Percona-XtraDB-Cluster+haproxy+keepalived 编译安装

系统：centos7

mysql 端口3306 Percona-XtraDB-Cluster 主节点

mysql端口3307 Percona-XtraDB-Cluster+haproxy+keepalived

MySQL端口3307 Percona-XtraDB-Cluster+haproxy+keepalived

/21 vip ip

软件版本：Percona-XtraDB-Cluster-- percona-xtrabackup- 

                  haproxy-   keepalived-

一、linux 系统限制配置

1、关闭系统防火墙  

systemctl stop  关闭防火墙systemctl disable   禁用防火墙

2、关闭SElinuxsed -i 's/SELINUX=.*/SELINUX=disabled/g' /etc/selinux/configsetenforce 0 selinux 立即生效

二、系统安装约定

源码包编译安装位置：/usr/local/软件名字

数据目录 /apps/data

三、下在源码安装包

下载 Percona-XtraDB-Cluster--

wget -P /usr/local/src /downloads/Percona-XtraDB-Cluster-57/Percona-XtraDB-Cluster--/rel-/source/tarball/Percona-XtraDB-Cluster--

下载percona-xtrabackup-

wget -P /usr/local/src /downloads/XtraBackup/Percona-XtraBackup-/source/tarball/percona-xtrabackup-

下载boost_1_59_0

mkdir -p /usr/local/boostwget -P /usr/local/boost /project/boost/boost//boost_1_59_

下载haproxy-

 wget -P /usr/local/src /download//src/haproxy-

下载keepalived-

wget -P /usr/local/src /software/keepalived-

下载automake-

wget -P /usr/local/src ftp:///gnu/automake/automake-

四、安装编译环境及编译依赖

1、yum -y install epel-release2、卸载旧版mysql或mariadbyum -y remove mariadb* mysql*3、安装编译环境yum -y install  libtool ncurses-devel libgcrypt-devel  libev-devel \git scons  gcc  gcc-c++  openssl  check  cmake  bison boost-devel  \asio-devel libaio-devel  ncurses-devel  readline-devel  pam-devel socat \libaio  automake autoconf vim redhat-lsb check-devel curl  curl-devel xinetd4、安装keepalived 依赖 （ ）yum -y install libnl-devel openssl-devel libnfnetlink-devel ipvsadm \popt-devel libnfnetlink kernel-devel popt-static iptraf

五、编译安装 Percona-XtraDB-Cluster-- percona-xtrabackup-

1、解压percona-xtrabackup-

tar -xvf /usr/local/src/percona-xtrabackup- -C /usr/local/src/

2、解压Percona-XtraDB-Cluster--

tar -xvf /usr/local/src/Percona-XtraDB-Cluster-- -C /usr/local/src/

3、编译percona-xtrabackup-

cd /usr/local/src/percona-xtrabackup-/cmake ./ -DBUILD_CONFIG=xtrabackup_release \         -DWITH_MAN_PAGES=OFF \         -DDOWNLOAD_BOOST=1 \         -DWITH_BOOST="/usr/local/boost" make -j4 && make installln -sf /usr/local/xtrabackup/bin/* /usr/sbin/

4、编译安装Percona-XtraDB-Cluster--

、创建mysql用户名及文件夹

mkdir /apps/{run,log}/mysqld -puseradd mysql -s /sbin/nologin -Mchown -R mysql:mysql /apps

、编译garbd,libgalera_

cd /usr/local/src/Percona-XtraDB-Cluster--cd "percona-xtradb-cluster-galera"revno值 cat GALERA-REVISIONscons -j4 psi=1 --config=force  revno="b98f92f"  boost_pool=0 libgalera_scons -j4 --config=force revno="b98f92f" garb/garbd创建pxc安装目录mkdir -p /usr/local/Percona-XtraDB-Cluster/{bin,lib}拷贝编译结果文件到pxc安装目录cp garb/garbd /usr/local/Percona-XtraDB-Cluster/bincp libgalera_ /usr/local/Percona-XtraDB-Cluster/lib

