前 言
摘 要
MHA 是当 master 出现故障,挑选一个 slave 作为新的 master 并构建成新的
主从架构的管理工具。从 master 出现故障到构建成新的主从架构时间是 10-30
秒。在 master 出现故障时可能会出现 slave 同步的数据不一致的现象,此工具
可以自动应用差异的中继日志到其他 slave 上保证数据的一致性。
一.Mha 优点
1.M aster crash 时可以快速的进行故障切换。
9-12 秒内可以检测到 master 故障, 7-10 秒内可以关闭 master 机器避免脑
裂,在几秒内可以应用差异日志,并构建新的主从架构,整个过程大约在 10-30
秒内可以完成,最大化的减少故障修复时间。
2.M aster crash 时可以最大化的减少数据丢失
当 master crash 时 MHA 自动检测选择数据同步最全的 slave,并把差异日
志应用到其他 slave 上, 以保障数据的一致性。结合使用 mysql
3.Semi-Synchronous Replication 可以最大化的减少数据的丢失。
MHA 的更改升级配置等不影响线上正在运行的数据库
使用 mha 不需要增加太多的服务器
MHA 由 MHA Manager 和 MHA Node 组成。 MHA Node 运行在 MYSQL 服务器上,
所以不会因为 MHA node 增加新的服务器。MHA Manager 通常需要独立运行在一
台服务器上,所以你需要增加一台服务器用于监控管理运行 MHA Manager,但
是一台服务器上的 MHA Manager 可以同时监控管理多达百台 master,所以总的
来说服务器增加不会太多。MHA Manger 也可以运行在一台 slave 上,这样总的
服务器数也不会增加。
4.原来应用系统整体性能不会降低太多
MHA 工作在异步或半同步的主从架构上。当监控 master 时,MHA 会每隔几
Mha 部署测试文档
5
秒 (默认 3 秒) 向 master 发出 ping 包并且不需要大的 sql 语句用于监控 master
的健康状况。Slave 需要开启 binlog,整体性能不会有太大的降低。
5.MHA 适合任何存储引擎
只要能主从复制的存储引擎它都支持,不限于支持事物的 innodb 引擎。
Mha 部署测试文档
二.安 装
共有 4 台服务器: 一个管理服务器,一个 master 服务器,两个 slave 服务器。
操作系统 Centos 6.4 64 bit 。
192.168.186.141 MYSQL.COM
192.168.186.142 SLAVE1.COM
192.168.186.146 SLAVE2.COM
192.168.186.144 MANAGER.COM
数据库版本mysql-5.6.10
1.首先在三台机器上装编译安装MYSQL-5.6.10
关闭selinux iptables 服务以便后期主从同步不出错
[root@MYSQL ~]# cd /usr/local/src/ [root@MYSQL src]# ls installmysql5.sh mysql-5.6.10 mysql-5.6.10.tar.gz [root@MYSQL src]# sh installmysql5.sh please enter you mysql version (eg:/mysql-5.5.34):mysql-5.6.10 please enter you mysql datadir (eg:/data/mysql/data):/date/mysql/data
2.配置HOSTS环境
[root@MANAGER ~]# vi /etc/hosts [root@MYSQL etc]# vi /etc/hosts 127.0.0.1 localhost localhost.localdomain localhost4 localhost4.localdomain4 ::1 localhost localhost.localdomain localhost6 localhost6.localdomain6 192.168.186.141 MYSQL.COM 192.168.186.142 SLAVE1.COM 192.168.186.146 SLAVE2.COM 192.168.186.144 MANAGER.COM [root@MYSQL etc]# for i in 142 146 144;do scp /etc/hosts 192.168.186.$i:/etc/;done [email protected]'s password: Permission denied, please try again. [email protected]'s password: Permission denied, please try again. [email protected]'s password: hosts 100% 266 0.3KB/s 00:00 [email protected]'s password: hosts 100% 266 0.3KB/s 00:00 [email protected]'s password:
3.安装MYSQL 主从半同步
# 所有mysql数据库服务器,安装半同步插件(semisync_master.so,semisync_slave.so) mysql> install plugin rpl_semi_sync_master soname 'semisync_master.so'; mysql> install plugin rpl_semi_sync_slave soname 'semisync_slave.so'; [root@MYSQL etc]vi /etc/my.cnf [mysqld] rpl_semi_sync_master_enabled=1 rpl_semi_sync_master_timeout=1000 rpl_semi_sync_slave_enabled=1 relay_log_purge=0 skip-name-resolve #socket=/usr/mysql.sock #auto_increment_offset = 2 #auto_increment_increment = 2 server-id = 1 log-bin=mysql-bin read_only=1 slave-skip-errors=1396
三台机器可以全部开启 仅仅server-id不同
mysql> show variables like '%sync%';
# 查看半同步状态:
mysql> show status like '%sync%';
# 有几个状态参数值得关注的:
rpl_semi_sync_master_status:显示主服务是异步复制模式还是半同步复制模式
rpl_semi_sync_master_clients:显示有多少个从服务器配置为半同步复制模式
rpl_semi_sync_master_yes_tx:显示从服务器确认成功提交的数量
rpl_semi_sync_master_no_tx:显示从服务器确认不成功提交的数量
rpl_semi_sync_master_tx_avg_wait_time:事务因开启semi_sync,平均需要额外等待的时间
rpl_semi_sync_master_net_avg_wait_time:事务进入等待队列后,到网络平均等待时间
[root@MYSQL src]# service mysqld restart 每台机器重启
4.每一台机器配置互相无交互
[root@MYSQL src]# cat /etc/hosts 127.0.0.1 localhost localhost.localdomain localhost4 localhost4.localdomain4 ::1 localhost localhost.localdomain localhost6 localhost6.localdomain6 192.168.186.141 MYSQL.COM 192.168.186.142 SLAVE1.COM 192.168.186.146 SLAVE2.COM 192.168.186.144 MANAGER.COM [root@MYSQL src]# ssh-keygen [root@MYSQL src]# ssh-copy-id 192.168.186.142 [root@MYSQL src]# ssh-copy-id 192.168.186.144 [root@MYSQL src]# ssh-copy-id 192.168.186.146 其他台配置方法与以上一致 保证每台互相无交互 配置完成后每一个都登录一次 [root@MYSQL ~]# ssh MANAGER.COM [root@MYSQL ~]# ssh SLAVE1.COM [root@MYSQL ~]# ssh SALVE2.COM 作用首次连接需要输入一次YES 在know-hosts问价记录,达成无交互
5.配置主从
执行主从脚本
建议自己做 用这个脚本要是后期带来各种不便 不要怪我
详见附件
[root@MYSQL src]# sh mslave.sh please enter you mysql SLAVEIP (eg:192.168.152.138):192.168.186.142 please enter you master mysql password (eg:yunwei123):123 please enter you slave mysql password (eg:yunwei123):123 please enter you master mysql binlog (eg:mysql-bin.000001):mysql-bin.000001 [root@MYSQL src]# sh mslave.sh please enter you mysql SLAVEIP (eg:192.168.152.138):192.168.186.146 please enter you master mysql password (eg:yunwei123):123 这个是你的主MYSQL 登录密码 please enter you slave mysql password (eg:yunwei123):123 这个是你的从的MYSQL登录密码 please enter you master mysql binlog (eg:mysql-bin.000001):mysql-bin.000001 至此MYSQL 安装主从半同步配置完成
