mysql5.6-gtid-半同步-ssl-mha-keepalived方案
Mysql5.6简介 在MySQL 5.5发布两年后，Oracle宣布MySQL 5.6正式版发布，首个正式版版本号为MySQL 5.6.10。在MySQL 5.5中使用的是InnoDB作为默认的存储引擎，而MySQL 5.6则对InnoDB引擎进行了改造，提供全文索引能力，使InnoDB适合各种应用场景。 1、运行环境配置及安装 1.1、部署环境 系统：CentOS6.4 （64位）最小化安装 1.2、下载地址及版本 http://dev.mysql.com/downloads/mysql/ 下载版本：mysql-5.6.19.tar.gz 1.3、安装 1.3.1编译环境准备 yum groupinstall -y"Desktop Platform Development" "Development tools" yum install -y vim cmakebison-devel ncurses-devel perl-DBD-MySQL openssl openssl-devel freetype-demos libcurl-devellibjpeg-turbo-static libpng-static wget 关闭selinux setenforce0 修改selinux配置文件 vim/etc/selinux/config SELINUX=disabled 编辑主从hosts文件 shell>vim/etc/hosts 192.168.40.128master.testdb.com 192.168.40.129slave.testdb.com 1.3.2建立mysql所需账号、目录及权限 groupadd mysql useradd -g mysql -s /bin/false -M mysql mkdir -pv /data/mysqldata mkdir -pv /data/mysqlbinlog chown -R mysql.mysql /data/mysqldata chown -R mysql.mysql /data/mysqlbinlog 1.3.3解压并安装mysql 注：在启动MySQL服务时，会按照一定次序搜索my.cnf，先在/etc目录下找，找不到则会搜索"$basedir/my.cnf"，在本例中就是 /usr/local/mysql/my.cnf，这是新版MySQL的配置文件的默认位置！注意：在CentOS 6.4版操作系统的最小安装完成后，在/etc目录下会存在一个my.cnf，需要将此文件更名为其他的名字，如：/usr/local/mysql/my.cnf.bak，否则，该文件会干扰源码安装的MySQL的正确配置，造成无法启动。 tar xf mysql-5.6.19.tar.gz cd mysql-5.6.19 cmake -DCMAKE_INSTALL_PREFIX=/usr/local/mysql\ -DDEFAULT_CHARSET=utf8\ -DDEFAULT_COLLATION=utf8_general_ci\ -DEXTRA_CHARSETS=all\ -DWITH_MYISAM_STORAGE_ENGINE=1\ -DWITH_INNOBASE_STORAGE_ENGINE=1\ -DWITH_ARCHIVE_STORAGE_ENGINE=1\ -DWITH_BLACKHOLE_STORAGE_ENGINE=1\ -DWITH_MEMORY_STORAGE_ENGINE=1\ -DWITH_FEDERATED_STORAGE_ENGINE=1\ -DWITH_READLINE=1\ -DENABLED_LOCAL_INFILE=1\ -DMYSQL_DATADIR=/data/mysqldata\ -DMYSQL_USER=mysql\ -DMYSQL_TCP_PORT=3306\ -DSYSCONFDIR=/etc\ -DWITH_SSL=yes 编译注解： -DCMAKE_INSTALL_PREFIX=/usr/local/mysql\ #安装目录 -DDEFAULT_CHARSET=utf8 \ #默认字符 -DDEFAULT_COLLATION=utf8_general_ci \ #校验字符 -DEXTRA_CHARSETS=all \ #安装所有扩展字符集 -DWITH_MYISAM_STORAGE_ENGINE=1 \ #安装myisam存储引擎 -DWITH_INNOBASE_STORAGE_ENGINE=1 \ #安装innodb存储引擎 -DWITH_ARCHIVE_STORAGE_ENGINE=1 \ #安装archive存储引擎 -DWITH_BLACKHOLE_STORAGE_ENGINE=1 \ #安装blackhole存储引擎 -DWITH_MEMORY_STORAGE_ENGINE=1 \ #安装memory存储引擎 -DWITH_FEDERATED_STORAGE_ENGINE=1 \ #安装frderated存储引擎 -DWITH_READLINE=1 \ #快捷键功能 -DENABLED_LOCAL_INFILE=1 \ #允许从本地导入数据 -DMYSQL_DATADIR=/data/mysqldata \ #数据库存放目录 -DMYSQL_USER=mysql \ #数据库属主 -DMYSQL_TCP_PORT=3306 \ #数据库端口 -DSYSCONFDIR=/etc \ #MySQL配辑文件 -DWITH_SSL=yes #数据库SSL make && make install 将mysql的库文件路径加入系统的库文件搜索路径中 利用ldconfig导入系统库 echo"/usr/local/mysql/lib" >> /etc/ld.so.conf.d/mysql.conf ldconfig 输出mysql的头文件到系统头文件 ln -s /usr/local/mysql/include/mysql/usr/include/mysql 进入安装路径,初始化配置脚本 cd /usr/local/mysql scripts/mysql_install_db --basedir=/usr/local/mysql --datadir=/data/mysqldata--user=mysql 复制mysql启动脚本到系统服务目录 cp/usr/local/mysql/support-files/mysql.server /etc/init.d/mysqld 系统启动项相关配置 chkconfig --add mysqld #添加开机启动服务 chkconfig mysqld on #设置mysql启动 配置mysql环境变量 vim /etc/profile export PATH=$PATH:/usr/local/mysql/bin 加载环境变量 source /etc/profile 2、配置mysql主从同步 Master：192.168.40.128 Host： master.testdb.com 防火墙添加规则，允许主从两机互相访问： iptables -I INPUT -s 192.168.40.129 -jACCEPT Slave：192.168.40.129 Host： slave.testdb.com 防火墙添加规则，允许主从两机互相访问： iptables -I INPUT -s 192.168.40.128 -jACCEPT 2.1修改mysql配置文件，配置GTID,创建同步账号 配置my.cnf Mysql5.6默认my.cnf在安装目录下/usr/local/mysql/ 根据服务器配置设置my.cnf内的参数,必须进行设置的为： \\主节点： [mysqld] binlog_format=row server-id = 1 datadir = /data/mysqldata innodb_file_per_table = 1 log-bin = /data/mysqlbinlog/master_bin.log log-slave-updates=true # 忽略mysql系统库复制 binlog-ignore-db=mysql binlog-ignore-db=information_schema binlog-ignore-db=performance_schema binlog-ignore-db=test replicate-ignore-db=test replicate-ignore-db=mysql replicate-ignore-db=information_schema