、编译 Percona-XtraDB-Cluster--

-DMYSQL_SERVER_SUFFIX 值得获取cd /usr/local/src/Percona-XtraDB-Cluster--WSREP_VERSION="$(grep WSREP_INTERFACE_VERSION wsrep/wsrep_ | cut -d '"' -f2).$(grep 'SET(WSREP_PATCH_VERSION'  "cmake/" | cut -d '"' -f2)"echo $WSREP_VERSION-DCOMPILATION_COMMENT 值获取cd /usr/local/src/Percona-XtraDB-Cluster--source VERSIONMYSQL_VERSION="$MYSQL_VERSION_MAJOR.$MYSQL_VERSION_MINOR.$MYSQL_VERSION_PATCH" echo $MYSQL_VERSION  REVISION="$(cd "$SOURCEDIR"; grep '^short: ' Docs/INFO_SRC |sed -e 's/short: //')" -DCOMPILATION_COMMENT 值 Percona XtraDB Cluster binary (GPL) $MYSQL_VERSION-$WSREP_VERSION Revision $REVISION 编译Percona-XtraDB-Cluster-- cd /usr/local/src/Percona-XtraDB-Cluster--cmake ./ -DBUILD_CONFIG=mysql_release \            -DCMAKE_BUILD_TYPE=RelWithDebInfo\            -DWITH_EMBEDDED_SERVER=OFF \            -DFEATURE_SET=community \            -DENABLE_DTRACE=OFF \            -DWITH_SSL=system -DWITH_ZLIB=system \            -DCMAKE_INSTALL_PREFIX="/usr/local/Percona-XtraDB-Cluster" \            -DMYSQL_DATADIR="/apps/data" \            -DMYSQL_SERVER_SUFFIX="" \            -DWITH_INNODB_DISALLOW_WRITES=ON \            -DWITH_WSREP=ON \            -DWITH_UNIT_TESTS=0 \            -DWITH_READLINE=system \            -DWITHOUT_TOKUDB=ON \    -DCOMPILATION_COMMENT="Percona XtraDB Cluster binary (GPL) -, Revision bec0879" \            -DWITH_PAM=ON \            -DWITH_INNODB_MEMCACHED=ON \            -DDOWNLOAD_BOOST=1 \            -DWITH_BOOST="/usr/local/boost" \            -DWITH_SCALABILITY_METRICS=ON   make -j4 && make install  拷贝文件到安装目录  cp -R /usr/local/src/Percona-XtraDB-Cluster--/percona-xtradb-cluster-tests /usr/local/Percona-XtraDB-Cluster/  创建软连接 mysql 系统环境  ln -sf /usr/local/Percona-XtraDB-Cluster/bin/* /usr/sbin/  拷贝启动文件到系统启动目录  cp -R /usr/local/Percona-XtraDB-Cluster/support-files/ /etc//mysqld  拷贝监控文件到xinetd管理目录  cp -R /usr/local/Percona-XtraDB-Cluster//mysqlchk /etc//  加入动态库  echo "/usr/local/Percona-XtraDB-Cluster/lib" >> /etc/  ldconfig

配置MySQL健康检查

vim /etc//mysqlchkserver          = /usr/bin/clustercheck修改为如果不修过可以cp /usr/local/Percona-XtraDB-Cluster/bin/clustercheck到 /usr/bin/server          = /usr/sbin/clustercheckvim /etc/services添加mysqlchk 9200/tcp # mysqlchk

