluster;
+------+--------------+---------------------+
| id   | name         | cdate               |
+------+--------------+---------------------+
|    2 | Cluster Test | 2018-03-06 23:07:50 |
|    3 | MyISAM Test  | 2018-03-06 23:15:02 |
|    1 | Alen         | 2018-03-06 23:07:50 |
+------+--------------+---------------------+
3 rows in set (0.00 sec)
数据节点的单点故障 在该测试环境中,数据节点有两个,那么它们对数据的存储是互相镜像还是一份数据分成几块呢?这个答案的关键在于配置文件中[ndbd default]组中的noofreplicas参数,如果该值为1,表示只有一份数据,但是分成n块分别存储在n个数据节点上;如果等于2,则表示数据被分成n/2块,每块数据都有两个备份,这样即使有任一个节点发生故障,只要它的备份节点正常,系统就可以正常运行。
--管理节点10.24.32.180的[ndbd default]组中的noofreplicas参数为1,将数据节点10.24.32.183的NDB进程停止:
[root@strong u01]# ps -ef |grep ndbd
root      8253     1  0 20:51 ?        00:00:01 ndbd
root      8254  8253  3 20:51 ?        00:03:36 ndbd
root      8515  8336  0 22:48 pts/1    00:00:00 grep --color=auto ndbd
[root@strong u01]# kill 8254 8253
--查看集群状态:
ndb_mgm> show
Cluster Configuration
---------------------
[ndbd(NDB)]	2 node(s)
id=2 (not connected, accepting connect from 10.24.32.183)
id=3 (not connected, accepting connect from 10.24.32.184)

[ndb_mgmd(MGM)]	1 node(s)
id=1	@10.24.32.180  (mysql-5.7.21 ndb-7.5.9)

[mysqld(API)]	2 node(s)
id=4 (not connected, accepting connect from 10.24.32.181)
id=5 (not connected, accepting connect from 10.24.32.182)
--在任一SQL节点(10.24.32.181)查询表t_cluster,结果如下:
mysql> select *from t_cluster;
ERROR 1296 (HY000): Got error 4009 'Cluster Failure' from NDBCLUSTER
mysql> 
--将noofreplicas参数改为2,按照前面的步骤重启集群:
NoOfReplicas=2
--停止数据节点10.24.32.183的NDB进程:
[root@strong ~]# ps aux |grep ndbd
root      1322  0.0  0.0  28584   128 ?        Ss   01:22   0:00 ndbd --initial
root      1323  1.8 56.6 1136400 281004 ?      Sl   01:22   0:04 ndbd --initial
root      1382  0.0  0.1 112632   964 pts/0    S+   01:26   0:00 grep --color=auto ndbd
[root@strong ~]# kill 1323
[root@strong ~]# ps aux |grep ndbd
root      1386  0.0  0.1 112632   964 pts/0    R+   01:26   0:00 grep --color=auto ndbd
[root@strong ~]# 
--查看集群状态:
ndb_mgm> show
Cluster Configuration
---------------------
[ndbd(NDB)]	2 node(s)
id=2 (not connected, accepting connect from 10.24.32.183)
id=3	@10.24.32.184  (mysql-5.7.21 ndb-7.5.9, Nodegroup: 0, *)

[ndb_mgmd(MGM)]	1 node(s)
id=1	@10.24.32.180  (mysql-5.7.21 ndb-7.5.9)

[mysqld(API)]	2 node(s)
id=4	@10.24.32.181  (mysql-5.7.21 ndb-7.5.9)
id=5	@10.24.32.182  (mysql-5.7.21 ndb-7.5.9)

ndb_mgm> 
--再次从任一SQL节点10.24.32.181查询,结果如下:
mysql> select *from t_cluster;
+------+--------------+---------------------+
| id   | name         | cdate               |
+------+--------------+---------------------+
|    1 | Alen         | 2018-03-07 01:40:04 |
|    3 | MyISAM Test  | 2018-03-07 01:40:19 |
|    2 | Cluster Test | 2018-03-07 01:40:04 |
+------+--------------+---------------------+
3 rows in set (0.01 sec)

11、Cluster的关闭;

[root@strong mysql-cluster]# ndb_mgm
-- NDB Cluster -- Management Client --
ndb_mgm> shutdown
Connected to Management Server at: localhost:1186
Node 2: Cluster shutdown initiated
Node 3: Cluster shutdown initiated
Node 3: Node shutdown completed.
3 NDB Cluster node(s) have shutdown.
Disconnecting to allow management server to shutdown.
Node 2: Node shutdown completed.
ndb_mgm> 

注意:集群关闭后,SQL节点不会关闭,需手工关闭;