replicate-ignore-db=performance_schema #创建function函数必须开启 log_bin_trust_function_creators=1 gtid-mode=on enforce-gtid-consistency=true master-info-repository=TABLE relay-log-info-repository=TABLE sync-master-info=1 slave-parallel-workers=4 binlog-checksum=CRC32 master-verify-checksum=1 slave-sql-verify-checksum=1 binlog-rows-query-log_events=1 report-port=3306 report-host=master.testdb.com \\从节点： [mysqld] binlog_format=row server-id = 2 datadir = /data/mysqldata innodb_file_per_table = 1 log-bin = /data/mysqlbinlog/slave_bin.log log-slave-updates=true # 忽略mysql系统库复制 binlog-ignore-db=mysql binlog-ignore-db=information_schema binlog-ignore-db=performance_schema binlog-ignore-db=test replicate-ignore-db=test replicate-ignore-db=mysql replicate-ignore-db=information_schema replicate-ignore-db=performance_schema #创建function函数必须开启 log_bin_trust_function_creators=1 gtid-mode=on enforce-gtid-consistency=true master-info-repository=TABLE relay-log-info-repository=TABLE sync-master-info=1 slave-parallel-workers=4 binlog-checksum=CRC32 master-verify-checksum=1 slave-sql-verify-checksum=1 binlog-rows-query-log_events=1 report-port=3306 report-host=slave.testdb.com 选项解析： datadir：数据文件存放路径 innodb_file_per_table = 1：独立表空间 log-bin = /data/mysqlbinlog/slave_bin.log：二进制日志存放路径 binlog-format：二进制日志的格式，有row、statement和mixed三种类型；需要注意的是：当设置隔离级别为READ-COMMITED必须设置二进制日志格式为ROW，现在MySQL官方认为STATEMENT这个已经不再适合继续使用；但mixed类型在默认的事务隔离级别下，可能会导致主从数据不一致； log-bin：启用二进制日志； server-id：同一个复制拓扑中的所有服务器的id号必须惟一； log-slave-updates：slave更新时是否记录到日志中； gtid-mode：指定GTID的类型，否则就是普通的复制架构 enforce-gtid-consistency：是否强制GTID的一致性 report-port：产生复制报告时在哪个端口上提供相关功能 report-host：产生复制报告时在哪个主机上提供相关功能，一般为自己的主机名 master-info-repository和relay-log-info-repository：启用此两项，可用于实现在崩溃时保证二进制及从服务器安全的功能； sync-master-info：启用之可确保服务器崩溃时无信息丢失； slave-paralles-workers：设定从服务器启动几个SQL复制线程数；0表示关闭多线程复制功能；数字太大也无意义最好与要复制的数据库的数目相同 binlog-checksum：复制时是否校验二进制文件的完整性等相关功能；binlog的校验格式校验算法（CRC32：循环冗余校验码32位） master-verify-checksum：检验主服务器二进制日志的相关功能 slave-sql-verify-checksum：校验从服务器中继日志的相关功能的 binlog-rows-query-log-events：启用之可用于在二进制日志详细记录事件相关的信息，可降低故障排除的复杂度； 启动mysql service mysqld start 创建sock链接 ln -s /var/lib/mysql/mysql.sock/tmp/mysql.sock 设置初始账户,并登陆后台 /usr/local/mysql/bin/mysqladmin-u root password 111111 #设置密码 登陆mysql Shell> mysql �Curoot �Cp 111111 mysql> grant all privileges on *.* toroot@'%' identified by '111111' with grant option; #给root用户非本地链接所有权限,并改密码和赋予其给其他人下发权限. mysql> flush privileges; #刷新 登入mysql，创建同步账号 shell> mysql �Curoot �Cp 111111 mysql> grant replicationclient,replication slave on *.* to 'repl'@'192.168.40.%' identified by 'mysql'; mysql> flush privileges; mysql> show global variables like'%UUID%'; +---------------+--------------------------------------+ | Variable_name | Value | +---------------+--------------------------------------+ | server_uuid | cecbb9e0-07e1-11e4-bb96-000c2923ee83 | +---------------+--------------------------------------+ 1 row in set (0.01 sec) 2.2开启主从同步 登入从服务器mysql shell> mysql �Curoot �Cp 111111 mysql> grant replication slave on *.* to'repl'@'192.168.40.%' identified by 'mysql'; mysql> flush privileges; 同步 mysql > stop slave; mysql > reset slave; mysql>CHANGE MASTER TO MASTER_HOST='192.168.40.128', MASTER_USER='repl',MASTER_PASSWORD='mysql', MASTER_AUTO_POSITION=1; mysql>show slave status \G *************************** 1. row*************************** Slave_IO_State: Waitingfor master to send event Master_Host: 192.168.40.128 Master_User: repl Master_Port: 3306 Connect_Retry: 60 Master_Log_File: master_bin.000002 Read_Master_Log_Pos: 1274 Relay_Log_File:slave-relay-bin.000002 Relay_Log_Pos: 550 Relay_Master_Log_File: master_bin.000002 Slave_IO_Running: Yes Slave_SQL_Running: Yes Replicate_Do_DB: Replicate_Ignore_DB: test,mysql,information_schema,performance_schema