Corosync+Pacemaker+DRBD+Mysql高可用HA配置

操作系统: CentOS 6.6 x64,本文采用rpm方式安装corosync+pacemaker+drbd,采用二进制版本安装mysql-5.6.29。本文是在Corosync+Pacemaker+DRBD+NFS高可用实例配置基础上进行配置修改，然后进行测试的安装过程。

一、双机配置

1. app1,app2配置hosts文件，以及主机名。

[root@app1 soft]# vi /etc/hosts  
127.0.0.1   localhost localhost.localdomain localhost4 localhost4.localdomain4   
::1         localhost localhost.localdomain localhost6 localhost6.localdomain6   
192.168.0.24         app1   
192.168.0.25         app2   
10.10.10.24          app1-priv   
10.10.10.25          app2-priv

说明：10段是心跳IP, 192.168段是业务IP, 采用VIP地址是192.168.0.26。

 

2. 关闭selinux与防火墙

sed -i '/SELINUX/s/enforcing/disabled/' /etc/selinux/config
setenforce 0
chkconfig iptables off
service iptables stop

 

3. 配置各节点ssh互信，好像可配\可不配,方便管理。

app1:
[root@app1 ~]# ssh-keygen  -t rsa -f ~/.ssh/id_rsa  -P '' 
[root@app1 ~]# ssh-copy-id -i .ssh/id_rsa.pub root@app2

app2:
[root@app2 ~]# ssh-keygen  -t rsa -f ~/.ssh/id_rsa  -P ''
[root@app2 ~]# ssh-copy-id -i .ssh/id_rsa.pub root@app1

 

二、DRDB安装配置

1. app1,app2配置hosts文件以及准备磁盘分区

app1: /dev/sdb1  ―> app2: /dev/sdb1

 

2. app1,app2安装drbd并安装

(1) 下载drbd安装包, CentOS6.6采用kmod-drbd84-8.4.5-504.1安装包才可用。

http://rpm.pbone.net/

drbd84-utils-8.9.1-1.el6.elrepo.x86_64.rpm
kmod-drbd84-8.4.5-504.1.el6.x86_64.rpm

# rpm -ivh drbd84-utils-8.9.5-1.el6.elrepo.x86_64.rpm kmod-drbd84-8.4.5-504.1.el6.x86_64.rpm
Preparing...                ########################################### [100%]
   1:drbd84-utils           ########################################### [ 50%]
   2:kmod-drbd84            ########################################### [100%]
Working. This may take some time ...
Done.
#

 

(2) 加载DRBD到内核模块

app1,app2分别操作，并加入到/etc/rc.local文件中。
modprobe drbd
lsmode |grep drbd

 

3. 创建修改配置文件。节点1，节点2一样配置。

[root@app1 ~]# vi /etc/drbd.d/global_common.conf
global {
        usage-count no;
}
common {
        protocol C;
        disk {
                on-io-error detach;
                no-disk-flushes;
                no-md-flushes; 
        }
        net {
                sndbuf-size 512k;
                max-buffers     8000;
                unplug-watermark   1024;
                max-epoch-size  8000;
                cram-hmac-alg "sha1";
                shared-secret "hdhwXes23sYEhart8t";
                after-sb-0pri disconnect;
                after-sb-1pri disconnect;
                after-sb-2pri disconnect;
                rr-conflict disconnect;
        }
        syncer {
                rate 300M;
                al-extents 517;
        }
}

resource data {
      on app1 {
               device    /dev/drbd0;
               disk      /dev/sdb1;
               address   10.10.10.24:7788;
               meta-disk internal;
      }
      on app2 {
               device     /dev/drbd0;
               disk       /dev/sdb1;
               address    10.10.10.25:7788;
               meta-disk internal;
      }
}

 

4. 初始化资源

在app1和app2上分别执行：

# drbdadm create-md data

initializing activity log
NOT initializing bitmap
Writing meta data...
New drbd meta data block successfully created.