三.安装配置MHA
1.安装
每台机器做如下操作
[root@SLAVE2 data]#rpm -ivh http://dl.fedoraproject.org/pub/epel/6/x86_64/epel-release-6-8.noarch.rpm [root@MANAGER src]# yum clean all Loaded plugins: fastestmirror, refresh-packagekit, security Cleaning repos: epel name Cleaning up Everything Cleaning up list of fastest mirrors [root@MANAGER src]# yum makecache [root@MANAGER src]# rpm --import /etc/pki/rpm-gpg/* [root@SLAVE2 data]# yum -y install perl-DBD-MySQL perl-Config-Tiny perl-Log-Dispatch perl-Parallel-ForkManager perl-Config-IniFiles ncftp perl-Params-Validate perl-CPAN perl-Test-Mock-LWP.noarch perl-LWP-Authen-Negotiate.noarch perl-devel [root@SLAVE2 data]#yum install perl-ExtUtils-CBuilder perl-ExtUtils-MakeMaker 这个百度文档没有但是显然是要装的
以下操作管理节点需要两个都安装,在3台数据库节点只要安装MHA的node节点:
1 # 如果安装下面包,报依赖关系错,请先安装mysql-share-compat包
2 # 先安装下面的 perl-dbd-mysql包
3 # 在下面执行perl时,如果出现报错,需要安装如下这几个perl包: perl-devel perl-CPAN
[root@MANAGER src]# tar -xf mha4mysql-node-0.53.tar.gz [root@MANAGER src]# cd mha4mysql-node-0.53 [root@MANAGER mha4mysql-node-0.53]# perl Makefile.PL [root@MANAGER mha4mysql-node-0.53]# make && make install ***字体这部分是每一台节点都要做的 [root@MANAGER src]# tar -xf mha4mysql-manager-0.53.tar.gz [root@MANAGER src]# cd mha4mysql-manager-0.53 [root@MANAGER mha4mysql-manager-0.53]# perl Makefile.PL [root@MANAGER mha4mysql-manager-0.53]# make && make install 根据提示输入 如果中间有卡顿现象 直接crtl+c 然后继续会出现下载的进度条,说明是正常的 [root@MANAGER src]# mkdir /etc/masterha [root@MANAGER mha]# mkdir -p /master/app1 [root@MANAGERmha]# mkdir -p /scripts [root@MANAGER mha]# cp samples/conf/* /etc/masterha/ [root@MANAGERmha]# cp samples/scripts/* /scripts [root@MANAGER mha4mysql-manager-0.53]# cp samples/conf/* /etc/masterha/
2.配置
[root@MANAGER masterha]# vi app1.cnf 内容如下; [server default] manager_workdir=/masterha/app1 manager_log=/masterha/app1/manager.log user=mha_mon password=123 ssh_user=root repl_user=slave 做主从的用户 这个也是每一台都要授权的 repl_password=yunwei123 做主从的密码 ping_interval=1 shutdown_script="" master_ip_online_change_script="" report_script="" [server1] hostname=192.168.186.141 master_binlog_dir=/data/mysql/data candidate_master=1 [server2] hostname=192.168.186.142 master_binlog_dir=/data/mysql/data candidate_master=1 [server3] hostname=192.168.186.146 master_binlog_dir=/data/mysql/data no_master=1 保存退出! [root@MANAGER masterha]# >masterha_default.cnf
3.测试ssh
[root@MANAGER masterha]# masterha_check_ssh --global_conf=/etc/masterha/masterha_default.cnf --conf=/etc/masterha/app1.cnf Wed Jul 9 02:26:57 2014 - [info] Reading default configuratoins from /etc/masterha/masterha_default.cnf.. Wed Jul 9 02:26:57 2014 - [info] Reading application default configurations from /etc/masterha/app1.cnf.. Wed Jul 9 02:26:57 2014 - [info] Reading server configurations from /etc/masterha/app1.cnf.. Wed Jul 9 02:26:57 2014 - [info] Starting SSH connection tests.. Wed Jul 9 02:26:58 2014 - [debug] Wed Jul 9 02:26:57 2014 - [debug] Connecting via SSH from [email protected](192.168.186.141:22) to [email protected](192.168.186.142:22).. Wed Jul 9 02:26:57 2014 - [debug] ok. Wed Jul 9 02:26:57 2014 - [debug] Connecting via SSH from [email protected](192.168.186.141:22) to [email protected](192.168.186.146:22).. Wed Jul 9 02:26:57 2014 - [debug] ok. Wed Jul 9 02:26:58 2014 - [debug] Wed Jul 9 02:26:57 2014 - [debug] Connecting via SSH from [email protected](192.168.186.142:22) to [email protected](192.168.186.141:22).. Wed Jul 9 02:26:57 2014 - [debug] ok. Wed Jul 9 02:26:57 2014 - [debug] Connecting via SSH from [email protected](192.168.186.142:22) to [email protected](192.168.186.146:22).. Wed Jul 9 02:26:58 2014 - [debug] ok. Wed Jul 9 02:26:58 2014 - [debug] Wed Jul 9 02:26:58 2014 - [debug] Connecting via SSH from [email protected](192.168.186.146:22) to [email protected](192.168.186.141:22).. Wed Jul 9 02:26:58 2014 - [debug] ok. Wed Jul 9 02:26:58 2014 - [debug] Connecting via SSH from [email protected](192.168.186.146:22) to [email protected](192.168.186.142:22).. Wed Jul 9 02:26:58 2014 - [debug] ok. Wed Jul 9 02:26:58 2014 - [info] All SSH connection tests passed successfully. 登入每台数据库 mysql> grant all privileges on *.* to mha_mon@'%' identified by '123'; Query OK, 0 rows affected (1.00 sec) mysql> flush privileges; Query OK, 0 rows affected (0.01 sec)
4.注意
[root@SLAVE1 ~]# ln -s /usr/local/mysql/bin/* /usr/bin
在每台MYSQL 服务器上做这件事情 极度重要哦
mysql>set global read_only=1; set global relay_log_purge=0;
在从上执行 或者干脆写到my.cnf文件里面最好
[root@SLAVE1 ~]# vi /etc/my.cnf read_only=1 slave-skip-errors=1396
