重起两节点或heartbeat服务,通过crm_mon –i2 命令查看资源状态..
此时,两个节点都充当主节点,占用共享IP和磁盘资源,同时做为对方的备用节点.这是为以下在每个节点同时安装两个MySQL做准备.其目的是同一时刻每个节点都运行一个MySQL数据库,如果其中一个节点出现问题,另一个节点则会接管对方的MySQL服务,即此时,正常的这个节点会同时运行两个MySQL服务.提高了服务器的使用效率.
@测试内容2配置,在每个节点上做以下修改.
分别关闭两个节点,然后增加一网卡eth1,用于心跳监测.
分别关闭两个节点,然后增加一网卡eth1,用于心跳监测.
具体实现:
修改/etc/ha.d/,原来是eth0,然后重新生成cib.xml(先把原来的删除)
|
bcast eth1
|
|
# /usr/lib/heartbeat/haresources2cib.py --stout -c /etc/ha.d/ /etc/ha.d/haresources.bak
|
验证是否生效,重起heartbeat,如图,bcast使用eth1通信.
@FAQ
1. WARN: log_data_element: default_cib_update_callback: update:failed
问题:来源于/var/log/message,同时,机器不断重启.
原因: /etc//haresources.bak配置有问题,我当时没有在haresources里定义虚拟IP
|
db_cluster1 IPaddr:: Filesystem::192.168.0.34:/mnt/da0/db1::/var/mysql/db1::nfs
db_cluster1 IPaddr:: Filesystem::192.168.0.34:/mnt/da0/db2::/var/mysql/db2::nfs
|
改为以下即可.
|
db_cluster1IPaddr::192.168.0.32 Filesystem::192.168.0.34:/mnt/da0/db1::/var/mysql/db1::nfs
db_cluster1IPaddr::192.168.0.33 Filesystem::192.168.0.34:/mnt/da0/db2::/var/mysql/db2::nfs
|
2. Resource IPaddr_192_168_0_32 cannot run anywhere
WARN: native_color: Resource Filesystem_2 cannot run anywhere
WARN: native_color: Resource IPaddr_192_168_0_33 cannot run anywhere
WARN: native_color: Resource Filesystem_12 cannot run anywhere
以下是原配置
|
db_cluster1IPaddr::192.168.0.32/24/eth0 Filesystem::192.168.0.34:/mnt/da0/db1::/var/mysql/db1::nfs
db_cluster1IPaddr::192.168.0.33/24/eth0 Filesystem::192.168.0.34:/mnt/da0/db1::/var/mysql/db1::nfs
|
解决:
@资源手工管理 1.查看所有资源(在任何一个节点执行都可以)
# crm_resource –L
2. 资源切换
# crm_resource -M -r group_11 -H db_cluster1
-M 表示迁移 –r 资源名,这里是指资源组名 –H 节点名
-M 表示迁移 –r 资源名,这里是指资源组名 –H 节点名
以上是切换前资源状态
以上是切换后资源状态
四．MySQL多实例在两个节点上的安装
安装前资源状态：#crm_mon –i2
具体安装位置说明：
|
|
db_cluster1(节点1)
|
db_cluster2(节点2)
|
|
mysqld_1
|
basedir: /usr/local/mysql
datadir: /var/mysql/db1
|
basedir: /usr/local/mysql
datadir: /var/mysql/db1
|
|
Mysqld_2
|
basedir: /usr/local/mysql
datadir: /var/mysql/db2
|
basedir: /usr/local/mysql
datadir: /var/mysql/db2
|
1．Db_cluster1上安装MySQL
@版本：mysql-5.1.35.tar.gz(官网下载)
刚开始采用的版本是5.0.41,但其有一个bug,没有mysqld_multi.server.sh文件，不易在HA上管理mysql多个实例。故选择了5.1以上版本。
@安装
# tar zxvf mysql-5.1.35.tar.gz
# cd mysql-5.1.35
#./configure --prefix=/usr/local/mysql
--with-extra-charsets=gb2312,gbk --with-unix-socket-path=/tmp/mysql.sock
--with-mysqld-ldflags=-all-static --with-client-ldflags=-all-static
--sysconfdir=/etc --enable-assembler
# ./ scripts/mysql_install_db --user=mysql #安装数据库，安装位置为/usr/local/mysql/var
基本配置：
|
# cp –r –p /usr/local/mysql/var /var/mysql/db1/
# cd /usr/local/mysql
# cp share/mysql/my-medium.cnf /etc/my.cnf
# more /etc/my.cnf
-----------------------------my.cnf开始---------------------------------------------- [client]
port = 3306
socket = /tmp/mysql.sock
[mysqld_multi]
mysqld = /usr/local/mysql/bin/mysqld_safe
mysqladmin = /usr/local/mysql/bin/mysqladmin
user = root
[mysqld1]
port = 3306
socket = /tmp/mysql.sock
pid-file = /var/mysql/db1/var/db_cluster1.pid
datadir = /var/mysql/db1/var
skip-locking
key_buffer = 16M
max_allowed_packet = 1M
table_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
server-id = 1
[mysqld2]
port = 3307
socket = /tmp/mysql2.sock
pid-file = /var/mysql/db2/var/db_cluster1.pid
datadir = /var/mysql/db2/var
skip-locking
key_buffer = 16M
max_allowed_packet = 1M
table_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
server-id = 1
[mysqldump]
quick
max_allowed_packet = 16M
[mysql]
no-auto-rehash
[isamchk]
key_buffer = 20M
sort_buffer_size = 20M
read_buffer = 2M
write_buffer = 2M
[myisamchk]
key_buffer = 20M
sort_buffer_size = 20M
read_buffer = 2M
write_buffer = 2M
[mysqlhotcopy]
interactive-timeout
-------------------------------------my.cnf结束------------------------------------------------------------
# export PATH=$PATH:/usr/local/mysql/bin:. (可以将其放置于/etc/profile文件)
|
起动测试： # mysqld_multi start 1-2 #两个实例同时起动
# mysqld_multi start 1 # 起动实例1
从上图可以看出，两个MySQL实例都正常起动了。
停止服务：# mysqld_multi stop 1-2 # 两个实例同时停止
# mysqld_multi stop 1 # 停止实例1
以服务方式管理：
# cp 源代码/support-files/mysqld_multi.server /etc/rc.d/init.d/mysqld_m
将export PATH=$PATH:/usr/local/mysql/bin添加到/etc/rc.d/init.d/mysqld_m文件中
# chmod +x /etc/rc.d/init.d/mysqld_m
操作命令：# service mysqld_m start|stop|report|restart
@整合mysqld_m系统服务到heartbeat资源管理中.