配置

vim /etc/# Template  for PXC# Edit to your requirements.[client]socket=/apps/data/port=3306[mysqld]user=mysqlserver-id=10055datadir=/apps/datasocket=/apps/data/log-error=/apps/log/mysqld/pid-file=/apps/run/mysqld/log-binlog_slave_updatesexpire_logs_days=7port=3306character_set_server = utf8# Disabling symbolic-links is recommended to prevent assorted security riskssymbolic-links=0# Path to Galera librarywsrep_provider=/usr/local/Percona-XtraDB-Cluster/lib/libgalera_# Cluster connection URL contains IPs of nodes#If no IP is found, this implies that a new cluster needs to be created,#in order to do that you need to bootstrap this nodewsrep_cluster_address='gcomm://,,'#wsrep_cluster_address=gcomm://# In order for Galera to work correctly binlog format should be ROWbinlog_format=ROW# MyISAM storage engine has only experimental supportdefault_storage_engine=InnoDBwsrep_sst_receive_address=:6010wsrep_node_incoming_address=wsrep_provider_options = "_addr=tcp://;_addr=:6020;"# Slave thread to usewsrep_slave_threads= 8wsrep_log_conflicts# This changes how InnoDB autoincrement locks are managed and is a requirement for Galerainnodb_autoinc_lock_mode=2# Node IP addresswsrep_node_address=# Cluster namewsrep_cluster_name=pxc-cluster#If wsrep_node_name is not specified,  then system hostname will be usedwsrep_node_name=pxc-cluster-node-1#pxc_strict_mode allowed values: DISABLED,PERMISSIVE,ENFORCING,MASTERpxc_strict_mode=ENFORCING# SST methodwsrep_sst_method=xtrabackup-v2#wsrep_sst_method=rsync#Authentication for SST method # 修改自己的创建的账号、密码wsrep_sst_auth="sstuser:456789"

vim /etc/# Template  for PXC# Edit to your requirements.[client]socket=/apps/data/port=3307[mysqld]user=mysqlserver-id=10056datadir=/apps/datasocket=/apps/data/log-error=/apps/log/mysqld/pid-file=/apps/run/mysqld/log-binlog_slave_updatesexpire_logs_days=7port=3307character_set_server = utf8# Disabling symbolic-links is recommended to prevent assorted security riskssymbolic-links=0# Path to Galera librarywsrep_provider=/usr/local/Percona-XtraDB-Cluster/lib/libgalera_# Cluster connection URL contains IPs of nodes#If no IP is found, this implies that a new cluster needs to be created,#in order to do that you need to bootstrap this nodewsrep_cluster_address='gcomm://,,'#wsrep_cluster_address=gcomm://# In order for Galera to work correctly binlog format should be ROWbinlog_format=ROW# MyISAM storage engine has only experimental supportdefault_storage_engine=InnoDBwsrep_sst_receive_address=:6010wsrep_node_incoming_address=wsrep_provider_options = "_addr=tcp://;_addr=:6020;"# Slave thread to usewsrep_slave_threads= 8wsrep_log_conflicts# This changes how InnoDB autoincrement locks are managed and is a requirement for Galerainnodb_autoinc_lock_mode=2# Node IP addresswsrep_node_address=# Cluster namewsrep_cluster_name=pxc-cluster#If wsrep_node_name is not specified,  then system hostname will be usedwsrep_node_name=pxc-cluster-node-2#pxc_strict_mode allowed values: DISABLED,PERMISSIVE,ENFORCING,MASTERpxc_strict_mode=ENFORCING# SST methodwsrep_sst_method=xtrabackup-v2#wsrep_sst_method=rsync#Authentication for SST method # 修改自己的创建的账号、密码wsrep_sst_auth="sstuser:456789"

