15、 Heartbeat+DRBD+MySQL高可用架构方案与实施过程细节
参考自：http://oldboy.blog.51cto.com/2561410/1240412
|
heartbeat和keepalived应用场景及区别 |
很多网友说为什么不使用keepalived而使用长期不更新的heartbeat，下面说一下它们之间的应用场景及区别： 1、对于web，db，负载均衡(lvs,haproxy,nginx)等，heartbeat和keepalived都可以实现 2、lvs最好和keepalived结合，因为keepalived最初就是为lvs产生的，(heartbeat没有对RS的健康检查功能，heartbeat可以通过ldircetord来进行健康检查的功能) 3、mysql双主多从，NFS/MFS存储，他们的特点是需要数据同步，这样的业务最好使用heartbeat，因为heartbeat有自带的drbd脚本 总结： 无数据同步的应用程序高可用可选择keepalived， 有数据同步的应用程序高可用可选择heartbeat（DRBD） 。 |
|
1、 安装部署准备 |
（1）架构拓扑
架构说明： 一主多从最常用的架构，多个从库可以使用lvs来提供读的负载均衡。 解决一主单点的问题，当主库宕机后，可以实现主库宕机后备节点自动接管，所有的从库会自动和新的主库进行同步，实现了mysql主库的热备方案 （2）系统环境：
（3）部署环境
（4）主库服务器数据分区信息
|
|
2、heatbeat安装部署 |
(1)、配置服务器间心跳连接路由 主节点 [[email protected] ~]# route add -host 172.16.4.3 dev eth2<==到对端的心跳路由 [[email protected] ~]# route add -host 172.168.4.3 dev eth3<==到对端的DRBD数据路由 备节点 [[email protected] ~]# route add -host 172.16.4.2 dev eth2 [[email protected] ~]# route add -host 172.168.4.2 dev eth3 (2)、安装heartbeat [[email protected] ~]# yum install heartbeat -y [[email protected] ~]# yum install heartbeat -y 提示：需要执行两遍安装heartbeat操作 (3)、配置heartbeat 主备节点两端的配置文件(ha.cfauthkeysharesources)完全相同 1)、ha.cf [[email protected] ~]# vim /etc/ha.d/ha.cf #log configure debugfile /var/log/ha-debug logfile /var/log/ha-log logfacility local1 #options configure keepalive 2 deadtime 30 warntime 10 initdead 120 #bcast eth2 mcast eth2 225.0.0.7 694 1 0 #node configure auto_failback on node master1 <==主节点主机名 node master2 <==备节点主机名 crm no 2)、配置authkeys [[email protected] ~]# vim /etc/ha.d/authkeys auth 1 1 sha1 47e9336850f1db6fa58bc470bc9b7810eb397f04 3)、配置haresources [[email protected] ~]# vim /etc/ha.d/haresources master1 IPaddr::192.168.4.1/16/eth1 #master1 IPaddr::192.168.4.1/16/eth1 drbddisk::data Filesystem::/dev/drbd1::/data::ext3 mysqld 说明： drbddisk::data <==启动drbd data资源，相当于执行/etc/ha.d/resource.d/drbddisk data stop/start操作 Filesystem::/dev/drbd1::/data::ext3 <==drbd分区挂载到/data目录，相当于执行/etc/ha.d/resource.d/Filesystem /dev/drbd1 /data ext3 stop/start <==相当于系统中执行mount /dev/drbd1 /data mysql <==启动mysql服务脚本，相当于/etc/init.d/mysql stop/start (4)、启动heartbeat [[email protected] ~]# /etc/init.d/heartbeat start [[email protected] ~]# chkconfig heartbeat off 说明：关闭开机自启动，当服务器重启时，需要人工去启动 (5)、测试heartbeat 正常状态 [[email protected]master1 ~]# ip addr|grep eth1 3: eth1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast qlen 1000 inet 192.168.4.2/16 brd 192.168.255.255 scope global eth1 inet 192.168.4.1/16 brd 192.168.255.255 scope global secondary eth1:0 [[email protected] ~]# ip addr|grep eth1 3: eth1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast qlen 1000 inet 192.168.4.3/16 brd 192.168.255.255 scope global eth1 说明：master1节点拥有vip地址，master2节点没有 