Replicate_Do_Table: Replicate_Ignore_Table: Replicate_Wild_Do_Table: Replicate_Wild_Ignore_Table: Last_Errno: 0 Last_Error: Skip_Counter: 0 Exec_Master_Log_Pos: 1274 Relay_Log_Space: 754 Until_Condition: None Until_Log_File: Until_Log_Pos: 0 Master_SSL_Allowed: No Master_SSL_CA_File: Master_SSL_CA_Path: Master_SSL_Cert: Master_SSL_Cipher: Master_SSL_Key: Seconds_Behind_Master: 0 Master_SSL_Verify_Server_Cert: No Last_IO_Errno: 0 Last_IO_Error: Last_SQL_Errno: 0 Last_SQL_Error: Replicate_Ignore_Server_Ids: Master_Server_Id: 1 Master_UUID:cecbb9e0-07e1-11e4-bb96-000c2923ee83 Master_Info_File: mysql.slave_master_info SQL_Delay: 0 SQL_Remaining_Delay: NULL Slave_SQL_Running_State: Slave has read all relay log; waiting for theslave I/O thread to update it Master_Retry_Count: 86400 Master_Bind: Last_IO_Error_Timestamp: Last_SQL_Error_Timestamp: Master_SSL_Crl: Master_SSL_Crlpath: Retrieved_Gtid_Set: cecbb9e0-07e1-11e4-bb96-000c2923ee83:7 Executed_Gtid_Set: cecbb9e0-07e1-11e4-bb96-000c2923ee83:1-7, f072628f-07e1-11e4-bb97-000c293e24ae:1-5 Auto_Position: 1 1 row in set (0.06 sec) 如果： Slave_IO_Running: Yes Slave_SQL_Running: Yes 则说明同步成功。 此时在master库上解锁表： mysql> unlock tables; 查看同步状态： ###########主节点：################ mysql> show processlist; +----+------+----------------------+------+------------------+------+-----------------------------------------------------------------------+------------------+ | Id | User | Host | db | Command | Time | State |Info | +----+------+----------------------+------+------------------+------+-----------------------------------------------------------------------+------------------+ | 3| root | localhost | NULL | Query | 0 | init | show processlist | | 4| repl | slave.gtid.com:43028 | NULL | Binlog Dump GTID | 259 | Master has sent all binlog to slave;waiting for binlog to be updated | NULL | +----+------+----------------------+------+------------------+------+-----------------------------------------------------------------------+------------------+ 2 rows in set (0.00 sec) ###########从节点：################ mysql> show processlist; +----+-------------+-----------+------+---------+------+-----------------------------------------------------------------------------+------------------+ | Id | User | Host | db | Command | Time | State |Info | +----+-------------+-----------+------+---------+------+-----------------------------------------------------------------------------+------------------+ | 1| root | localhost | NULL |Query | 0 | init | show processlist | | 8| system user | | NULL |Connect | 387 | Waiting for master tosend event | NULL | | 9| system user | | NULL |Connect | 387 | Slave has read all relaylog; waiting for the slave I/O thread to update it | NULL | | 10 | system user | | NULL | Connect | 387 | Waiting for an event fromCoordinator |NULL | | 11 | system user | | NULL | Connect | 387 | Waiting for an event fromCoordinator | NULL | | 12 | system user | | NULL | Connect | 387 | Waiting for an event fromCoordinator | NULL | | 13 | system user | | NULL | Connect | 431 | Waiting for an event fromCoordinator | NULL | +----+-------------+-----------+------+---------+------+-----------------------------------------------------------------------------+------------------+ 7 rows in set (0.05 sec) 测试： 可以在master数据库建立一个库和表，然后到slave上show databases进行验证。 