为什么要跳过这个错误呢 因为啊在主里面删除用户的时候 从会报错说没有这个用户所以跳过这个错误吧
如果数据库存在空的用户 域名的用户 一定要删除否则MHA 连接MYSQL 会报错连不上 一般只要在从上面删除 如果直接没删除也OK 那就OK 吧如果报错登录不了就删除掉吧 或者跳过域名解析,或者你授权的时候记得也授权域名等等方法多种 我的方式是跳过域名解析的 skip-name-resolve
mysql> select user,host from mysql.user; +---------+---------------+ | user | host | +---------+---------------+ | root | 127.0.0.1 | | mha_mon | 192.168.186.% | | repl | 192.168.186.% | | slave | 192.168.186.% | | root | ::1 | | | SLAVE2.COM | | root | SLAVE2.COM | | root | localhost | +---------+---------------+ 8 rows in set (0.00 sec) mysql> drop user 'root'@SLAVE2.COM; Query OK, 0 rows affected (0.00 sec)
5.测试mysql
[root@MANAGER masterha]# masterha_check_repl --conf=/etc/masterha/app1.cnf Wed Jul 9 04:23:16 2014 - [warning] Global configuration file /etc/masterha_default.cnf not found. Skipping. Wed Jul 9 04:23:16 2014 - [info] Reading application default configurations from /etc/masterha/app1.cnf.. Wed Jul 9 04:23:16 2014 - [info] Reading server configurations from /etc/masterha/app1.cnf.. Wed Jul 9 04:23:16 2014 - [info] MHA::MasterMonitor version 0.53. Wed Jul 9 04:23:17 2014 - [info] Dead Servers: Wed Jul 9 04:23:17 2014 - [info] Alive Servers: Wed Jul 9 04:23:17 2014 - [info] 192.168.186.141(192.168.186.141:3306) Wed Jul 9 04:23:17 2014 - [info] 192.168.186.142(192.168.186.142:3306) Wed Jul 9 04:23:17 2014 - [info] SLAVE2.COM(192.168.186.146:3306) Wed Jul 9 04:23:17 2014 - [info] Alive Slaves: Wed Jul 9 04:23:17 2014 - [info] 192.168.186.142(192.168.186.142:3306) Version=5.6.10-log (oldest major version between slaves) log-bin:enabled Wed Jul 9 04:23:17 2014 - [info] Replicating from 192.168.186.141(192.168.186.141:3306) Wed Jul 9 04:23:17 2014 - [info] Primary candidate for the new Master (candidate_master is set) Wed Jul 9 04:23:17 2014 - [info] SLAVE2.COM(192.168.186.146:3306) Version=5.6.10-log (oldest major version between slaves) log-bin:enabled Wed Jul 9 04:23:17 2014 - [info] Replicating from 192.168.186.141(192.168.186.141:3306) Wed Jul 9 04:23:17 2014 - [info] Not candidate for the new Master (no_master is set) Wed Jul 9 04:23:17 2014 - [info] Current Alive Master: 192.168.186.141(192.168.186.141:3306) Wed Jul 9 04:23:17 2014 - [info] Checking slave configurations.. Wed Jul 9 04:23:17 2014 - [info] Checking replication filtering settings.. Wed Jul 9 04:23:17 2014 - [info] binlog_do_db= , binlog_ignore_db= Wed Jul 9 04:23:17 2014 - [info] Replication filtering check ok. Wed Jul 9 04:23:17 2014 - [info] Starting SSH connection tests.. Wed Jul 9 04:23:18 2014 - [info] All SSH connection tests passed successfully. Wed Jul 9 04:23:18 2014 - [info] Checking MHA Node version.. Wed Jul 9 04:23:19 2014 - [info] Version check ok. Wed Jul 9 04:23:19 2014 - [info] Checking SSH publickey authentication settings on the current master.. Wed Jul 9 04:23:19 2014 - [info] HealthCheck: SSH to 192.168.186.141 is reachable. Wed Jul 9 04:23:19 2014 - [info] Master MHA Node version is 0.53. Wed Jul 9 04:23:19 2014 - [info] Checking recovery script configurations on the current master.. Wed Jul 9 04:23:19 2014 - [info] Executing command: save_binary_logs --command=test --start_pos=4 --binlog_dir=/data/mysql/data --output_file=/var/tmp/save_binary_logs_test --manager_version=0.53 --start_file=mysql-bin.000001 Wed Jul 9 04:23:19 2014 - [info] Connecting to [email protected](192.168.186.141).. Creating /var/tmp if not exists.. ok. Checking output directory is accessible or not.. ok. Binlog found at /data/mysql/data, up to mysql-bin.000001 Wed Jul 9 04:23:20 2014 - [info] Master setting check done. Wed Jul 9 04:23:20 2014 - [info] Checking SSH publickey authentication and checking recovery script configurations on all alive slave servers.. Wed Jul 9 04:23:20 2014 - [info] Executing command : apply_diff_relay_logs --command=test --slave_user=mha_mon --slave_host=192.168.186.142 --slave_ip=192.168.186.142 --slave_port=3306 --workdir=/var/tmp --target_version=5.6.10-log --manager_version=0.53 --relay_log_info=/data/mysql/data/relay-log.info --relay_dir=/data/mysql/data/ --slave_pass=xxx Wed Jul 9 04:23:20 2014 - [info] Connecting to [email protected](192.168.186.142:22).. Checking slave recovery environment settings.. Opening /data/mysql/data/relay-log.info ... ok. Relay log found at /data/mysql/data, up to SLAVE1-relay-bin.000002 Temporary relay log file is /data/mysql/data/SLAVE1-relay-bin.000002 Testing mysql connection and privileges..Warning: Using a password on the command line interface can be insecure. done. Testing mysqlbinlog output.. done. Cleaning