corosync+pacemaker+主从模式drbd构建高可用mysql

---------------------------------DRBD安装配置及测试--------------------------------

一、准备工作

    1.更改node1、node2的hosts文件为:

 192.168.10.22 node1.wsh.com node1
 192.168.10.23 node2.wsh.com node2

    2.更改主机名,确保重启后主机名不变

node1, node2 # sed -i 's@\(HOSTNAME=\).*@\1node1.wsh.com@g' /etc/sysconfig/network
node1, node2 # sed -i 's@\(HOSTNAME=\).*@\1node2.wsh.com@g' /etc/sysconfig/network

    3.配置ssh双机互信

    4.同步时间,确保时间保持一致

二、安装drbd

    1.安装软件

node1, node2 # tar xf drbd-8.4.3.tar.gz
node1, node2 # cd drbd-8.4.3
node1, node2 # ./configure --prefix=/usr/local/drbd --sysconfdir=/etc --libdir=/usr/lib --with-km --with-pacemaker
node1, node2 # make && make install

    2.装载drbd模块

node1, node2 # modprobe drbd
node1, node2 # lsmod | grep drbd
       	       drbd                  325626  0
       	       libcrc32c               1246  1 drbd

        确保开机自动装载模块:

node1, node2 # echo "modprobe drbd" > /etc/sysconfig/modules/drbd.modules
node1, node2 # chmod 755 /etc/sysconfig/modules/drbd.modules

    3.将/usr/local/drbd/lib软连接至/usr/lib/drbd

node1, node2 # ln -sv /usr/local/drbd/lib/drbd /usr/lib/drbd

    4.将/usr/local/drbd/sbin加入PATH路径

node1, node2 # echo "export PATH=/usr/local/drbd/sbin:$PATH" > /etc/profile.d/drbd.sh

三、配置drbd

        node1上操作:

        1.更改global_common.conf配置文件,如下:

node1 # vim /etc/drbd.d/global_common.conf

            global {
                        usage-count no;
            }
            common {
                        handlers {
                                     pri-on-incon-degr 
                                     pri-lost-after-sb
                                     local-io-error
                         }
                         disk {
                                     on-io-error detach;
                         }
                          net {
                                     protocol C;
                                     cram-hmac-alg "sha1";
                                     shared-secret "e1680f714c4e6d71";
                          }
                          syncer {
                                     rate 1000M;
                          }
                }

        2.定义drbd的资源mystore.res,如下:  

node1 #vim /etc/drbd.d/mystore.res
                resource mystore {
                    on node1.wsh.com {
                             device /dev/drbd0;
                             disk /dev/sdb1;
                             address 192.168.10.22:7789;
                             meta-disk internal;
                    }
                    on node2.wsh.com {
                             device /dev/drbd0;
                             disk /dev/sdb1;
                             address 192.168.10.23:7789;
                             meta-disk internal;
                    }
                 }

       将global_common.conf和mystore.res复制到node2的/etc/drbd.d下:

# scp /etc/drbd.d/* node2:/etc/drbd.d/

四、初始化drbd资源并启动服务

        1.初始化资源,在node1、node2上分别执行:       

node1, node2 # drbdadm create-md mystore
              Writing meta data...
              initializing activity log
              NOT initializing bitmap
              New drbd meta data block successfully created.