2.3搭建mysql半复制(Semisynchronous Replication) 上面主从复制环境已经搭建成功，下面搭建mysql半复制(SemisynchronousReplication) //master上安装半同步复制插件 mysql> install plugin rpl_semi_sync_master soname 'semisync_master.so'; mysql> set global rpl_semi_sync_master_enabled=1; mysql> set global rpl_semi_sync_master_timeout=1000; mysql> show variables like '%semi%';+------------------------------------+-------+| Variable_name | Value |+------------------------------------+-------+| rpl_semi_sync_master_enabled | ON || rpl_semi_sync_master_timeout | 1000 || rpl_semi_sync_master_trace_level | 32 || rpl_semi_sync_master_wait_no_slave | ON |+------------------------------------+-------+4 rows in set (0.00 sec) 要想下次重启mysql自动生效,可以加入配置文件sehll> vim /etc/my.cnf[msyqld]rpl_semi_sync_master_enabled=1rpl_semi_sync_master_timeout=1000 //slave上安装插件 mysql> install pluginrpl_semi_sync_slave soname 'semisync_slave.so'; mysql> set globalrpl_semi_sync_slave_enabled = 1; mysql> stop slave; mysql> start slave; mysql> show variables like '%semi%'; +---------------------------------+-------+ | Variable_name | Value | +---------------------------------+-------+ | rpl_semi_sync_slave_enabled | ON | | rpl_semi_sync_slave_trace_level | 32 | +---------------------------------+-------+ 2 rows in set (0.00 sec) 同理，要想下次重启mysql自动生效,可以加入配置文件 sehll> vim /etc/my.cnf [msyqld] rpl_semi_sync_slave_enabled = 1 //在master服务器上测试半同步是否生效 mysql> show global status like 'rpl_semi%'; +--------------------------------------------+-------+ | Variable_name | Value | +--------------------------------------------+-------+ | Rpl_semi_sync_master_clients | 1 | |Rpl_semi_sync_master_net_avg_wait_time | 0 | | Rpl_semi_sync_master_net_wait_time | 0 | | Rpl_semi_sync_master_net_waits | 0 | | Rpl_semi_sync_master_no_times | 0 | | Rpl_semi_sync_master_no_tx | 0 | | Rpl_semi_sync_master_status | ON | |Rpl_semi_sync_master_timefunc_failures | 0 | |Rpl_semi_sync_master_tx_avg_wait_time | 0 | | Rpl_semi_sync_master_tx_wait_time | 0 | | Rpl_semi_sync_master_tx_waits | 0 | |Rpl_semi_sync_master_wait_pos_backtraverse | 0 | | Rpl_semi_sync_master_wait_sessions | 0 | | Rpl_semi_sync_master_yes_tx | 0 | +--------------------------------------------+-------+ 14 rows in set (0.00 sec) 2.4基于ssl的主从同步 1.主服务器上生成私钥 shell> cd /etc/pki/CA/ shell> touch index.txt shell> touch serial shell> echo "00" > serial shell> umask 077;openssl genrsa -out private/cakey.pem2048 Generating RSA private key, 2048 bit longmodulus ......................................+++ ................................+++ e is 65537 (0x10001) 2.主服务器上生成自签证书 shell> openssl req -new -x509 -key private/cakey.pem-out cacert.pem -days36500 3.主服务器上申请证书 shell> cd /usr/local/mysql/ shell> mkdirssl shell> cp /etc/pki/CA/cacert.pem /usr/local/mysql/ssl/ shell> umask 077; openssl genrsa -out/usr/local/mysql/ssl/master.key 2048 shell> openssl req -new -key/usr/local/mysql/ssl/master.key -out /usr/local/mysql/ssl/master.csr shell> opensslca -in /usr/local/mysql/ssl/master.csr -out /usr/local/mysql/ssl/master.crt -days 36500 shell> chown -Rmysql.mysql /usr/local/mysql/ssl 4.配置文件中添加ssl支持 shell> vim /usr/local/mysql/my.cnf [msyqld] ssl ssl-ca=/etc/pki/CA/cacert.pem ssl-cert=/usr/local/mysql/ssl/master.crt ssl-key=/usr/local/mysql/ssl/master.key 5.从服务器上 shell> cd /usr/local/mysql/ shell> mkdir ssl shell> umask 077; openssl genrsa -out/usr/local/mysql/ssl/mysql.key 2048 shell> openssl req -new -key /usr/local/mysql/ssl/mysql.key -out /usr/local/mysql/ssl/mysql.csr shell> chown -R mysql.mysql /usr/local/mysql/ssl/ 6.把证书传给主服务器，在主服务器上签署证书后在传给从服务器 shell> cd /usr/local/mysql/ssl shell> scp mysql.csr [email protected]:/etc/pki/CA 7.主服务器上签署证书 Shell> cd /etc/pki/CA/ shell> openssl ca -in mysql.csr -out mysql.crt -days 36500 8.传回给从服务器 shell> scp cacert.pem mysql.crt [email protected]:/usr/local/mysql/ssl 请确保主、从服务器中mysql路径下的ssl目录下的文件属主属组都是mysql shell> chown -R mysql.mysql /usr/local/mysql/ssl/ 9.配置从服务器my.cnf文件(证书路径建议不要在配置文件中指定，而是在change to命令中指定） 在my.cnf中添加ssl支持 [mysqld] ssl #添加这一行 10.