up test file(s).. done. Wed Jul 9 04:23:20 2014 - [info] Executing command : apply_diff_relay_logs --command=test --slave_user=mha_mon --slave_host=SLAVE2.COM --slave_ip=192.168.186.146 --slave_port=3306 --workdir=/var/tmp --target_version=5.6.10-log --manager_version=0.53 --relay_log_info=/data/mysql/data/relay-log.info --relay_dir=/data/mysql/data/ --slave_pass=xxx Wed Jul 9 04:23:20 2014 - [info] Connecting to [email protected](SLAVE2.COM:22).. Checking slave recovery environment settings.. Opening /data/mysql/data/relay-log.info ... ok. Relay log found at /data/mysql/data, up to slave2-relay-bin.000002 Temporary relay log file is /data/mysql/data/slave2-relay-bin.000002 Testing mysql connection and privileges..Warning: Using a password on the command line interface can be insecure. done. Testing mysqlbinlog output.. done. Cleaning up test file(s).. done. Wed Jul 9 04:23:21 2014 - [info] Slaves settings check done. Wed Jul 9 04:23:21 2014 - [info] 192.168.186.141 (current master) +--192.168.186.142 +--SLAVE2.COM Wed Jul 9 04:23:21 2014 - [info] Checking replication health on 192.168.186.142.. Wed Jul 9 04:23:21 2014 - [info] ok. Wed Jul 9 04:23:21 2014 - [info] Checking replication health on SLAVE2.COM.. Wed Jul 9 04:23:21 2014 - [info] ok. Wed Jul 9 04:23:21 2014 - [warning] master_ip_failover_script is not defined. Wed Jul 9 04:23:21 2014 - [warning] shutdown_script is not defined. Wed Jul 9 04:23:21 2014 - [info] Got exit code 0 (Not master dead). MySQL Replication Health is OK. 至此说明你的MHA 已经配置好了
6.启动
[root@MANAGER ~]# nohup masterha_manager --conf=/etc/mastermha/app1.cnf > /tmp/mha_manager.log &1 &
启动MHA
四.测试重构
1.测试
测试 将MYSQL.COM机器上的MYSQL服务关闭 ,注意观察 manager.log 日志会发现 切换到了SLAVE1.COM 并且SLAVE1.COM变成了主 而SLAVE2.COM 则变成了SLAVE1.COM 的从
root@MANAGER app1]# tail -f manager.log 这是启动后还没关闭主数据库的日志内容 192.168.186.141 (current master) +--192.168.186.142 +--SLAVE2.COM Wed Jul 9 18:52:32 2014 - [warning] master_ip_failover_script is not defined. Wed Jul 9 18:52:32 2014 - [warning] shutdown_script is not defined. Wed Jul 9 18:52:32 2014 - [info] Set master ping interval 1 seconds. Wed Jul 9 18:52:32 2014 - [warning] secondary_check_script is not defined. It is highly recommended setting it to check master reachability from two or more routes. Wed Jul 9 18:52:32 2014 - [info] Starting ping health check on 192.168.186.141(192.168.186.141:3306).. Wed Jul 9 18:52:32 2014 - [info] Ping(SELECT) succeeded, waiting until MySQL doesn't respond.. [root@MYSQL ~]# service mysqld stop Shutting down MySQL..... SUCCESS! [root@MANAGER app1]# tail -f manager.log 最要看最后几行 就知道有没有切换成功 192.168.186.141 (current master) +--192.168.186.142 +--SLAVE2.COM Wed Jul 9 18:56:47 2014 - [info] Dead Servers: Wed Jul 9 18:56:47 2014 - [info] 192.168.186.141(192.168.186.141:3306) Wed Jul 9 18:56:47 2014 - [info] Alive Servers: Wed Jul 9 18:56:47 2014 - [info] 192.168.186.142(192.168.186.142:3306) Wed Jul 9 18:56:47 2014 - [info] SLAVE2.COM(192.168.186.146:3306) Wed Jul 9 18:56:47 2014 - [info] Alive Slaves: Wed Jul 9 18:56:47 2014 - [info] 192.168.186.142(192.168.186.142:3306) Version=5.6.10-log (oldest major version between slaves) log-bin:enabled Wed Jul 9 18:56:47 2014 - [info] Replicating from 192.168.186.141(192.168.186.141:3306) Wed Jul 9 18:56:47 2014 - [info] Primary candidate for the new Master (candidate_master is set) Wed Jul 9 18:56:47 2014 - [info] SLAVE2.COM(192.168.186.146:3306) Version=5.6.10-log (oldest major version between slaves) log-bin:enabled Wed Jul 9 18:56:47 2014 - [info] Replicating from 192.168.186.141(192.168.186.141:3306) Wed Jul 9 18:56:47 2014 - [info] Not candidate for the new Master (no_master is set) Wed Jul 9 18:56:47 2014 - [info] Checking slave configurations.. Wed Jul 9 18:56:47 2014 - [info] Checking replication filtering settings.. Wed Jul 9 18:56:47 2014 - [info] Replication filtering check ok. Wed Jul 9 18:56:47 2014 - [info] Master is down! Wed Jul 9 18:56:47 2014 - [info] Terminating monitoring script. Wed Jul 9 18:56:47 2014 - [info] Got exit code 20 (Master dead). Wed Jul 9 18:56:47 2014 - [info] MHA::MasterFailover version 0.53. Wed Jul 9 18:56:47 2014 - [info] Starting master failover. Wed Jul 9 18:56:47 2014 - [info] * Phase 2: Dead Master Shutdown Phase completed. Wed Jul 9 18:56:47 2014 - [info] 192.168.186.142(192.168.186.142:3306) Version=5.6.10-log (oldest major version between slaves) log-bin:enabled Wed Jul 9 18:56:47 2014 - [info] Replicating from 192.168.186.141(192.168.186.141:3306) Wed Jul 9 18:56:47 2014 - [info] Primary candidate