vim /etc/# Template  for PXC# Edit to your requirements.[client]socket=/apps/data/port=3307[mysqld]user=mysqlserver-id=10057datadir=/apps/datasocket=/apps/data/log-error=/apps/log/mysqld/pid-file=/apps/run/mysqld/log-binlog_slave_updatesexpire_logs_days=7port=3307character_set_server = utf8# Disabling symbolic-links is recommended to prevent assorted security riskssymbolic-links=0# Path to Galera librarywsrep_provider=/usr/local/Percona-XtraDB-Cluster/lib/libgalera_# Cluster connection URL contains IPs of nodes#If no IP is found, this implies that a new cluster needs to be created,#in order to do that you need to bootstrap this nodewsrep_cluster_address='gcomm://,,'#wsrep_cluster_address=gcomm://# In order for Galera to work correctly binlog format should be ROWbinlog_format=ROW# MyISAM storage engine has only experimental supportdefault_storage_engine=InnoDBwsrep_sst_receive_address=:6010wsrep_node_incoming_address=wsrep_provider_options = "_addr=tcp://;_addr=:6020;"# Slave thread to usewsrep_slave_threads= 8wsrep_log_conflicts# This changes how InnoDB autoincrement locks are managed and is a requirement for Galerainnodb_autoinc_lock_mode=2# Node IP addresswsrep_node_address=# Cluster namewsrep_cluster_name=pxc-cluster#If wsrep_node_name is not specified,  then system hostname will be usedwsrep_node_name=pxc-cluster-node-3#pxc_strict_mode allowed values: DISABLED,PERMISSIVE,ENFORCING,MASTERpxc_strict_mode=ENFORCING# SST methodwsrep_sst_method=xtrabackup-v2#wsrep_sst_method=rsync#Authentication for SST methodwsrep_sst_auth="sstuser:456789" # 修改自己的创建的账号、密码

初始化数据库

初始化数据mysqld --initialize --user=mysql  --datadir="/apps/data"启动数据库/etc//mysqld start --wsrep-cluster-address="gcomm://"关闭数据库 以后都要密码才能进去/etc//mysqld stop安全模式期待MySQL修改密码mysqld_safe --wsrep-cluster-address=gcomm://  --user=mysql --skip-grant-tables --skip-networking &use mysql；update user set authentication_string=Password('123456') where user='root';flush privileges;shutdownexit 退出MySQLps -ef | grep mysql直到MySQL完全关闭 再次启动MySQL/etc//mysqld start --wsrep-cluster-address="gcomm://"进入MySQL 创建同步账号及修改root密码mysql -u root -p123456创建MySQL同步账号及密码GRANT ALL PRIVILEGES  ON *.* TO sstuser@'%' IDENTIFIED BY '456789'; FLUSH PRIVILEGES;GRANT ALL PRIVILEGES  ON *.* TO sstuser@'localhost' IDENTIFIED BY '456789'; FLUSH PRIVILEGES;修改root 远程能够连接GRANT ALL PRIVILEGES ON *.* TO root@'%' IDENTIFIED BY '123456'; FLUSH PRIVILEGES;创建健康检查数据库账号密码GRANT PROCESS ON *.* TO 'clustercheckuser'@'localhost' IDENTIFIED BY 'clustercheckpassword!';如果使用自己账号密码请修改vim /usr/sbin/clustercheck 对应的账号密码  其它节点直接启动service mysqld start  重启第一个节点  service mysqld restart 出现错误 /etc//mysqld bootstrap-pxc 进行启动  启动完成登陆MySQL mysql -u root -p123456 查看MySQL 节点数 mysql> show global status like 'wsrep_cluster_size';+--------------------+-------+| Variable_name      | Value |+--------------------+-------+| wsrep_cluster_size | 3     |+--------------------+-------+1 row in set ( sec)mysql> show global status like 'wsrep_incoming_addresses';+--------------------------+-------------------------------------------------------+| Variable_name            | Value                                                 |+--------------------------+-------------------------------------------------------+| wsrep_incoming_addresses | :3307,:3307,:3306 |+--------------------------+-------------------------------------------------------+1 row in set ( sec)创建测试数据看看是否同步mysql> create database test;Query OK, 1 row affected ( sec)