修改以下内容:
|
# more /etc/ha.d/haresources.bak
db_cluster1 IPaddr::192.168.0.32 Filesystem::192.168.0.34:/mnt/da0/db1::/var/mysql/db1::nfs mysqld_m::1
db_cluster2 IPaddr::192.168.0.33 Filesystem::192.168.0.34:/mnt/da0/db2::/var/mysql/db2::nfs mysqld_m::2
# /usr/lib/heartbeat/haresources2cib.py --stout -c /etc/ha.d/ /etc/ha.d/haresources.bak
重新生成cib.xml
|
2．Db_cluster2上安装MySQL(实例:mysqld1和mysqld2)
@安装
# tar zxvf mysql-5.0.41.tar.gz
# cd mysql-5.0.41
#./configure --prefix=/usr/local/mysql
--with-extra-charsets=gb2312,gbk --with-unix-socket-path=/tmp/mysql.sock
--with-mysqld-ldflags=-all-static --with-client-ldflags=-all-static
--sysconfdir=/etc --enable-assembler
# make && make install
# ./ scripts/mysql_install_db --user=mysql #安装数据库，安装位置为/usr/local/mysql/var
基本配置：
|
# cp –r –p /usr/local/mysql/var /var/mysql/db2/
# cd /usr/local/mysql
# cp share/mysql/my-medium.cnf /etc/my.cnf
# more /etc/my.cnf
-----------------------------my.cnf开始---------------------------------------------- # 与db_cluster1机器上的配置是一样的,从上边拷贝一下即可 # 可以这样: # scp 192.168.0.30:/etc/my.cnf /etc/ -------------------------------------my.cnf结束---------------------------------------
# export PATH=$PATH:/usr/local/mysql/bin:. (可以将其放置于/etc/profile文件)
|
以服务方式管理：
# cp 源代码/support-files/mysqld_multi.server /etc/rc.d/init.d/mysqld_m
# chmod +x /etc/rc.d/init.d/mysqld_m
操作命令：# service mysqld_m start|stop|report|restart
@整合mysqld_m系统服务到heartbeat资源管理中.
修改以下内容:
|
# more /etc/ha.d/haresources.bak
db_cluster1 IPaddr::192.168.0.32 Filesystem::192.168.0.34:/mnt/da0/db1::/var/mysql/db1::nfs mysqld_m::1
db_cluster2 IPaddr::192.168.0.33 Filesystem::192.168.0.34:/mnt/da0/db2::/var/mysql/db2::nfs mysqld_m::2
# /usr/lib/heartbeat/haresources2cib.py --stout -c /etc/ha.d/ /etc/ha.d/haresources.bak
重新生成cib.xml,建议先把原来的删除(/var/lib/heartbeat/crm/)
|
3.FAQ
@WARNING: my_print_defaults command not found.
Please make sure you have this command available and
in your path. The command is available from the latest
MySQL distribution.
ABORT: Can't find command 'my_print_defaults'.
This command is available from the latest MySQL
distribution. Please make sure you have the command
in your PATH.
问题出现：/usr/local/mysql/bin/mysqld_multi start 1-2
解决：export PATH=$PATH:/usr/local/mysql/bin:.
如果添加到/etc/profile文件中不能解决，可以将其加入到/etc/rc.d/init.d/mysqld_m文件中
五．配置到此完毕，做最后测试
@测试一：起动两个节点，同时或不同时都可。
资源均起动正常。（包括VIP，NFS挂载和数据库服务）
@测试二,停止其中一个节点，稍等片刻再看资源状态，这里关闭第一个节点(db_cluster1)
Db_cluster1显示已OFFLINE,其资源均由db_cluster2接管。
@测试三，起动关闭的节点1，待节点1起动以后，稍等片刻再看资源状态，已恢复到了测试一状态。
本文出自 “漫步人生” 博客，请务必保留此出处http://anxiongbo.blog.51cto.com/805770/167706