for the new Master (candidate_master is set) Wed Jul 9 18:56:47 2014 - [info] SLAVE2.COM(192.168.186.146:3306) Version=5.6.10-log (oldest major version between slaves) log-bin:enabled Wed Jul 9 18:56:47 2014 - [info] Replicating from 192.168.186.141(192.168.186.141:3306) Wed Jul 9 18:56:47 2014 - [info] Not candidate for the new Master (no_master is set) Wed Jul 9 18:56:47 2014 - [info] The oldest binary log file/position on all slaves is mysql-bin.000001:214 Wed Jul 9 18:56:47 2014 - [info] Oldest slaves: Wed Jul 9 18:56:47 2014 - [info] 192.168.186.142(192.168.186.142:3306) Version=5.6.10-log (oldest major version between slaves) log-bin:enabled Wed Jul 9 18:56:47 2014 - [info] Replicating from 192.168.186.141(192.168.186.141:3306) Wed Jul 9 18:56:47 2014 - [info] Primary candidate for the new Master (candidate_master is set) Wed Jul 9 18:56:47 2014 - [info] SLAVE2.COM(192.168.186.146:3306) Version=5.6.10-log (oldest major version between slaves) log-bin:enabled Wed Jul 9 18:56:47 2014 - [info] Replicating from 192.168.186.141(192.168.186.141:3306) Wed Jul 9 18:56:47 2014 - [info] Not candidate for the new Master (no_master is set) Wed Jul 9 18:56:47 2014 - [info] Wed Jul 9 18:56:47 2014 - [info] * Phase 3.2: Saving Dead Master's Binlog Phase.. Wed Jul 9 18:56:47 2014 - [info] Wed Jul 9 18:56:48 2014 - [info] Fetching dead master's binary logs.. Wed Jul 9 18:56:48 2014 - [info] Executing command on the dead master 192.168.186.141(192.168.186.141:3306): save_binary_logs --command=save --start_file=mysql-bin.000001 --start_pos=214 --binlog_dir=/data/mysql/data --output_file=/var/tmp/saved_master_binlog_from_192.168.186.141_3306_20140709185647.binlog --handle_raw_binlog=1 --disable_log_bin=0 --manager_version=0.53 Creating /var/tmp if not exists.. ok. Concat binary/relay logs from mysql-bin.000001 pos 214 to mysql-bin.000001 EOF into /var/tmp/saved_master_binlog_from_192.168.186.141_3306_20140709185647.binlog .. Dumping binlog format description event, from position 0 to 120.. ok. Dumping effective binlog data from /data/mysql/data/mysql-bin.000001 position 214 to tail(237).. ok. Concat succeeded. Wed Jul 9 18:56:48 2014 - [info] scp from [email protected]:/var/tmp/saved_master_binlog_from_192.168.186.141_3306_20140709185647.binlog to local:/masterha/app1/saved_master_binlog_from_192.168.186.141_3306_20140709185647.binlog succeeded. Wed Jul 9 18:56:49 2014 - [info] HealthCheck: SSH to 192.168.186.142 is reachable. Wed Jul 9 18:56:49 2014 - [info] HealthCheck: SSH to SLAVE2.COM is reachable. Wed Jul 9 18:56:49 2014 - [info] Wed Jul 9 18:56:49 2014 - [info] * Phase 3.3: Determining New Master Phase.. Wed Jul 9 18:56:49 2014 - [info] Wed Jul 9 18:56:49 2014 - [info] Finding the latest slave that has all relay logs for recovering other slaves.. Wed Jul 9 18:56:49 2014 - [info] All slaves received relay logs to the same position. No need to resync each other. Wed Jul 9 18:56:49 2014 - [info] Searching new master from slaves.. Wed Jul 9 18:56:49 2014 - [info] Candidate masters from the configuration file: Wed Jul 9 18:56:49 2014 - [info] 192.168.186.142(192.168.186.142:3306) Version=5.6.10-log (oldest major version between slaves) log-bin:enabled Wed Jul 9 18:56:49 2014 - [info] Replicating from 192.168.186.141(192.168.186.141:3306) Wed Jul 9 18:56:49 2014 - [info] Primary candidate for the new Master (candidate_master is set) Wed Jul 9 18:56:49 2014 - [info] Non-candidate masters: Wed Jul 9 18:56:49 2014 - [info] SLAVE2.COM(192.168.186.146:3306) Version=5.6.10-log (oldest major version between slaves) log-bin:enabled Wed Jul 9 18:56:49 2014 - [info] Replicating from 192.168.186.141(192.168.186.141:3306) Wed Jul 9 18:56:49 2014 - [info] Not candidate for the new Master (no_master is set) Wed Jul 9 18:56:49 2014 - [info] Searching from candidate_master slaves which have received the latest relay log events.. Wed Jul 9 18:56:49 2014 - [info] New master is 192.168.186.142(192.168.186.142:3306) Wed Jul 9 18:56:49 2014 - [info] Starting master failover.. Wed Jul 9 18:56:49 2014 - [info] From: 192.168.186.141 (current master) +--192.168.186.142 +--SLAVE2.COM To: 192.168.186.142 (new master) +--SLAVE2.COM Wed Jul 9 18:56:49 2014 - [info] Wed Jul 9 18:56:49 2014 - [info] * Phase 3.3: New Master Diff Log Generation Phase.. Wed Jul 9 18:56:49 2014 - [info] Wed Jul 9 18:56:49 2014 - [info] This server has all relay logs. No need to generate diff files from the latest slave. Wed Jul 9 18:56:49 2014 - [info] Sending binlog.. Wed Jul 9 18:56:50 2014 - [info] scp