 

5. 启动服务

在app1和app2上分别执行：或采用 drbdadm up data

# service drbd start

Starting DRBD resources: [
     create res: data
   prepare disk: data
    adjust disk: data
     adjust net: data
]
..........
#

 

6. 查看启动状态, 两节点应均处于Secondary状态。

cat /proc/drbd       #或者直接使用命令drbd-overview

节点1:
[root@app1 drbd.d]# cat /proc/drbd 
version: 8.4.5 (api:1/proto:86-101)
GIT-hash: 1d360bde0e095d495786eaeb2a1ac76888e4db96 build by [email protected], 2015-01-02 12:06:20
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:d oos:20964116

节点2:
[root@app2 drbd.d]# cat /proc/drbd 
version: 8.4.5 (api:1/proto:86-101)
GIT-hash: 1d360bde0e095d495786eaeb2a1ac76888e4db96 build by [email protected], 2015-01-02 12:06:20
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:d oos:20964116

 

7. 将其中一个节点配置为主节点

我们需要将其中一个节点设置为Primary，在要设置为Primary的节点上执行如下两条命令均可：
drbdadm -- --overwrite-data-of-peer primary data  
drbdadm primary --force data

主节点查看同步状态：
[root@app1 drbd.d]# cat /proc/drbd 
version: 8.4.5 (api:1/proto:86-101)
GIT-hash: 1d360bde0e095d495786eaeb2a1ac76888e4db96 build by [email protected], 2015-01-02 12:06:20
0: cs:SyncSource ro:Primary/Secondary ds:UpToDate/Inconsistent C r-----
    ns:1229428 nr:0 dw:0 dr:1230100 al:0 bm:0 lo:0 pe:2 ua:0 ap:0 ep:1 wo:d oos:19735828
        [>...................] sync'ed:  5.9% (19272/20472)M
        finish: 0:27:58 speed: 11,744 (11,808) K/sec
[root@app1 drbd.d]#

 

8. 创建文件系统

文件系统的挂载只能在Primary节点进行，只有在设置了主节点后才能对drbd设备进行格式化, 格式化与手动挂载测试。

[root@app1 ~]# mkfs.ext4 /dev/drbd0
[root@app1 ~]# mount /dev/drbd0 /data

 

三、安装配置Mysql-5.6.x

1. app1\app2下载编译版本mysql安装

wget http://mirrors.sohu.com/mysql/MySQL-5.6/mysql-5.6.29-linux-glibc2.5-x86_64.tar.gz
tar zxvf mysql-5.6.29-linux-glibc2.5-x86_64.tar.gz  -C /usr/local
cd /usr/local/
ln -sv mysql-5.6.29-linux-glibc2.5-x86_64 mysql
groupadd mysql
useradd -g mysql -M -s /sbin/nologin mysql
chown -R mysql:mysql /usr/local/mysql

 

2. app1下初始化数据库（初始化目录为drbd0同步目录中）

/usr/local/mysql/scripts/mysql_install_db --user=mysql --basedir=/usr/local/mysql --datadir=/data/mysql3306

 

3, app1,app2下创建配置文件及服务

cd /usr/local/mysql
cp support-files/my-default.cnf /etc/my.cnf
cp support-files/mysql.server  /etc/rc.d/init.d/mysqld
chkconfig --add mysqld

 

4. app1,app2配置Mysql命令链接，也可以采用加入环境变量中，该方式可以略过。

ln -sf /usr/local/mysql/bin/mysql /usr/bin/mysql
ln -sf /usr/local/mysql/bin/mysqldump /usr/bin/mysqldump
ln -sf /usr/local/mysql/bin/myisamchk /usr/bin/myisamchk
ln -sf /usr/local/mysql/bin/mysqld_safe /usr/bin/mysqld_safe

或通过加入环境变量中解决。

# vi /etc/profile
export PATH=/usr/local/mysql/bin/:$PATH
# source /etc/profile

ln -sv /usr/local/mysql/include  /usr/include/mysql
echo '/usr/local/mysql/lib' > /etc/ld.so.conf.d/mysql.conf
ldconfig