数据库健康检查启动

systemctl start  查看9200端口是否打开netstat -anlp | grep 9200tcp6       0      0 :::9200                 :::*                    LISTEN      73323/xinetd 测试端口是否正常telnet  9200Escape character is '^]'.HTTP/ 200 OKContent-Type: text/plainConnection: closeContent-Length: 40Percona XtraDB Cluster Node is synced.Connection closed by foreign host.返回503 首先执行 /usr/sbin/clustercheck是否返回200 如果不是检查链接数据库的账号密码返回200正常 就请修改 vim /etc//mysqlchkuser root重启systemctl restart  再次测试如果返回200正常就ok 还是报错就请查看系统日志修复

六、编译安装 haproxy (,)

安装 haproxy 日志记录支持yum -y install rsyslog解压 haproxy- tar -xvf /usr/local/src/haproxy- -C /usr/local/src/cd /usr/local/src/haproxy-make TARGET=linux31 PREFIX=/usr/local/haproxymake install PREFIX=/usr/local/haproxy创建文件夹haproxymkdir -pv /usr/local/haproxy/{conf,run,log}

创建haproxy 启动脚本vim /etc//haproxy

#! /bin/bash# chkconfig: - 85 15# description: haproxy is a World Wide Web server. It is used to servePROGDIR=/usr/local/haproxyPROGNAME=haproxyDAEMON=$PROGDIR/sbin/$PROGNAMECONFIG=$PROGDIR/conf/$PIDFILE=$PROGDIR/run/$DESC="HAProxy daemon"SCRIPTNAME=/usr/local/haproxy//$PROGNAME# Gracefully exit if the package has been removed.test -x $DAEMON || exit 0start() {    echo -n "Starting $DESC: $PROGNAME"   $DAEMON -f $CONFIG   echo "."        }stop() {  echo -n "Stopping $DESC: $PROGNAME"    cat $PIDFILE | xargs kill    echo "."}reload() { echo -n "reloading $DESC: $PROGNAME"   $DAEMON -f $CONFIG -p $PIDFILE -sf $(cat $PIDFILE)}case "$1" in  start)  start  ;; stop)  stop   ;;  restart)   stop   start    ;;  reload) reload   ;;*) echo "Usage: $SCRIPTNAME {start|stop|restart|reload}" >&2 exit 1 ;;esacexit 0

chmod +x  /etc//haproxy # 可执行权限

创建

vim /usr/local/haproxy/conf/

 global                log    local0                log   local1 notice                #log loghost    local0 info                maxconn 50000                chroot /usr/local/haproxy                uid 99                gid 99                daemon                nbproc 2                pidfile /usr/local/haproxy/run/                #debug                #quiet defaults                log     global                mode    tcp                option  tcplog                option  dontlognull                option  forwardfor                option  redispatch                retries 3                timeout connect     3000                timeout client      50000                timeout server 50000frontend admin_stat         bind *:8888         mode http         default_backend stats-backfrontend pxc-front         bind    *:3306         mode    tcp         default_backend pxc-backfrontend pxc-onenode-front         bind *:3308         mode tcp         default_backend pxc-onenodebackend pxc-back        mode    tcp        balance roundrobin        option httpchk        server node1 :3306 check port 9200 inter 12000 rise 3 fall 3        server node2 :3307 check port 9200 inter 12000 rise 3 fall 3        server node3 :3307 check port 9200 inter 12000 rise 3 fall 3backend pxc-onenode        mode tcp        balance roundrobin        option httpchk        server node1 :3306 check port 9200 inter 12000 rise 3 fall 3        server node2 :3307 check port 9200 inter 12000 rise 3 fall 3  backup        server node3 :3307 check port 9200 inter 12000 rise 3 fall 3  backupbackend stats-back        mode http        balance roundrobin        stats uri /admin?stats        stats auth admin:admin        stats realm Haproxy\ Statistics