         2.启动drbd服务,在node1、node2上执行:

node1, node2 #service drbd start

        3.查看启动状态          

 # drbd-overview
   0:mystore/0  Connected Secondary/Secondary Inconsistent/Inconsistent C r-----
# cat /proc/drbd
 version: 8.4.3 (api:1/proto:86-101)
 GIT-hash: 89a294209144b68adb3ee85a73221f964d3ee515 build by [email protected], 2015-08-01 14:04:51
  0: cs:Connected ro:Secondary/Secondary ds:Inconsistent/Inconsistent C r-----
   ns:0 nr:0 dw:0 dr:0 al:0 bm:0 lo:0 pe:0 ua:0 ap:0 ep:1 wo:f oos:4200796

         4. promote node1为Primary

            从上述状态来看,两个节点均为Secondary状态,下面我们提升node1节点为Primary,需要在设置为Primary节点上执行:           

  node1 # drbdadm primary --force mystore  第一次提升主节点,需要使用--force
  node1 # drbd-overview
          0:mystore/0  SyncSource Primary/Secondary UpToDate/Inconsistent C r---n-
          [===>................] sync'ed: 22.5% (3184/4100)M   此时正在同步数据

             稍等片刻,数据同步完成,在此查看状态:           

 node1 # drbd-overview
         0:mystore/0  Connected Primary/Secondary UpToDate/UpToDate C r-----

五、创建文件系统

        1.格式化drbd设备

        文件系统的挂载只能在Primary节点进行,因此,也只有在设置了主节点后才能对文件系统进行格式化:

node1 # mke2fs -t ext4 /dev/drbd0

        2.创建/mydata/data,并挂载/dev/drbd0       

node1, node2 # mkdir -v /mydata
       node1 # mount /dev/drbd0 /mydata

         3.测试       

node1 # cp /etc/inittab /mydata
node1 # # sed -i 's@\(id:\)3\(:initdefault:\)@\15\2@g' /mydata/inittab
node1 # drbd-overview
        0:mystore/0  Connected Primary/Secondary UpToDate/UpToDate C r----- /mydata ext4 4.0G 137M 3.7G 4%

六、切换Primary和Secondary节点

        对于Primary/Secondary模型的drbd来讲,在某个时刻只能有一个Primary节点,因此,要切换两个节点的角色,需要先将原有的Primary节点demote为Secondary后,才能将另一个节点promote为Primary:       

node1 demote:
        node1 # umount /mydata
       node1 # drbdadm secondary mystore
       node1 # drbd-overview
               0:mystore/0  Connected Secondary/Secondary UpToDate/UpToDate C r-----
 node2 promote:
        node2 # drbdadm primary mystore
       node2 # mount /dev/drbd0 /mydata
       node2 # tail -n1 /mydata/inittab
               id:5:initdefault:

         文件及文件内容都正确;   

#删除测试文件
node2 # rm -f /mydata/inittab
#创建data目录,作为mysql的数据目录
node2 # mkdir /mydata/data

-----------------------------------DRBD完成-----------------------------------------

---------------------------------MySQL安装及配置-----------------------------------


一、编译安装MySQL

    1.创建运行MySQL的用户和组  

node1, node2 # groupadd -r -g 306 mysql
node1, node2 # useradd -r -u 306 -g mysql mysql

     2.开始编译安装  

node1, node2 # yum -y install cmake
node1, node2 # tar xf mysql-5.6.26.tar.gz
node1, node2 # cd mysql-5.6.26
node1, node2 # cmake . -DCMAKE_INSTALL_PREFIX=/usr/local/mysql \
               > -DMYSQL_DATADIR=/mydata/data \
               > -DSYSCONFDIR=/etc \
               > -DWITH_INNOBASE_STORAGE_ENGINE=1 \
               > -DWITH_ARCHIVE_STORAGE_ENGINE=1 \
               > -DWITH_BLACKHOLE_STORAGE_ENGINE=1 \
               > -DWITH_READLINE=1 \
               > -DWITH_SSL=system \
               > -DWITH_ZLIB=system \
               > -DWITH_LIBWRAP=0 \
               > -DWITH_UNIX_ADDR=/tmp/mysql.sock \
               > -DDEFAULT_CHARSET=utf8 \
               > -DDEFAULT_COLLATION=utf8_general_ci
node1, node2 # make && make install