 

5. app1上Mysql配置文件（两边保持配置文件一致）

vi /etc/my.cnf

[client]
port        = 3306
default-character-set  = utf8
socket      = /tmp/mysql.sock
[mysqld]
character-set-server   = utf8
collation-server       = utf8_general_ci
port                   = 3306
socket                 = /tmp/mysql.sock
basedir                = /usr/local/mysql
datadir                = /data/mysql3306
skip-external-locking
key_buffer_size        = 16M
max_allowed_packet     = 1M
table_open_cache       = 64
sort_buffer_size       = 512K
net_buffer_length      = 8K
read_buffer_size       = 256K
read_rnd_buffer_size    = 512K
myisam_sort_buffer_size = 8M
log-bin                 = mysql-bin
binlog_format           = mixed
server-id               = 1
[mysqldump]
quick
max_allowed_packet = 16M
[mysql]
no-auto-rehash

[myisamchk]
key_buffer_size = 20M
sort_buffer_size = 20M
read_buffer = 2M
write_buffer = 2M

[mysqlhotcopy]
interactive-timeout

 

6. 启动mysql,不要配置开机自启动。

service mysqld start

 

7. 修改管理员密码并测试

# /usr/local/mysql/bin/mysqladmin -u root password 'admin' #设置管理员密码
# /usr/local/mysql/bin/mysql -u root -p   #测试密码输入

 

8. 复制配置文件到app2

# scp /etc/my.cnf app2:/etc/

 

9. app1关闭mysql并设置开机不启动

[root@node1 ~]# service mysqld stop
[root@node1 data]# chkconfig mysqld off

 

10.将node2节点上的DRBD设置为主节点并挂载

(1) app1卸载/dev/drbd0

# umount /data/
# drbdadm secondary data 
# drbd-overview  
  0:web/0  Connected Secondary/Secondary UpToDate/UpToDate C r-----

(2) app2配置drbd为主后，测试mysql的启动。

# drbdadm primary data
# drbd-overview  
  0:web/0  Connected Primary/Secondary UpToDate/UpToDate C r-----  

# mkdir /data 
# mount /dev/drbd0 /data/ 
# service mysqld start

 

四、corosync+pacemaker

1. app1,app2配置安装corosync pacemaker 

# yum install corosync pacemaker -y

2. app1,app2安装crmsh

RHEL自6.4起不再提供集群的命令行配置工具crmsh，要实现对集群资源管理，还需要独立安装crmsh。
crmsh的rpm安装可从如下地址下载：http://download.opensuse.org/repositories/network:/ha-clustering:/Stable/CentOS_CentOS-6/

[root@app1 crm]# yum install python-dateutil -y  
说明：python-pssh、pssh依懒于python-dateutil包

[root@app1 crm]# rpm -ivh pssh-2.3.1-4.2.x86_64.rpm python-pssh-2.3.1-4.2.x86_64.rpm crmsh-2.1-1.6.x86_64.rpm
warning: pssh-2.3.1-4.2.x86_64.rpm: Header V3 RSA/SHA1 Signature, key ID 17280ddf: NOKEY
Preparing...                ########################################### [100%]
   1:python-pssh            ########################################### [ 33%]
   2:pssh                   ########################################### [ 67%]
   3:crmsh                  ########################################### [100%]
[root@app1 crm]#
[root@app1 crm]#

 