模拟主节点宕机后的状态 [[email protected] ~]# /etc/init.d/heartbeat stop [[email protected]master2 ~]# ip addr|grep eth1 3: eth1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast qlen 1000 inet 192.168.4.3/16 brd 192.168.255.255 scope global eth1 inet 192.168.4.1/16 brd 192.168.255.255 scope global secondary eth1:0 说明：master1宕机后，vip地址漂移到master2节点上，master2成为主节点 模拟主节点故障恢复后的状态 [[email protected] ~]# /etc/init.d/heartbeat start [[email protected]master1 ~]# ip addr|grep eth1 3: eth1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast qlen 1000 inet 192.168.4.2/16 brd 192.168.255.255 scope global eth1 inet 192.168.4.1/16 brd 192.168.255.255 scope global secondary eth1:0 说明：master1抢占vip资源 |
|
3、DRBD 安装部署 |
(1)、新添加硬盘 [[email protected] ~]# fdisk /dev/sdb #sdb磁盘分两个分区sdb1和sdb2 [[email protected] ~]# partprobe [[email protected] ~]# mkfs.ext3 /dev/sdb1 #sdb2分区为meta data分区，不需要格式化操作 [[email protected] ~]# tune2fs -c -1 /dev/sdb1 #设置最大挂载数为-1 (2)、安装DRBD [[email protected] ~]# yum install kmod-drbd83 drbd83 -y [[email protected] ~]# modprobe drbd #注意：不要设置echo ‘modprobe drbd‘ >>/etc/rc.loca开机自动加载drbd模块，如果drbd服务是开机自启动的，会先启动drbd服务在加载drbd的顺序，导致drbd启动不了出现的问题 (3)、配置DRBD 主备节点两端配置文件完全一致 [[email protected] ~]# cat /etc/drbd.conf global { # minor-count 64; # dialog-refresh 5; # 5 seconds # disable-ip-verification; usage-count no; } common { protocol C; disk { on-io-error detach; #size 454G; no-disk-flushes; no-md-flushes; } net { sndbuf-size 512k; # timeout 60; # 6 seconds (unit = 0.1 seconds) # connect-int 10; # 10 seconds (unit = 1 second) # ping-int 10; # 10 seconds (unit = 1 second) # ping-timeout 5; # 500 ms (unit = 0.1 seconds) max-buffers 8000; unplug-watermark 1024; max-epoch-size 8000; # ko-count 4; # allow-two-primaries; cram-hmac-alg "sha1"; shared-secret "hdhwXes23sYEhart8t"; after-sb-0pri disconnect; after-sb-1pri disconnect; after-sb-2pri disconnect; rr-conflict disconnect; # data-integrity-alg "md5"; # no-tcp-cork; } syncer { rate 120M; al-extents 517; } } resource data { on master1 { device /dev/drbd1; disk /dev/sdb1; address 192.168.4.2:7788; meta-disk /dev/sdb2 [0]; } on master2 { device /dev/drbd1; disk /dev/sdb1; address 192.168.4.3:7788; meta-disk /dev/sdb2 [0]; } } (4)、初始化meta分区 [[email protected] ~]# drbdadm create-md data Writing meta data... initializing activity log NOT initialized bitmap New drbd meta data block successfully created. (5)、初始化设备同步(覆盖备节点，保持数据一致) [[email protected] ~]# drbdadm -- --overwrite-data-of-peer primary data (6)、启动drbd [[email protected] ~]# drbdadm up all [[email protected] ~]# chkconfig drbd off (7)、挂载drbd分区到data数据目录 [[email protected] ~]# drbdadm primary all [[email protected] ~]# mount /dev/drbd1 /data #说明：/data目录为数据库的数据目录 (8)、测试DRBD 正常状态 [[email protected] ~]# cat /proc/drbd version: 8.3.13 (api:88/proto:86-96) GIT-hash: 83ca112086600faacab2f157bc5a9324f7bd7f77 build by [email protected], 2012-05-07 11:56:36 1: cs:Connected ro:Primary/Secondary ds:UpToDate/UpToDate C r----- ns:497984 nr:0 dw:1 dr:498116 al:1 bm:31 lo:0 pe:0 ua:0 ap:0 ep:1 wo:b oos:0 [[email protected] ~]# cat /proc/drbd version: 8.3.13 (api:88/proto:86-96) GIT-hash: 83ca112086600faacab2f157bc5a9324f7bd7f77 build by [email protected], 2012-05-07 11:56:36 1: cs:Connected ro:Secondary/Primary ds:UpToDate/UpToDate C r----- ns:0 nr:497984 dw:497984 dr:0 al:0 bm:30 lo:0 pe:0 ua:0 ap:0 ep:1 wo:b oos:0 #说明：master1为主节点，master2为备节点 模拟master1宕机 [[email protected] ~]# umount /dev/drbd1 [[email protected] ~]# drbdadm down all [[email protected] ~]# cat /proc/drbd version: 8.3.13 (api:88/proto:86-96) GIT-hash: 83ca112086600faacab2f157bc5a9324f7bd7f77 build by [email protected], 2012-05-07 11:56:36 1: cs:WFConnection ro:Secondary/Unknown ds:UpToDate/DUnknown C r----- ns:0 nr:497985 dw:497985 dr:0 al:0 bm:30 lo:0 pe:0 ua:0 ap:0 ep:1 wo:b oos:0 [[email protected] ~]# drbdadm primary all [[email protected] ~]# mount /dev/drbd1 /data [[email protected] ~]# df -h Filesystem Size Used Avail Use% Mounted on /dev/sda3 19G 5.1G 13G 29% / /dev/sda1 190M 18M 163M 10% /boot tmpfs 60M 0 60M 0% /dev/shm /dev/drbd1 471M 11M 437M 3% /data #说明：master1宕机后，master2可以升级为主节点，可挂载drbd分区继续使用 |
|
4、MySQL 安装部署 |
注意：三台数据库都安装mysql服务，master2只安装到makeinstall即可，mysqld服务不要设置为开机自启动 (1)、解决perl编译问题 echo ‘export LC_ALL=C‘>> /etc/profile source /etc/profile (2)、安装CAMKE cd /home/xu/tools wget http://www.cmake.org/files/v2.8/cmake-2.8.4.tar.gz tar zxf cmake-2.8.4.tar.gz cd cmake-2.8.4 ./configure make & make install (3)、创建用户 groupadd mysql useradd -g mysql mysql (4)、编译安装mysql wget http://mysql.ntu.edu.tw/Downloads/MySQL-5.5/mysql-5.5.27.tar.gz tar zxf mysql-5.5.27.tar.gz cd mysql-5.5.27 cmake -DCMAKE_INSTALL_PREFIX=/usr/local/mysql \ -DMYSQL_UNIX_ADDR=/tmp/mysql.sock \ -DDEFAULT_CHARSET=utf8 \ -DDEFAULT_COLLATION=utf8_general_ci \ -DWITH_EXTRA_CHARSETS=complex \ -DWITH_READLINE=1 \ -DENABLED_LOCAL_INFILE=1 make -j 4 make install (5)、设置mysql环境变量 [[email protected] ~]# echo ‘PATH=$PATH:/usr/local/mysql/bin‘ >>/etc/profile [[email protected] ~]# source /etc/profile (6)、初始化数据库 [[email protected] ~]# mount /dev/drbd1 /data #说明：数据库存放数据的目录是drbd分区 [[email protected] ~]# cd /usr/local/mysql/ [[email protected] ~]# ./scripts/mysql_install_db --datadir=/data/ --user=mysql (7)、启动数据库 [[email protected] ~]# vim /etc/init.d/mysqld datadir=/data #说明：修改mysql启动脚本，指定数据库的目录为/data [[email protected] ~]# /etc/init.d/mysqld start [[email protected] ~]# chkconfig mysqld off (8)、测试数据库 [[email protected] ~]# mysql -uroot -e "show databases;" +--------------------+ | Database | +--------------------+ | information_schema | | mysql | | performance_schema | +--------------------+ |