二、配置启动MySQL

    1.配置MySQL  

node1, node2 # cd /usr/local/mysql/
node1, node2 # chown -R mysql.mysql .
node1, node2 # cp support-files/my-default.cnf /etc/my.cnf
node1, node2 # sed -i '/^\[mysqld\]/a \datadir=\/mydata\/data' /etc/my.cnf
node1, node2 # sed -i '/^\[mysqld\]/a \socket=\/tmp\/mysql.sock' /etc/my.cnf
node1, node2 # echo -e "[mysql]\nsocket=/tmp/mysql.sock" >> /etc/my.cnf
node1, node2 # cp support-files/mysql.server /etc/init.d/mysqld
node1, node2 # chmod +x /etc/init.d/mysqld
node1, node2 # echo "export PATH=/usr/local/mysql/bin:$PATH" > /etc/profile.d/mysql.sh
node1, node2 #ln -sv /usr/local/mysql/include /usr/include/mysql
node1, node2 #echo '/usr/local/mysql/lib' /etc/ld.so.conf.d/mysql.conf

     2.初始化MySQL

    因为目前drbd的Primary在node2节点上,并且挂载至/mydata/data,所以在node2上初始化mysql: 

node2 # chown -R mysql.mysql /mydata/data
node2 # scripts/mysql_install_db --user=mysql --datadir=/mydata/data/

     在node2上尝试启动mysql:   

node2 # service mysqld start
        Starting MySQL....................         [  OK  ]

corosync+pacemaker+主从模式drbd构建高可用mysql_第1张图片

   用户登录授权:

mysql>GRANT ALL ON *.* TO 'root'@'%' IDENTIFIED BY 'redhat';

   3.切换node1为Primary,并启动mysql   

node2 # service mysqld stop
        Shutting down MySQL..           [  OK  ]
node2 # umount /mydata
node2 # drbdadm secondary mystore
node1 # drbdadm primary mystore
node1 # mount /dev/drbd0 /mydata/
node1 # service mysqld start
        Starting MySQL..........       [  OK  ]

corosync+pacemaker+主从模式drbd构建高可用mysql_第2张图片


三、停止服务并禁止开机自动启动

       为了保证drbd和mysqld的启动关闭等是由集群来控制的,所以需要手动设置:       

node1, node2 # chkconfig --add drbd
node1, node2 # chkconfig drbd off
node1, node2 # service drbd stop

node1, node2 # chkconfig --add mysqld
node1, node2 # chkconfig mysqld off
node1, node2 # service mysqld stop


------------------------------------MySQL完成--------------------------------------

-------------------------------------集群配置-----------------------------------------

一、安装corosync、pacemaker、crmsh

    1.安装corosync、pacemaker

node1, node2 # yum -y install corosync pacemaker

    2.安装crmsh

      只在管理端安装即可,这里两个节点都安装:

node1, node2 # yum install python-pssh-2.3.1-4.2.x86_64.rpm pssh-2.3.1-4.2.x86_64.rpm crmsh-2.1-1.6.x86_64.rpm

二、配置corosync

    1.编辑corosync配置文件     

node1 # cp /etc/corosync/corosync.conf.example /etc/corosync/corosync.conf
node1 # vim /etc/corosync/corosync.conf
                  compatibility: whitetank
                  totem {
                                version: 2
                                secauth: on
                                threads: 0
                                interface {
                                                    ringnumber: 0
                                                    bindnetaddr: 192.168.10.0
                                                    mcastaddr: 239.2.3.4
                                                    mcastport: 5405
                                                    ttl: 1
                                }
                    }
                    logging {
                                    fileline: off
                                    to_stderr: no
                                    to_logfile: yes
                                    logfile: /var/log/cluster/corosync.log
                                    to_syslog: no
                                    debug: off
                                    timestamp: on
                                    logger_subsys {
                                                                subsys: AMF
                                                                debug: off
                                    }
                    }