3. 创建corosync配置文件，app1,app2一样。

cd /etc/corosync/
cp corosync.conf.example corosync.conf

vi /etc/corosync/corosync.conf
# Please read the corosync.conf.5 manual page
compatibility: whitetank
totem {   
        version: 2
        secauth: on
        threads: 0
        interface {
                ringnumber: 0
                bindnetaddr: 10.10.10.0
                mcastaddr: 226.94.8.8
                mcastport: 5405
                ttl: 1
        }
}

logging {
        fileline: off
        to_stderr: no
        to_logfile: yes
        to_syslog: no
        logfile: /var/log/cluster/corosync.log
        debug: off
        timestamp: on
        logger_subsys {
                subsys: AMF
                debug: off
        }
}

amf {
        mode: disabled
}

service {
        ver:  1                  
        name: pacemaker       
}
aisexec {
        user: root
        group:  root
}

 

4. 创建认证文件，app1,app2一样

各节点之间通信需要安全认证，需要安全密钥，生成后会自动保存至当前目录下，命名为authkey,权限为400。

[root@app1 corosync]# corosync-keygen
Corosync Cluster Engine Authentication key generator.
Gathering 1024 bits for key from /dev/random.
Press keys on your keyboard to generate entropy.
Press keys on your keyboard to generate entropy (bits = 128).
Press keys on your keyboard to generate entropy (bits = 192).
Press keys on your keyboard to generate entropy (bits = 256).
Press keys on your keyboard to generate entropy (bits = 320).
Press keys on your keyboard to generate entropy (bits = 384).
Press keys on your keyboard to generate entropy (bits = 448).
Press keys on your keyboard to generate entropy (bits = 512).
Press keys on your keyboard to generate entropy (bits = 576).
Press keys on your keyboard to generate entropy (bits = 640).
Press keys on your keyboard to generate entropy (bits = 704).
Press keys on your keyboard to generate entropy (bits = 768).
Press keys on your keyboard to generate entropy (bits = 832).
Press keys on your keyboard to generate entropy (bits = 896).
Press keys on your keyboard to generate entropy (bits = 960).
Writing corosync key to /etc/corosync/authkey.
[root@app1 corosync]#

 

5. 将刚才配置的三个文件同步至app2，同步过去后要修改ha.cf文件中的心跳IP

# scp authkeys corosync.conf  root@app2:/etc/corosync/  

 

6. 启动corosync\pacemaker服务，测试能否正常提供服务

节点1:  
[root@app1 ~]# service corosync start   
Starting Corosync Cluster Engine (corosync):               [OK]

[root@app1 ~]# service pacemaker start
Starting Pacemaker Cluster Manager                         [OK]

配置服务开机自启动：
chkconfig corosync on
chkconfig pacemaker on

节点2:  
[root@app2 ~]# service corosync start   
Starting Corosync Cluster Engine (corosync):               [OK]

[root@app1 ~]# service pacemaker start
Starting Pacemaker Cluster Manager                         [OK]

配置服务开机自启动：
chkconfig corosync on
chkconfig pacemaker on

 

7. 测试corosync,pacemaker,crmsh安装情况

(1) 查看节点情况

[root@app1 ~]# crm status
Last updated: Tue Jan 26 13:13:19 2016
Last change: Mon Jan 25 17:46:04 2016 via cibadmin on app1
Stack: classic openais (with plugin)
Current DC: app1 - partition with quorum
Version: 1.1.10-14.el6-368c726
2 Nodes configured, 2 expected votes
0 Resources configured

Online: [ app1 app2 ]

 

(2) 查看端口启动情况

# netstat -tunlp
Active Internet connections (only servers)
Proto Recv-Q Send-Q Local Address               Foreign Address             State       PID/Program name  
udp        0      0 10.10.10.25:5404            0.0.0.0:*                               2828/corosync      
udp        0      0 10.10.10.25:5405            0.0.0.0:*                               2828/corosync      
udp        0      0 226.94.8.8:5405             0.0.0.0:*                               2828/corosync      

 