主从均重启mysql服务 shell> service mysqld restart 主服务器登录mysql： mysql> show global variables like'%ssl%'; +---------------+---------------------------------+ | Variable_name | Value | +---------------+---------------------------------+ | have_openssl | DISABLED | | have_ssl | DISABLED | | ssl_ca | /etc/pki/CA/cacert.pem | | ssl_capath | | | ssl_cert | /usr/local/mysql/ssl/master.crt | | ssl_cipher | | | ssl_crl | | | ssl_crlpath | | | ssl_key | /usr/local/mysql/ssl/master.key | +---------------+---------------------------------+ 9 rows in set (0.07 sec) 11.主、从服务器上创建ssl链接用户 mysql> grant replication client,replication slave on*.* to "ssl"@"192.168.40.%" identified by 'mysql'; mysql> flush privileges; 12.从服务器上配置 mysql > stop slave; mysql > reset slave; mysql > change master to master_host='192.168.40.128', master_user='ssl', master_password='mysql',MASTER_AUTO_POSITION=1,master_ssl=1,master_ssl_ca='/usr/local/mysql/ssl/cacert.pem',master_ssl_cert='/usr/local/mysql/ssl/mysql.crt',master_ssl_key='/usr/local/mysql/ssl/mysql.key'; mysql> start slave; mysql> show slave status \G *************************** 1. row*************************** Slave_IO_State: Waiting formaster to send event Master_Host: 192.168.40.128 Master_User: ssl Master_Port: 3306 Connect_Retry: 60 Master_Log_File:master_bin.000003 Read_Master_Log_Pos: 734 Relay_Log_File:slave-relay-bin.000002 Relay_Log_Pos: 550 Relay_Master_Log_File: master_bin.000003 Slave_IO_Running: Yes Slave_SQL_Running: Yes Replicate_Do_DB: Replicate_Ignore_DB: test,mysql,information_schema,performance_schema Replicate_Do_Table: Replicate_Ignore_Table: Replicate_Wild_Do_Table: Replicate_Wild_Ignore_Table: Last_Errno: 0 Last_Error: Skip_Counter: 0 Exec_Master_Log_Pos: 734 Relay_Log_Space: 754 Until_Condition: None Until_Log_File: Until_Log_Pos: 0 Master_SSL_Allowed: Yes Master_SSL_CA_File: /usr/local/mysql/ssl/cacert.pem Master_SSL_CA_Path: Master_SSL_Cert:/usr/local/mysql/ssl/mysql.crt Master_SSL_Cipher: Master_SSL_Key:/usr/local/mysql/ssl/mysql.key Seconds_Behind_Master: 0 Master_SSL_Verify_Server_Cert: No Last_IO_Errno: 0 Last_IO_Error: Last_SQL_Errno: 0 Last_SQL_Error: Replicate_Ignore_Server_Ids: Master_Server_Id: 1 Master_UUID:cecbb9e0-07e1-11e4-bb96-000c2923ee83 Master_Info_File: mysql.slave_master_info SQL_Delay: 0 SQL_Remaining_Delay: NULL Slave_SQL_Running_State: Slave has read all relay log; waiting for theslave I/O thread to update it Master_Retry_Count: 86400 Master_Bind: Last_IO_Error_Timestamp: Last_SQL_Error_Timestamp: Master_SSL_Crl: Master_SSL_Crlpath: Retrieved_Gtid_Set: cecbb9e0-07e1-11e4-bb96-000c2923ee83:16 Executed_Gtid_Set: cecbb9e0-07e1-11e4-bb96-000c2923ee83:1-16, f072628f-07e1-11e4-bb97-000c293e24ae:1-7 Auto_Position: 1 1 row in set (0.00 sec) 3、搭建MHA环境 MHA介绍 MHA(MasterHigh Availability)是自动的master故障转移和Slave提升的软件包.它是基于标准的MySQL复制(异步/半同步). MHA有两部分组成:MHA Manager(管理节点)和MHA Node(数据节点). MHAManager可以单独部署在一台独立机器上管理多个master-slave集群,也可以部署在一台slave上.MHA Manager探测集群的node节点,当发现master出现故障的时候,它可以自动将具有最新数据的slave提升为新的master,然后将所有其 它的slave导向新的master上.整个故障转移过程对应用程序是透明的。 MHA node运行在每台MySQL服务器上(master/slave/manager),它通过监控具备解析和清理logs功能的脚本来加快故障转移的。 MHA节点包含三个脚本，依赖perl模块。 save_binary_logs:保存和复制当掉的主服务器二进制日志。 apply_diff_relay_logs:识别差异的relay log事件，并应用于其他salve服务器。 purge_relay_logs:清除relay log文件。 需要在所有mysql服务器上安装MHA节点，MHA管理服务器也需要安装。MHA管理节点模块内部依赖MHA节点模块。MHA管理节点通过ssh连接管理mysql服务器和执行MHA节点脚本。MHA节点依赖perl的DBD::mysql模块。 MHA工作原理 -从宕机崩溃的master保存二进制日志事件(binlogevents)。 -识别含有最新更新的slave。 -应用差异的中继日志(relay log)到其它slave。 -应用从master保存的二进制日志事件(binlogevents)。 -提升一个slave为新master。 -使其它的slave连接新的master进行复制。 