Haproxy rsyslog 日志配置vim /etc/ #$ModLoad imudp      修改 $ModLoad imudp #$UDPServerRun 514  修改 $UDPServerRun 514#添加 Haproxy  日志记录路径local0.* /usr/local/haproxy/log/

重启rsyslogsystemctl restart  启动Haproxy service haproxy start加入开机启动chkconfig haproxy on查看haproxy  是否启动ps -ef | grep haproxy haproxy 访问端口8888 账号密码 admin:admin:8888/admin?stats:8888/admin?stats

七、编译安装keepalived

安装automake-tar -xvf /usr/local/src/automake- -C /usr/local/srccd /usr/local/src/automake-./configure && make && make install

编译keepalived

tar -xvf /usr/local/src/keepalived- -C /usr/local/srccd /usr/local/src/keepalived-./configure --prefix=/usr/local/keepalivedvmware 虚拟机安装 修改./lib/HAVE_DECL_RTA_ENCAP 1 修改为HAVE_DECL_RTA_ENCAP 0 如果不修改编译会报错make && make install

拷贝keepalived 可执行文件到/usr/sbin/cp /usr/local/keepalived/sbin/keepalived /usr/sbin/

创建keepalived 启动脚本

vim /etc///keepalived

#!/bin/sh## Startup script for the Keepalived daemon## processname: keepalived# pidfile: /var/run/# config: /etc/keepalived/# chkconfig: - 21 79# description: Start and stop Keepalived# Source function library. /etc///functions# Source configuration file (we set KEEPALIVED_OPTIONS there). /etc/sysconfig/keepalivedRETVAL=0prog="keepalived"start() {    echo -n $"Starting $prog: "    daemon keepalived ${KEEPALIVED_OPTIONS}    RETVAL=$?    echo    [ $RETVAL -eq 0 ] && touch /var/lock/subsys/$prog}stop() {    echo -n $"Stopping $prog: "    killproc keepalived    RETVAL=$?    echo    [ $RETVAL -eq 0 ] && rm -f /var/lock/subsys/$prog}reload() {    echo -n $"Reloading $prog: "    killproc keepalived -1    RETVAL=$?    echo}# See how we were called.case "$1" in    start)        start        ;;    stop)        stop        ;;    reload)        ;;    restart)        stop        start        ;;    condrestart)        if [ -f /var/lock/subsys/$prog ]; then            stop            start        fi        ;;    status)        status keepalived        RETVAL=$?        ;;    *)        echo "Usage: $0 {start|stop|reload|restart|condrestart|status}"        RETVAL=1esacexit $RETVAL

chmod +x /etc///keepalived #keepalived 可执行

配置keepalived 

cp /usr/local/keepalived/etc/sysconfig/keepalived  /etc/sysconfig/创建/etc/keepalived 文件夹mkdir -p /etc/keepalived

创建/etc/keepalived/ （）

vim /etc/keepalived/! Configuration File for keepalivedglobal_defs {   notification_email {        admin@    #发生故障时，接受信息的email地址   }   notification_email_from admin@   smtp_server    smtp_connect_timeout 30   router_id LVS_DEVEL}vrrp_script chk_haproxy {                           #自定义的监控脚本    script "/etc/keepalived/chk_"    interval 2    weight 2}vrrp_instance VI_1 {    state BACKUP                                         #让VIP不抢回    nopreempt    interface eth0                                       #监听接口    virtual_router_id 51    priority 100                                            #优先级，backup机器上的优先级要小与这个值    advert_int 1                                          #检查间隔    authentication {        auth_type PASS        auth_pass 1111    }    track_script {                                        #自定义的监控脚本        chk_haproxy    }    virtual_ipaddress {                              #VIP地址，可以设多个        /21    }    notify_backup "/etc//haproxy restart"    notify_fault "/etc//haproxy stop"}