from local:/masterha/app1/saved_master_binlog_from_192.168.186.141_3306_20140709185647.binlog to [email protected]:/var/tmp/saved_master_binlog_from_192.168.186.141_3306_20140709185647.binlog succeeded. Wed Jul 9 18:56:50 2014 - [info] Wed Jul 9 18:56:50 2014 - [info] * Phase 3.4: Master Log Apply Phase.. Wed Jul 9 18:56:50 2014 - [info] Wed Jul 9 18:56:50 2014 - [info] *NOTICE: If any error happens from this phase, manual recovery is needed. Wed Jul 9 18:56:50 2014 - [info] Starting recovery on 192.168.186.142(192.168.186.142:3306).. Wed Jul 9 18:56:50 2014 - [info] Generating diffs succeeded. Wed Jul 9 18:56:50 2014 - [info] Waiting until all relay logs are applied. Wed Jul 9 18:56:50 2014 - [info] done. Wed Jul 9 18:56:50 2014 - [info] Getting slave status.. Wed Jul 9 18:56:50 2014 - [info] This slave(192.168.186.142)'s Exec_Master_Log_Pos equals to Read_Master_Log_Pos(mysql-bin.000001:214). No need to recover from Exec_Master_Log_Pos. Wed Jul 9 18:56:50 2014 - [info] Connecting to the target slave host 192.168.186.142, running recover script.. Wed Jul 9 18:56:50 2014 - [info] Executing command: apply_diff_relay_logs --command=apply --slave_user=mha_mon --slave_host=192.168.186.142 --slave_ip=192.168.186.142 --slave_port=3306 --apply_files=/var/tmp/saved_master_binlog_from_192.168.186.141_3306_20140709185647.binlog --workdir=/var/tmp --target_version=5.6.10-log --timestamp=20140709185647 --handle_raw_binlog=1 --disable_log_bin=0 --manager_version=0.53 --slave_pass=xxx Wed Jul 9 18:56:50 2014 - [info] MySQL client version is 5.6.10. Using --binary-mode. Applying differential binary/relay log files /var/tmp/saved_master_binlog_from_192.168.186.141_3306_20140709185647.binlog on 192.168.186.142:3306. This may take long time... Applying log files succeeded. Wed Jul 9 18:56:50 2014 - [info] All relay logs were successfully applied. Wed Jul 9 18:56:50 2014 - [info] Getting new master's binlog name and position.. Wed Jul 9 18:56:50 2014 - [info] mysql-bin.000007:504 Wed Jul 9 18:56:50 2014 - [info] All other slaves should start replication from here. Statement should be: CHANGE MASTER TO MASTER_HOST='192.168.186.142', MASTER_PORT=3306, MASTER_LOG_FILE='mysql-bin.000007', MASTER_LOG_POS=504, MASTER_USER='repl', MASTER_PASSWORD='xxx'; Wed Jul 9 18:56:50 2014 - [warning] master_ip_failover_script is not set. Skipping taking over new master ip address. Wed Jul 9 18:56:50 2014 - [info] Setting read_only=0 on 192.168.186.142(192.168.186.142:3306).. Wed Jul 9 18:56:50 2014 - [info] ok. Wed Jul 9 18:56:50 2014 - [info] ** Finished master recovery successfully. Wed Jul 9 18:56:50 2014 - [info] * Phase 3: Master Recovery Phase completed. Wed Jul 9 18:56:50 2014 - [info] Wed Jul 9 18:56:50 2014 - [info] * Phase 4: Slaves Recovery Phase.. Wed Jul 9 18:56:50 2014 - [info] Wed Jul 9 18:56:50 2014 - [info] * Phase 4.1: Starting Parallel Slave Diff Log Generation Phase.. Wed Jul 9 18:56:50 2014 - [info] Wed Jul 9 18:56:50 2014 - [info] -- Slave diff file generation on host SLAVE2.COM(192.168.186.146:3306) started, pid: 3135. Check tmp log /masterha/app1/SLAVE2.COM_3306_20140709185647.log if it takes time.. Wed Jul 9 18:56:50 2014 - [info] Wed Jul 9 18:56:50 2014 - [info] Log messages from SLAVE2.COM ... Wed Jul 9 18:56:50 2014 - [info] Wed Jul 9 18:56:50 2014 - [info] This server has all relay logs. No need to generate diff files from the latest slave. Wed Jul 9 18:56:50 2014 - [info] End of log messages from SLAVE2.COM. Wed Jul 9 18:56:50 2014 - [info] -- SLAVE2.COM(192.168.186.146:3306) has the latest relay log events. Wed Jul 9 18:56:50 2014 - [info] Generating relay diff files from the latest slave succeeded. Wed Jul 9 18:56:50 2014 - [info] Wed Jul 9 18:56:50 2014 - [info] * Phase 4.2: Starting Parallel Slave Log Apply Phase.. Wed Jul 9 18:56:50 2014 - [info] Wed Jul 9 18:56:50 2014 - [info] -- Slave recovery on host SLAVE2.COM(192.168.186.146:3306) started, pid: 3137. Check tmp log /masterha/app1/SLAVE2.COM_3306_20140709185647.log if it takes time.. Wed Jul 9 18:56:51 2014 - [info] Wed Jul 9 18:56:51 2014 - [info] Log messages from SLAVE2.COM ... Wed Jul 9 18:56:51 2014 - [info] Wed Jul 9 18:56:50 2014 - [info] Sending binlog.. Wed Jul 9 18:56:51 2014 - [info] scp from local:/masterha/app1/saved_master_binlog_from_192.168.186.141_3306_20140709185647.binlog to [email protected]:/var/tmp/saved_master_binlog_from_192.168.186.141_3306_20140709185647.binlog succeeded. Wed Jul 9 18:56:51 2014 - [info] Starting recovery on SLAVE2.COM(192.168.186.146:3306).. Wed Jul 9 