1、安装MHA MHAmanager版本mha4mysql-manager-0.55-0.el6.noarch.rpm MHA节点版本mha4mysql-node-0.54-0.el6.noarch.rpm 主服务器安装： yum install -y perl-DBD-MySQL 从服务安装： yum install -y perl-DBD-MySQLperl-Config-Tiny perl-Log-Dispatch perl-Parallel-ForkManager perl-MIME-Lite perl-Params-Validate 安装MHA manager （目前规定manager部署在slave服务器上192.168.40.129） shell> yum install perl wget shell> yum install cpan shell> wgetftp://ftp.muug.mb.ca/mirror/centos/6.5/os/x86_64/Packages/compat-db43-4.3.29-15.el6.x86_64.rpm shell> wgethttp://downloads.naulinux.ru/pub/NauLinux/6x/i386/sites/School/RPMS/perl-Log-Dispatch-2.27-1.el6.noarch.rpm shell> wgethttp://dl.fedoraproject.org/pub/epel/6/i386/perl-Parallel-ForkManager-0.7.9-1.el6.noarch.rpm shell> wgethttp://dl.fedoraproject.org/pub/epel/6/i386/perl-Mail-Sender-0.8.16-3.el6.noarch.rpm shell> wgethttp://dl.fedoraproject.org/pub/epel/6/i386/perl-Mail-Sendmail-0.79-12.el6.noarch.rpm shell> wget http://mirror.centos.org/centos/6/os/x86_64/Packages/perl-Time-HiRes-1.9721-136.el6.x86_64.rpm shell> rpm -ivh perl-5.8.8-41.el5.x86_64.rpmcompat-db43-4.3.29-15.el6.x86_64.rpm perl-Log-Dispatch-2.27-1.el6.noarch.rpmperl-Parallel-ForkManager-0.7.9-1.el6.noarch.rpmperl-Mail-Sender-0.8.16-3.el6.noarch.rpmperl-Mail-Sendmail-0.79-12.el6.noarch.rpmperl-Time-HiRes-1.9721-136.el6.x86_64.rpm shell> rpm -ivhmha4mysql-manager-0.55-0.el6.noarch.rpm shell> rpm -ivhmha4mysql-manager-0.55-0.el6.noarch.rpm 2、建立ssh无密码登录环境 主mysql（192.168.40.128） ssh-keygen -t rsa ssh-copy-id -i .ssh/[email protected] 从mysql（192.168.40.129） ssh-keygen -t rsa ssh-copy-id -i .ssh/[email protected] ssh-copy-id -i .ssh/id_rsa.pub [email protected] 因为mha运行时需要用到/usr/bin下的两个mysql命令（貌似路径写死了，更改/etc/profile对mha无效），所以建立两个软连接 ln -s /usr/local/mysql/bin/mysqlbinlog/usr/bin/mysqlbinlog ln -s /usr/local/mysql/bin/mysql/usr/bin/mysql 从服务器的mysql要设置read_only=1 mysql> set global read_only=1; 3、配置MHA 1、在mha-manager上（本例是slave192.168.40.129上） shell> mkdir �Cp /usr/local/mha-manager/app1 shell> mkdir �Cpv /etc/mha-manager shell> vim /etc/mha-manager/app1.cnf [server default] manager_workdir=/usr/local/mha-manager/app1 manager_log=/usr/local/mha-manager/app1/manager.log user=root password=111111 ssh_user=root repl_user=repl repl_password=mysql remote_workdir=/usr/local/mha-manager/app1 ping_interval=1 # master_ip_failover_script=/etc/mha/scripts/master_ip_failover # shutdown_script=/etc/mha/scripts/power_manager #report_script=/etc/mha/scripts/send_report #master_ip_online_change_script=/etc/mha/scripts/master_ip_online_change [server1] hostname=192.168.40.128 master_binlog_dir=/data/mysqlbinlog candidate_master=1 [server2] hostname=192.168.40.129 master_binlog_dir=/data/mysqlbinlog candidate_master=1 2、masterha_check_ssh工具验证ssh信任登录是否成功 shell>masterha_check_ssh--conf=/etc/mha-manager/app1.cnf 注意：用ssh-keygen实现三台主机之间相互免密钥登录决定这一步是否成功。 Tue Jun 10 09:26:23 2014 - [warning] Global configurationfile /etc/masterha_default.cnf not found. Skipping. Tue Jun 10 09:26:23 2014 - [info] Reading applicationdefault configurations from /etc/mha-manager/app1.cnf.. Tue Jun 10 09:26:23 2014 - [info] Reading serverconfigurations from /etc/mha-manager/app1.cnf.. Tue Jun 10 09:26:23 2014 - [info] Starting SSH connectiontests.. Tue Jun 10 09:26:24 2014 - [debug] Tue Jun 10 09:26:23 2014 - [debug] Connecting via SSH [email protected](192.168.40.128:22) [email protected](192.168.40.129:22).. Tue Jun 10 09:26:24 2014 - [debug] ok. Tue Jun 10 09:26:24 2014 - [debug] Tue Jun 10 09:26:23 2014 - [debug] Connecting via SSH [email protected](192.168.40.129:22) [email protected](192.168.40.128:22).. Tue Jun 10 09:26:24 2014 - [debug] ok. Tue Jun 10 09:26:24 2014 - [info] All SSH connectiontests passed successfully. 成功！ 