| |
Recently a project needs to do a highly available MySQL + Keepalived master-slave synchronization with it hundreds of times before, of course, but think of how to hang in the main library, with the fastest speed to restore the data from the database to take over and get back to take over the main library, so compiled under the idea to write the script as follows, for your reference
A. Backup the current database from the library
NOTE: All of the following configuration requires a database is not running in the state, it is recommended to do a iptables rule prohibiting non-administrator IP access from the main server IP, as well as VIP, so can guarantee the administrator can configure the SSH remote login, but also to restore the primary from the environment, to be configured to lift the ban.
# Vi /etc/rc.d/mysql_bak.sh
-------------------------
#! / Bin / bash
# Environment Variables
PATH = / bin: / usr / bin: / sbin: / usr / sbin; export PATH
export LANG = C
Daily backup directory #
basedir = / backup / daily /
Database Account #
user = root
passwd = 123456
# Mysql backup directory
mysql_db_bak = $ basedir / $ (date +% F_% H% M)
[! -d "$ Mysql_db_bak"] && mkdir -p $ mysql_db_bak
All library under the name # mysql
db_name = ( "information_schema" "mysql" "test_db1" "test_db2" "test_db3" "test_db4")
# 1 MySQL database backup
for ((i = 0; i = $ {# db_name [@]};! ++ i))
{
mysqldump -u $ user -p $ passwd --opt --skip-lock-tables --flush-logs --database $ {db_name [i]}> $ mysql_db_bak / $ {db_name [i]}. sql
}
-------------------------
B. Copied from (192.168.7.9) database backup to the main library (192.168.7.12)
# Vi /etc/rc.d/scp_mysql.sh
----------------------
# / Bin / bash
# Environment Variables
PATH = / bin: / usr / bin: / sbin: / usr / sbin; export PATH
scp / backup / daily / 2013-09-10_1258 / * 192.168.7.9:/tmp
----------------------
III. Recovery main library (192.168.7.9)
# Vi /etc/rc.d/recover_mysql.sh
------------------------------------
# / Bin / bash
# Environment Variables
PATH = / bin: / usr / bin: / sbin: / usr / sbin; export PATH
export LANG = C
Database Account #
user = root
passwd = 123456
All library under the name # # mysql
db_name = ( "information_schema" "mysql" "test_db1" "test_db2" "test_db3" "test_db4")
# Restore MySQL database
cd / tmp
for ((i = 0; i = $ {# db_name [@]};! ++ i))
{
mysql -u $ user -p $ passwd -e "drop database $ {db_name [i]};"
mysql -u $ user -p $ passwd -e "create database $ {db_name [i]};"
mysql -u $ user -p $ passwd $ {db_name [i]} <$ {db_name [i]}. sql
}
# Open from Library
mysql -u $ user -p $ passwd -e "stop slave;"
mysql -u $ user -p $ passwd -e "reset slave;"
# Set the connection parameters of the main library from the library, you can use the relevant command
mysql -u $ user -p $ passwd -e "change master to master_host = '192.168.7.9', master_user = 'slave', master_password = '123456 #', master_log_file = 'bin.000029', master_log_pos = 106";
mysql -u $ user -p $ passwd -e "start slave;"
# Check whether the connection is successful master-slave
mysql -u $ user -p $ passwd -e "show slave status G;"
------------------------------------
C. Check master-slave synchronization information
1. Check the main library binlog log and offsets
# Mysql -u root -p123456 -e "show master status"
2. View from the library is connected to the main library
# Mysql -u root -p123456 -e "show slave status G;"
Note: The actual operation found that once the VIP have a chance to cause the main switch at the same time will be obtained from the server to the virtual IP, leading to the master database from the conflict, whether or not to restart keepalived can not get rid of the VIP, currently my temporary solution the two servers are restarted, if we have a better way, please leave a message tell me, be grateful.
Follow-up:
Read a lot of information on the Internet, the initial judge in post-conflict VIP switching caused because pkill keepalived switching operation is triggered when used to kill keepaliaved process, causing the system just kill keepalived own process, not enough time to switch the VIP,
The current solution is to use a regular keepalived startup script /etc/init.d/keepalived stop to close
Initial tests found no abnormalities. . .
If the Lord did not automatically turn off when you switch from the VIP, VIP Lord appeared, from the normal open, causing both to get VIP, resulting in conflicting situations, you can manually enter the following command to delete one of the VIP address
# Ip addr del "virtual ip" dev eth0
However, this method is only a temporary solution, when manually delete the VIP, restart keepalived services will not normally open vip, need to restart the server in order to recover. |
|
|
|