18:56:51 2014 - [info] Generating diffs succeeded. Wed Jul 9 18:56:51 2014 - [info] Waiting until all relay logs are applied. Wed Jul 9 18:56:51 2014 - [info] done. Wed Jul 9 18:56:51 2014 - [info] Getting slave status.. Wed Jul 9 18:56:51 2014 - [info] This slave(SLAVE2.COM)'s Exec_Master_Log_Pos equals to Read_Master_Log_Pos(mysql-bin.000001:214). No need to recover from Exec_Master_Log_Pos. Wed Jul 9 18:56:51 2014 - [info] Connecting to the target slave host SLAVE2.COM, running recover script.. Wed Jul 9 18:56:51 2014 - [info] Executing command: apply_diff_relay_logs --command=apply --slave_user=mha_mon --slave_host=SLAVE2.COM --slave_ip=192.168.186.146 --slave_port=3306 --apply_files=/var/tmp/saved_master_binlog_from_192.168.186.141_3306_20140709185647.binlog --workdir=/var/tmp --target_version=5.6.10-log --timestamp=20140709185647 --handle_raw_binlog=1 --disable_log_bin=0 --manager_version=0.53 --slave_pass=xxx Wed Jul 9 18:56:51 2014 - [info] MySQL client version is 5.6.10. Using --binary-mode. Applying differential binary/relay log files /var/tmp/saved_master_binlog_from_192.168.186.141_3306_20140709185647.binlog on SLAVE2.COM:3306. This may take long time... Applying log files succeeded. Wed Jul 9 18:56:51 2014 - [info] All relay logs were successfully applied. Wed Jul 9 18:56:51 2014 - [info] Resetting slave SLAVE2.COM(192.168.186.146:3306) and starting replication from the new master 192.168.186.142(192.168.186.142:3306).. Wed Jul 9 18:56:51 2014 - [info] Executed CHANGE MASTER. Wed Jul 9 18:56:51 2014 - [info] Slave started. Wed Jul 9 18:56:51 2014 - [info] End of log messages from SLAVE2.COM. Wed Jul 9 18:56:51 2014 - [info] -- Slave recovery on host SLAVE2.COM(192.168.186.146:3306) succeeded. Wed Jul 9 18:56:51 2014 - [info] All new slave servers recovered successfully. Wed Jul 9 18:56:51 2014 - [info] Wed Jul 9 18:56:51 2014 - [info] * Phase 5: New master cleanup phease.. Wed Jul 9 18:56:51 2014 - [info] Wed Jul 9 18:56:51 2014 - [info] Resetting slave info on the new master.. Wed Jul 9 18:56:51 2014 - [info] 192.168.186.142: Resetting slave info succeeded. Wed Jul 9 18:56:51 2014 - [info] Master failover to 192.168.186.142(192.168.186.142:3306) completed successfully. Wed Jul 9 18:56:51 2014 - [info] ----- Failover Report ----- app1: MySQL Master failover 192.168.186.141 to 192.168.186.142 succeeded Master 192.168.186.141 is down! Check MHA Manager logs at MANAGER.COM:/masterha/app1/manager.log for details. Started automated(non-interactive) failover. The latest slave 192.168.186.142(192.168.186.142:3306) has all relay logs for recovery. Selected 192.168.186.142 as a new master. 192.168.186.142: OK: Applying all logs succeeded. SLAVE2.COM: This host has the latest relay log events. Generating relay diff files from the latest slave succeeded. SLAVE2.COM: OK: Applying all logs succeeded. Slave started, replicating from 192.168.186.142. 192.168.186.142: Resetting slave info succeeded. 192.168.186.142: Resetting slave info succeeded.
看到192.168.186.142: Resetting slave info succeeded.
Master failover to 192.168.186.142(192.168.186.142:3306) completed successfully. 说明切换成功了
[root@slave2 ~]# mysql -e "show slave status\G" *************************** 1. row *************************** Slave_IO_State: Waiting for master to send event Master_Host: 192.168.186.142 Master_User: repl Master_Port: 3306 Connect_Retry: 60 Master_Log_File: mysql-bin.000007 Read_Master_Log_Pos: 504 Relay_Log_File: slave2-relay-bin.000002 Relay_Log_Pos: 283 Relay_Master_Log_File: mysql-bin.000007 Slave_IO_Running: Yes Slave_SQL_Running: Yes 看到已经切换到和142同步了 本来是和141同步的 此时的SALVE1.COM已经变成主了 说明已然生效
2.重构
重构我想就不要我做了吧,就是这时候等于你的主挂了 切换在SLAVE1.COM 上变成了主 因此重构我提供一种方案(方案多种),拿出一台新的服务器重新加入以142为主做成从 再app1.conf配置文件中加入
[root@MANAGER app1]# rm -rf app1.failover.complete
删除该文件后 再次启动manager端即可
五.扩展Keepalived
1.安装keepalived
[root@MYSQL src]# wget http://www.keepalived.org/software/keepalived-1.2.12.tar.gz [root@MYSQL src]# tar -xf keepalived-1.2.12.tar.gz [root@MYSQL src]# cd keepalived-1.2.12 [root@MYSQL src]# yum -y install gcc gcc-c++ gcc-g77 ncurses-devel bison libaio-devel cmake libnl* libpopt* popt-static openssl-devel [root@MYSQL keepalived-1.2.12]# ./configure [root@MYSQL keepalived-1.2.12]# make && make install [root@MYSQL src]#mkdir /etc/keepalived/ [root@MYSQL src]# cp /usr/local/etc/keepalived/keepalived.conf /etc/keepalived/ [root@MYSQL src]# cp /usr/local/etc/rc.d/init.d/keepalived /etc/init.d/ [root@MYSQL src]# cp /usr/local/etc/sysconfig/keepalived /etc/sysconfig/ [root@MYSQL src]#cp /usr/local/sbin/keepalived /usr/sbin/