3、masterha_check_repl工具验证mysql复制是否成功 shell> masterha_check_repl--conf=/etc/mha-manager/app1.cnf Tue Jun 10 09:27:31 2014 - [info] Slaves settings checkdone. Tue Jun 10 09:27:31 2014 - [info] 192.168.40.128 (current master) +--192.168.40.129 Tue Jun 10 09:27:31 2014 - [info] Checking replicationhealth on 192.168.40.129.. Tue Jun 10 09:27:31 2014 - [info] ok. Tue Jun 10 09:27:31 2014 - [warning]master_ip_failover_script is not defined. Tue Jun 10 09:27:31 2014 - [warning] shutdown_script isnot defined. Tue Jun 10 09:27:31 2014 - [info] Got exit code 0 (Notmaster dead). MySQL Replication Health is OK. 4、启动MHA manager,并监控日志文件 shell> masterha_manager --conf=/etc/mha-manager/app1.cnf 后台运行可以使用 shell> nohup masterha_manager--conf=/etc/mha-manager/app1.cnf > /tmp/mha_manager.log 2>&1 shell> tail -f /usr/local/mha-manager/app1/manager.log //这里最好在新窗口执行 192.168.40.128 (current master) +--192.168.40.129 Tue Jun 10 09:29:04 2014 - [warning]master_ip_failover_script is not defined. Tue Jun 10 09:29:04 2014 - [warning] shutdown_script isnot defined. Tue Jun 10 09:29:04 2014 - [info] Set master pinginterval 1 seconds. Tue Jun 10 09:29:04 2014 - [warning]secondary_check_script is not defined. It is highly recommended setting it tocheck master reachability from two or more routes. Tue Jun 10 09:29:04 2014 - [info] Starting ping healthcheck on 192.168.40.128(192.168.40.128:3306).. Tue Jun 10 09:29:04 2014 - [info] Ping(SELECT) succeeded,waiting until MySQL doesn't respond.. 成功。 5、测试 [master--128] shell>service mysql stop [manager--129] shell>tail �Cf /usr/local/mha-manager/app1/manager.log -----日志显示如下------ ----- Failover Report ----- app1: MySQL Master failover 192.168.40.128 to192.168.40.129 succeeded Master 192.168.40.128 is down! Check MHA Manager logs atslave.testdb.com:/usr/local/mha-manager/app1/manager.log for details. Started automated(non-interactive) failover. The latest slave 192.168.40.129(192.168.40.129:3306) hasall relay logs for recovery. Selected 192.168.40.129 as a new master. 192.168.40.129: OK: Applying all logs succeeded. Generating relay diff files from the latest slavesucceeded. 192.168.40.129: Resetting slave info succeeded. Master failover to 192.168.40.129(192.168.40.129:3306)completed successfully. 转移后mha-manager服务器上的mha-manager进程自动跳出。原来的slave（129）上的mysql自动变成了mater。 6、故障转移后，用命令恢复原来的master 1)在旧master上执行 shell>service mysql start //数据库启动 shell>mysql -uroot �Cp111111 mysql> reset master; //清除binlog，重新记录binlog mysql> changemaster tomaster_host='192.168.40.129',master_user='ssl',master_password='mysql',master_log_file='slave_bin.000009',master_port=3306,master_log_pos=299, master_ssl=1,master_ssl_ca='/usr/local/mysql/ssl/cacert.pem',master_ssl_cert='/usr/local/mysql/ssl/mysql.crt',master_ssl_key='/usr/local/mysql/ssl/mysql.key'; mysql> start slave; #暂时先把旧master变为slave 2)然后在manager节点上： shell> masterha_master_switch--master_state=alive --conf=/etc/mha-manager/app1.cnf 此时master会转到128上，但是129的slave会断开，129需要重新设置为slave。 重新设置主从后，删除/usr/local/mha-manager/app1/app1.failover.complete文件，否则再次运行mha-manager后，如果主再出问题，会导致主从切换不成功。 恢复后在从（如果就两台服务器的话，这里指的是备主那台服务器）mysql里执行以下命令重新同步 mysql> stop slave; mysql> reset slave; mysql> change master to master_host='192.168.40.128',master_user='ssl',master_password='mysql',master_log_file='master_bin.000001',master_port=3306,master_log_pos=120,master_ssl=1,master_ssl_ca='/usr/local/mysql/ssl/cacert.pem',master_ssl_cert='/usr/local/mysql/ssl/mysql.crt',master_ssl_key='/usr/local/mysql/ssl/mysql.key'; mysql>start slave; mysql> show slave status; 同步设置好后,从服务器执行 mysql> set global read_only=1; 4、部署、安装keepalive 概述 keepalived介绍：Keepalived的作用是检测web服务器的状态，如果有一台web服务器死机，或工作出现故障，Keepalived将检测到，并将有故障的web 服务器从系统中剔除，当web服务器工作正常后Keepalived自动将web服务器加入到服务器群中，这些工作全部自动完成，不需要人工干涉，需要人工做的只是修复故障的web服务器。 