(3) 查看日志

[root@app1 corosync]# tail -f  /var/log/cluster/corosync.log

可以查看日志中关键信息：
Jan 23 16:09:30 corosync [MAIN  ] Corosync Cluster Engine ('1.4.7'): started and ready to provide service.
Jan 23 16:09:30 corosync [MAIN  ] Successfully read main configuration file '/etc/corosync/corosync.conf'.
....
Jan 23 16:09:30 corosync [TOTEM ] Initializing transmit/receive security: libtomcrypt SOBER128/SHA1HMAC (mode 0).
Jan 23 16:09:31 corosync [TOTEM ] The network interface [10.10.10.24] is now up.
Jan 23 16:09:31 corosync [TOTEM ] A processor joined or left the membership and a new membership was formed.
Jan 23 16:09:48 corosync [TOTEM ] A processor joined or left the membership and a new membership was formed.
[root@app1 corosync]#

 

五、配置pacemaker

1. 基本配置

corosync默认启用了stonith功能，而我们要配置的集群并没有stonith设备，因此在配置集群的全局属性时要对其禁用。

# crm
crm(live)# configure                                      ##进入配置模式
crm(live)configure# property stonith-enabled=false        ##禁用stonith设备
crm(live)configure# property no-quorum-policy=ignore      ##不具备法定票数时采取的动作
crm(live)configure# rsc_defaults resource-stickiness=100  ##设置默认的资源黏性，只对当前节点有效。
crm(live)configure# verify                                ##校验
crm(live)configure# commit                                ##校验没有错误再提交
crm(live)configure# show                                  ##查看当前配置
node app1
node app2
property cib-bootstrap-options: \
        dc-version=1.1.11-97629de \
        cluster-infrastructure="classic openais (with plugin)" \
        expected-quorum-votes=2 \
        stonith-enabled=false \
        default-resource-stickiness=100 \
        no-quorum-policy=ignore

 

2. 资源配置

#命令使用经验说明：verify报错的，可以直接退出，也可以采用edit编辑，修改正确为止。
# crm configure edit  可以直接编辑配置文件

 

(1) 添加VIP

不要单个资源提交，等所有资源及约束一起建立之后提交。
crm(live)configure# primitive vip ocf:heartbeat:IPaddr params ip=192.168.0.26 cidr_netmask=24 nic=eth0:1 op monitor interval=30s timeout=20s on-fail=restart
crm(live)configure# verify 

 

(2) 添加drdb服务

crm(live)configure# primitive mydrbd ocf:linbit:drbd params drbd_resource=data op monitor role=Master interval=20 timeout=30 op monitor role=Slave interval=30 timeout=30 op start timeout=240 op stop timeout=100
crm(live)configure# verify

把drbd设为主从资源：
crm(live)configure# ms ms_mydrbd mydrbd meta master-max=1 master-node-max=1 clone-max=2  clone-node-max=1 notify=true
crm(live)configure# verify

 

(3) 文件系统挂载服务：

crm(live)configure# primitive mystore ocf:heartbeat:Filesystem params device=/dev/drbd0 directory=/data fstype=ext4 op start timeout=60s op stop timeout=60s op monitor interval=30s timeout=40s on-fail=restart
crm(live)configure# verify

 

(4) 创建约束,很关键，VIP，DRBD, 目录挂载均在一台节点上，而且VIP，目录挂载均依懒于主DRBD.

创建组资源，vip与mystore一起。                   
crm(live)configure# group g_service vip mystore
crm(live)configure# verify

创建位置约束，组资源的启动依懒于drbd主节点
crm(live)configure# colocation c_g_service inf: g_service ms_mydrbd:Master

创建位置约整，mystore存储挂载依赖于drbd主节点

crm(live)configure# colocation mystore_with_drbd_master inf: mystore ms_mydrbd:Master

启动顺序依懒，drbd启动后，创建g_service组资源

crm(live)configure# order o_g_service inf: ms_mydrbd:promote g_service:start
crm(live)configure# verify
crm(live)configure# commit

 

(5) 增加mysql资源

crm(live)# configure  
crm(live)configure# primitive mysqld lsb:mysqld  op monitor interval=20 timeout=20 on-fail=restart
 