                    #这里将pacemaker作为corosync的插件启动(以后将会不支持)
                    service {
                                    ver:    0
                                    name:   pacemaker
                    }
                    aisexec {
                                    user:   root
                                    group:  root
                    }

     2.生成节点间通信时的认证秘钥文件

node1 # corosync-keygen

        自动生存authkey文件,权限为400

    3.将corosync.conf和authkey文件复制到node2的/etc/corosync下,并保持文件权限

node1 # scp -p /etc/corosync/{corosync.conf,authkey} node2:/etc/corosync

    4.在node1、node2上启动corosync

node1, node2 # service corosync start

    5.使用crmsh查看状态

corosync+pacemaker+主从模式drbd构建高可用mysql_第3张图片

三、配置属性

    1.关闭stonith设备

node1 # crm configure property stonith-enabled=false

    2.配置不满足法定票数时的动作为忽略

node1 # crm configure property no-quorum-policy=ignore

    3.配置资源默认黏性值为100

node1 # crm configure rsc_defaults resource-stickiness=100

四、定义资源

    1.定义主资源:drbd   

node1 # crm configure
crm(live)configure# primitive mysqlstore ocf:linbit:drbd params drbd_resource=mystore op monitor role=Master            interval=10s timeout=20s op monitor role=Slave interval=20s timeout=20s op start timeout=240s op stop timeout=100s on-fail=restart

     2.定义drbd为主从资源   

node1 # crm cnfigure
crm(live)configure# master ms_mysqlstore mysqlstore meta clone-max="2" clone-node-max="1" master-max="1" master-node-max="1" notify="true"

     3.定义主资源:VIP   

node1 # crm configure
crm(live)configure# primitive VIP ocf:heartbeat:IPaddr params ip=192.168.10.11 op monitor interval=10s timeout=20s op start timeout=20s op stop timeout=20s on-fail=restart

     4.定义主资源:Filesystem   

node1 # crm configure
crm(live)configure# primitive MYDATA ocf:heartbeat:Filesystem params device=/dev/drbd0 directory=/mydata fstype=ext4 op monitor interval=20s timeout=40s op start timeout=60s op stop timeout=60s on-fail=restart

     5.定义主资源:mysqld   

node1 # crm configure
crm(live)configure# primitive MYSQLD lsb:mysqld op monitor interval=30s timeout=60s op start timeout=60s op stop timeout=60s on-fail=restart

五、定于约束

    资源VIP、MYDATA、MYSQLD需要运行于drbd的Primary节点上,定义排列约束:

node1 # crm configure

    所有资源必须和ms_mysqlstore的主节点在一起(即drbd promote为Primary的节点)

crm(live)configure# collocation ALL_with_ms_mysqlstore_master inf: MYSQLD MYDATA VIP ms_mysqlstore:Master

    资源的启动顺序:

    ms_mysqlstore提升为Master(drbd promote为Primary)之后,MYDATA才能挂载,定义顺序约束:

crm(live)configure# order ms_mysqlstore_master_before_VIP_before_MYDATA_before_MYSQLD inf: ms_mysqlstore:promote VIP:start MYDATA:start MYSQLD:start

    校验:

crm(live)configure# verify

    校验没有问题,提交:

 crm(live)configure# commit

六、状态查看、角色切换及测试

corosync+pacemaker+主从模式drbd构建高可用mysql_第4张图片

corosync+pacemaker+主从模式drbd构建高可用mysql_第5张图片

将node1.wsh.com切换为standby:

node1 # crm node standby node1.wsh.com

corosync+pacemaker+主从模式drbd构建高可用mysql_第6张图片

corosync+pacemaker+主从模式drbd构建高可用mysql_第7张图片

------------------------------------集群配置完成--------------------------------------

至此,全部配置已完成。

文章仅作为学习笔记,如果疏漏,欢迎指正,谢谢!

你可能感兴趣的:(corosync+pacemaker+主从模式drbd构建高可用mysql)