创建/etc/keepalived/ （）

! Configuration File for keepalivedglobal_defs {   notification_email {        admin@    #发生故障时，接受信息的email地址   }   notification_email_from admin@   smtp_server    smtp_connect_timeout 30   router_id LVS_DEVEL}vrrp_script chk_haproxy {                           #自定义的监控脚本    script "/etc/keepalived/chk_"    interval 2    weight 2}vrrp_instance VI_1 {    state BACKUP                                         #让VIP不抢回    nopreempt    interface eth0                                       #监听接口    virtual_router_id 51    priority 99                                            #优先级，backup机器上的优先级要小与这个值    advert_int 1                                          #检查间隔    authentication {        auth_type PASS        auth_pass 1111    }    track_script {                                        #自定义的监控脚本        chk_haproxy    }    virtual_ipaddress {                              #VIP地址，可以设多个        /21    }    notify_backup "/etc//haproxy restart"    notify_fault "/etc//haproxy stop"}

创建/etc/keepalived/chk_

vim /etc/keepalived/chk_#!/bin/bashif [ $(ps -C haproxy --no-header | wc -l) -eq 0 ];then    /etc//haproxy startfisleep 2if [ $(ps -C haproxy --no-header | wc -l) -eq 0 ]; then    /etc//keepalived stopfi

chmod +x /etc/keepalived/chk_ #可执行权限

启动 keepalived

service keepalived start ip a1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN     link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00    inet /8 scope host lo       valid_lft forever preferred_lft forever    inet6 ::1/128 scope host        valid_lft forever preferred_lft forever2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen 1000    link/ether 00:0c:29:1b:50:a3 brd ff:ff:ff:ff:ff:ff    inet /21 brd  scope global dynamic eth0       valid_lft 44444sec preferred_lft 44444sec    inet /21 scope global secondary eth0       valid_lft forever preferred_lft forever    inet6 fe80::20c:29ff:fe1b:50a3/64 scope link        valid_lft forever preferred_lft forever

service keepalived start 1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN     link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00    inet /8 scope host lo       valid_lft forever preferred_lft forever    inet6 ::1/128 scope host        valid_lft forever preferred_lft forever2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen 1000    link/ether 00:0c:29:63:ae:85 brd ff:ff:ff:ff:ff:ff    inet /21 brd  scope global dynamic eth0       valid_lft 58142sec preferred_lft 58142sec    inet6 fe80::20c:29ff:fe63:ae85/64 scope link        valid_lft forever preferred_lft forever

关闭  keepalived 进程

service keepalived stop[root@56~]# ip a1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN     link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00    inet /8 scope host lo       valid_lft forever preferred_lft forever    inet6 ::1/128 scope host        valid_lft forever preferred_lft forever2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen 1000    link/ether 00:0c:29:1b:50:a3 brd ff:ff:ff:ff:ff:ff    inet /21 brd  scope global dynamic eth0       valid_lft 44258sec preferred_lft 44258sec    inet6 fe80::20c:29ff:fe1b:50a3/64 scope link        valid_lft forever preferred_lft forever

查看

[root@57~]# ip a 1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN     link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00    inet /8 scope host lo       valid_lft forever preferred_lft forever    inet6 ::1/128 scope host        valid_lft forever preferred_lft forever2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen 1000    link/ether 00:0c:29:63:ae:85 brd ff:ff:ff:ff:ff:ff    inet /21 brd  scope global dynamic eth0       valid_lft 57869sec preferred_lft 57869sec    inet /21 scope global secondary eth0       valid_lft forever preferred_lft forever    inet6 fe80::20c:29ff:fe63:ae85/64 scope link        valid_lft forever preferred_lft forever

其它 keepalived  两个都配置为BACKUP 当第一个出现故障的时候恢复了不用抢占正在提供服务的IP 服务不会闪断。     haproxy 配置双端口链接 3306 是同时负载3台服务器 3308 只是一台使用当第一台挂掉才会使用第二台。