创建mysql服务与g_service组在一起
crm(live)configure# colocation mysqld_with_g_service inf: mysqld g_service  
crm(live)configure# verify   
crm(live)configure# show   

创建启动顺序，mysql服务在g_service组启动之后再启动
crm(live)configure# order mysqld_after_g_service mandatory: g_service mysqld 
crm(live)configure# verify   
crm(live)configure# show   
crm(live)configure# commit

 

3. 配置完成后，查看状态

[root@app1 ~]# crm status
Last updated: Fri Apr 29 14:59:14 2016
Last change: Fri Apr 29 14:59:05 2016 via cibadmin on app1
Stack: classic openais (with plugin)
Current DC: app1 - partition with quorum
Version: 1.1.10-14.el6-368c726
2 Nodes configured, 2 expected votes
5 Resources configured

Online: [ app1 app2 ]

Master/Slave Set: ms_mydrbd [mydrbd]
     Masters: [ app1 ]
     Slaves: [ app2 ]
mysqld (lsb:mysqld):   Started app1
Resource Group: g_service
     vip        (ocf::heartbeat:IPaddr):        Started app1
     mystore    (ocf::heartbeat:Filesystem):    Started app1

[root@app1 ~]#

 

4. 模拟故障切换

(1) app1上操作standby

[root@app1 mysql]# crm node standby app1

(2) app1再查看切换状态：状态转移都很成功。

[root@app1 ~]# crm status
Last updated: Fri Apr 29 15:12:01 2016
Last change: Fri Apr 29 15:01:49 2016 via crm_attribute on app1
Stack: classic openais (with plugin)
Current DC: app1 - partition with quorum
Version: 1.1.10-14.el6-368c726
2 Nodes configured, 2 expected votes
5 Resources configured

Node app1: standby
Online: [ app2 ]

Master/Slave Set: ms_mydrbd [mydrbd]
     Masters: [ app2 ]
     Stopped: [ app1 ]
mysqld (lsb:mysqld):   Started app2
Resource Group: g_service
     vip        (ocf::heartbeat:IPaddr):        Started app2
     mystore    (ocf::heartbeat:Filesystem):    Started app2
[root@app1 ~]#

 

(3) app2上就可以测试mysql登录了：

[root@app2 ~]# mysql -uroot -padmin
Warning: Using a password on the command line interface can be insecure.
Welcome to the MySQL monitor.  Commands end with ; or \g.
Your MySQL connection id is 1
Server version: 5.6.29-log MySQL Community Server (GPL)

Copyright (c) 2000, 2016, 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> \q
Bye

 

(4) app2上查看drbd挂载目录情况

[root@app2 ~]# df -h
Filesystem                   Size  Used Avail Use% Mounted on
/dev/mapper/vg_app2-lv_root   36G  5.0G   29G  16% /
tmpfs                       1004M   29M  976M   3% /dev/shm
/dev/sda1                    485M   39M  421M   9% /boot
/dev/drbd0                   5.0G  249M  4.5G   6% /data
[root@app2 ~]#
[root@app2 ~]#

#说明：切换测试时有时会出现警告提示，影响真实状态查看，可以采用如下方式清除，提示哪个资源报警就清哪个，清理后，再次crm status查看状态显示正常。
Failed actions:
mystore_stop_0 on app1 'unknown error' (1): call=97, status=complete, last-rc-change='Tue Jan 26 14:39:21 2016', queued=6390ms, exec=0ms

[root@app1 ~]# crm resource cleanup mystore
Cleaning up mystore on app1
Cleaning up mystore on app2
Waiting for 2 replies from the CRMd.. OK
[root@app1 ~]#

 

5. 配置小结

在切换的过程中最大的问题就是DRBD的同步问题，必竟数据都在磁盘上，如果不同步就会造成数据不一致的问题，standby模拟切换其实不能真实模拟drbd的故障转移的。因为在故障转移之后，drbd被stop之后，从库接管主节点会从因stop之后会出现unknownn状态，这时候需要做会数据初始化同步。