1、keepalived安装 安装步骤两台机mysql-master、mysql-slave一样，如下： shell>wgethttp://www.keepalived.org/software/keepalived-1.2.12.tar.gz shell>tar xf keepalived-1.2.12.tar.gz shell>cd keepalived-1.2.12 shell>./configure --prefix=/usr/local/keepalived--with-kernel-dir=/usr/src/kernels/2.6.32-431.17.1.el6.x86_64 shell>make && make install 设置keepalived开机启动脚本 shell>cp/usr/local/keepalived/etc/rc.d/init.d/keepalived /etc/rc.d/init.d/ shell>cp/usr/local/keepalived/etc/sysconfig/keepalived /etc/sysconfig/ shell>cp/usr/local/keepalived/sbin/keepalived /usr/sbin/ shell>chkconfig keepalived off \\不能让keepalived开机自动启动 2、keepalived配置 1）主服务上（128）： shell>mkdir /etc/keepalived shell> vim/etc/keepalived/keepalived.conf global_defs { router_id mysql-master #修改为自己的主机名 notification_email { [email protected] #接收邮件，可以有多个，一行一个 } #当主、备份设备发生改变时，通过邮件通知 notification_email_from [email protected] #发送邮箱服务器 smtp_server 124.127.106.5 #发送邮箱超时时间 smtp_connect_timeout 30 } ##################第一部分################### vrrp_instance VI_1 { state BACKUP #都修改成BACKUP interface eth1 #绑定的网卡 virtual_router_id 60 #默认51 主从都修改为60 priority 100 #优先级，在mysql-slave上LVS上修改成80 advert_int 1 nopreempt #不抢占资源，意思就是它活了之后也不会再把主抢回来 authentication { # 认证方式，可以是PASS或AH两种认证方式 auth_typePASS # 认证密码 auth_pass 1111 } virtual_ipaddress { 192.168.40.130/24 #这可以增加多个VIP } } ##################第二部分################### virtual_server 192.168.40.130 3306 { delay_loop 6 lb_algo wrr lb_kind DR nat_mask 255.255.255.0 persistence_timeout 50 protocol TCP real_server 192.168.40.128 3306 { weight 1 notify_down /root/mysql_down.sh TCP_CHECK { connect_timeout 10 nb_get_retry 3 connect_port 3306 } } } 2）从服务器上（129）： shell>mkdir /etc/keepalived shell> vim/etc/keepalived/keepalived.conf global_defs { router_id mysql-master #修改为自己的主机名 notification_email { [email protected] #接收邮件，可以有多个，一行一个 } #当主、备份设备发生改变时，通过邮件通知 notification_email_from [email protected] #发送邮箱服务器 smtp_server 124.127.106.5 #发送邮箱超时时间 smtp_connect_timeout 30 } ##################第一部分################### vrrp_instance VI_1 { state BACKUP #都修改成BACKUP interface eth0 #绑定的网卡 virtual_router_id 60 #默认51 主从都修改为60 priority80 #优先级，在mysql-master上LVS上修改成100 advert_int 1 authentication { # 认证方式，可以是PASS或AH两种认证方式 auth_type PASS # 认证密码 auth_pass 1111 } virtual_ipaddress { 192.168.40.130/24 #这可以增加多个VIP } } ##################第二部分################### virtual_server 192.168.40.130 3306 { delay_loop 6 lb_algo wrr lb_kind DR nat_mask 255.255.255.0 persistence_timeout 50 protocol TCP real_server 192.168.40.129 3306 { weight 1 notify_down /root/mysql_down.sh TCP_CHECK { connect_timeout 10 nb_get_retry 3 connect_port 3306 } } } 3）myslq_down.sh配置，两台机都要配置这一步！！！ 这里需要注意的是，notify_down /root/mysql_down.sh这个选项，这个是是在keepalived检测不到mysql的时候要执行的脚本，从上面的配置文件来看real服务器只有本机。那么，keeaplived如果启动，客户端也只是访问本机的mysql。nopreempt这个选下也得注意，这个是不抢占资源在优先级高的机器上配置就可以。 看下这个脚本的内容： shell> vim/root/mysql_down.sh #!/bin/bash pkillkeepalived echo"VIP moved to 129 because of the 128-mysql is down !!!" |mail -s"VIP change warning from 128 !!!" [email protected] shell> chmod +x /root/mysql_down.sh #授权可执行权限 脚本内容就一条命令：pkill keepalived，主要作用是如果本机的mysql挂掉了，那么同时会杀死本机的keepalived，然后发送通知邮件（邮件smtp需要配置/etc/mail.rc），这样另外一台就会接替他工作，虚拟IP也会被另一台接管，如果不杀死keepalived虚拟IP不会被另一台接管，mysql访问也就不会切换过去。 注意：必须要先启动MYSQL,再启动keepalived，否则keepalived启动后会运行mysql_down.sh脚本，等于自杀。 4）主、从启动keepalived shell>service keepalived start 查看ip地址： shell> ip addr 或者简写 shell> ip a 3: eth1:<BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen1000 link/ether 00:0c:29:4a:f6:d3 brd ff:ff:ff:ff:ff:ff inet 192.168.40.128/24 brd 192.168.40.255 scope global eth1 inet 192.168.40.130/32 scope global eth1 inet6 fe80::20c:29ff:fe4a:f6d3/64 scope link valid_lft forever preferred_lft forever 由于master的keepalived的配置priority为100，比slave的80高，所以master抢到了VIP地址。 用VIP连接mysql进行测试： shell> mysql �Curoot �Ch 192.168.40.130�Cp111111 5）keepalived切换测试 1、 关闭master的keepalived服务，在slave服务器上查看地址，看到VIP已经切换到slave服务器上。 2、 启动master的keepalived服务，在slave服务器上查看地址，看到VIP仍然在slave服务器上，因为我们在master的keepalived的配置文件里设置了nopreempt（即优先级高的服务器不抢占资源）。 关闭slave的keepalived服务，在master服务器上查看地址，看到VIP已经切换到master服务器上。