2.配置keepalived
[root@MYSQL keepalived]# vi keepalived.conf ! Configuration File for keepalived global_defs { notification_email { [email protected] [email protected] [email protected] } notification_email_from [email protected] smtp_server 192.168.200.1 smtp_connect_timeout 30 router_id LVS_DEVEL ##配置是为了标识当前节点,两个节点的此项设置可相同,也可不相同 } vrrp_instance VI_1 { state MASTER #指定A节点为主节点 备用节点上设置为BACKUP即可 interface eth0 virtual_router_id 51 #VRRP组名,两个节点的设置必须一样,以指明各个节点属于同一VRRP priority 100 #主节点的优先级(1-254之间),备用节点必须比主节点优先级低 advert_int 1 authentication { #设置验证信息,两个节点必须一致 auth_type PASS auth_pass 1111 } virtual_ipaddress { 192.168.200.16 } }
3.检测mysql服务脚本
[root@MYSQL keepalived]# vi /root/check_mysql.sh #!/bin/bash MYSQL=/usr/local/mysql/bin/mysql MYSQL_HOST=127.0.0.1 MYSQL_USER=root MYSQL_PASSWORD=123 CHECK_TIME=3 #mysql is working MYSQL_OK is 1 , mysql down MYSQL_OK is 0 MYSQL_OK=1 function check_mysql_helth (){ $MYSQL -h $MYSQL_HOST -u $MYSQL_USER -e "show status;" >/dev/null 2>&1 if [ $? = 0 ] ;then MYSQL_OK=1 else MYSQL_OK=0 fi return $MYSQL_OK } while [ $CHECK_TIME -ne 0 ] do let "CHECK_TIME -= 1" check_mysql_helth if [ $MYSQL_OK = 1 ] ; then CHECK_TIME=0 exit 0 fi if [ $MYSQL_OK -eq 0 ] && [ $CHECK_TIME -eq 0 ] then pkill keepalived exit 1 fi sleep 1 Done
该脚本做一个计划任务每分钟做一次检查MYSQL 服务如果挂掉了的话 VIP 就会在SALVE1.COM 上生效 这样的话 client 端连接的VIP 就会从MYSQL.COM 上变到了SLAVE1.COM 上 此时SLAVE1.COM 由于MHA的生效 已经变为主对外提供服务了 VIP 也在SALVE1.COM上 所以从头到尾 客户端只需要连接VIP 就可以了
[root@MYSQL keepalived]# yum -y install cronie [root@MYSQL ~]# crontab -l */1 * * * * bash /root/checkmysql.sh
以上步骤再在下一次切换的SLAVE上做一次也就是SLAVE1.COM上做一次 keepalived.conf配置文件不同的地方已做注释
4.启动测试
两边启动keepalived
[root@MYSQL keepalived]# keepalived -f /etc/keepalived/keepalived.conf [root@MYSQL keepalived]# ps -ef |grep keep root 3230 1 0 23:27 ? 00:00:00 keepalived -f /etc/keepalived/keepalived.conf root 3231 3230 0 23:27 ? 00:00:00 keepalived -f /etc/keepalived/keepalived.conf root 3232 3230 0 23:27 ? 00:00:00 keepalived -f /etc/keepalived/keepalived.conf root 3234 2538 0 23:27 pts/0 00:00:00 grep keep [root@MYSQL keepalived]# ip a 1: lo:mtu 16436 qdisc noqueue state UNKNOWN link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00 inet 127.0.0.1/8 scope host lo inet6 ::1/128 scope host valid_lft forever preferred_lft forever 2: eth0: mtu 1500 qdisc pfifo_fast state UP qlen 1000 link/ether 00:0c:29:c9:85:ba brd ff:ff:ff:ff:ff:ff inet 192.168.186.141/24 brd 192.168.186.255 scope global eth0 inet 192.168.200.16/32 scope global eth0 inet6 fe80::20c:29ff:fec9:85ba/64 scope link valid_lft forever preferred_lft forever
测试关闭该主机的MYSQL 服务看看VIP 是否飘到配好keepalive的SLAVE1.COM上 MHA中的MYSQL 主是不是也变成了SLAVE1.COM
如是 至此全部完成。。。。。。。。。。。。。。。。。。。。。。。。。。。。。。。。。。。。。。。。。。。。谢谢!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
mha日常维护命令
1.查看ssh登陆是否成功 masterha_check_ssh --conf=/etc/masterha/app1.cnf 2.查看复制是否建立好 masterha_check_repl --conf=/etc/masterha/app1.cnf 3.启动mha nohup masterha_manager --conf=/etc/masterha/app1.cnf > /tmp/mha_manager.log < /dev/null 2>&1 & 当有slave节点宕掉的情况是启动不了的,加上--ignore_fail_on_start即使有节点宕掉也能启动mha nohup masterha_manager --conf=/etc/masterha/app1.cnf --ignore_fail_on_start > /tmp/mha_manager.log < /dev/null 2>&1 & 4.检查启动的状态 masterha_check_status --conf=/etc/masterha/app1.cnf 5.停止mha masterha_stop --conf=/etc/masterha/app1.cnf 6.failover后下次重启 每次failover切换后会在管理目录生成文件app1.failover.complete ,下次在切换的时候会发现有这个文件导致切换不成功,需要手动清理掉。 rm -rf /masterha/app1/app1.failover.complete 也可以加上参数--ignore_last_failover 7.手工failover 手工failover场景,master死掉,但是masterha_manager没有开启,可以通过手工failover: masterha_master_switch --conf=/etc/masterha/app1.cnf --dead_master_host=10.50.2.10 --master_state=dead --new_master_host=10.50.2.12 --ignore_last_failover 8.masterha_manager是一种监视和故障转移的程序。另一方面,masterha_master_switch程序不监控主库。 masterha_master_switch可以用于主库故障转移,也可用于在线总开关。 9.手动在线切换 masterha_master_switch --conf=/etc/app1.cnf --master_state=alive --new_master_host=192.168.119.74 --orig_master_is_new_slave 或者 masterha_master_switch --conf=/etc/app1.cnf --master_state=alive --new_master_host=192.168.119.74 --orig_master_is_new_slave --running_updates_limit=10000 --orig_master_is_new_slave切换时加上此参数是将原master变为slave节点,如果不加此参数,原来的master将不启动 --running_updates_limit=10000 切换时候选master如果有延迟的话,mha切换不能成功,加上此参数表示延迟在此时间范围内都可切换(单位为s),但是切换的时间长短是由recover时relay日志的大小决定 手动在线切换mha,切换时需要将在运行的mha停掉后才能切换。 在备库先执行DDL,一般先stop slave,一般不记录mysql日志,可以通过set SQL_LOG_BIN = 0实现。然后进行一次主备切换操作,再在原来的主库上执行DDL。这种方法适用于增减索引,如果是增加字段就需要额外注意。 可以通过如下命令停止mha masterha_stop --conf=/etc/app1.cnf