|
5、故障切换测试 |
(1)、架构正常状态 master1主节点正常状态 [[email protected] ~]# ip addr|grep eth1 3: eth1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast qlen 1000 inet 192.168.4.2/16 brd 192.168.255.255 scope global eth1 inet 192.168.4.1/16 brd 192.168.255.255 scope global secondary eth1:0 [[email protected] ~]# cat /proc/drbd version: 8.3.13 (api:88/proto:86-96) GIT-hash: 83ca112086600faacab2f157bc5a9324f7bd7f77 build by [email protected], 2012-05-07 11:56:36 1: cs:Connected ro:Primary/Secondary ds:UpToDate/UpToDate C r----- ns:39558 nr:12 dw:39570 dr:151 al:16 bm:1 lo:0 pe:0 ua:0 ap:0 ep:1 wo:b oos:0 [[email protected] ~]# mysql -uroot -e "create database coral;" [[email protected] ~]# mysql -uroot -e "show databases like ‘coral‘;" +------------------+ | Database (coral) | +------------------+ | coral | +------------------+ #说明：master1为主节点，拥有VIP地址，为drbd的主节点 master2备节点正常状态 [[email protected] ~]# ip addr|grep eth1 3: eth1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast qlen 1000 inet 192.168.4.3/16 brd 192.168.255.255 scope global eth1 [[email protected] ~]# cat /proc/drbd version: 8.3.13 (api:88/proto:86-96) GIT-hash: 83ca112086600faacab2f157bc5a9324f7bd7f77 build by [email protected], 2012-05-07 11:56:36 1: cs:Connected ro:Secondary/Primary ds:UpToDate/UpToDate C r----- ns:0 nr:48 dw:48 dr:0 al:0 bm:0 lo:0 pe:0 ua:0 ap:0 ep:1 wo:b oos:0 #说明：master2备节点没有VIP地址，为drbd备节点 (2)、模拟master1宕机故障状态 [[email protected] ~]# /etc/init.d/heartbeat stop <==模拟master1故障宕机 [[email protected] ~]# tailf /var/log/ha-log <==查看备节点接管日志 heartbeat[13209]: 2013/01/23_04:09:36 info: Received shutdown notice from ‘master1‘. heartbeat[13209]: 2013/01/23_04:09:36 info: Resources being acquired from master1. heartbeat[15293]: 2013/01/23_04:09:36 info: acquire local HA resources (standby). heartbeat[15294]: 2013/01/23_04:09:37 info: No local resources [/usr/share/heartbeat/ResourceManager listkeys master2] to acquire. heartbeat[15293]: 2013/01/23_04:09:37 info: local HA resource acquisition completed (standby). heartbeat[13209]: 2013/01/23_04:09:37 info: Standby resource acquisition done [foreign]. harc[15319]: 2013/01/23_04:09:37 info: Running /etc/ha.d/rc.d/status status mach_down[15335]: 2013/01/23_04:09:37 info: Taking over resource group IPaddr::192.168.4.1/16/eth1 ResourceManager[15361]: 2013/01/23_04:09:37 info: Acquiring resource group: master1 IPaddr::192.168.4.1/16/eth1 drbddisk::data Filesystem::/dev/drbd1::/data::ext3 mysqld IPaddr[15388]: 2013/01/23_04:09:37 INFO: Resource is stopped ResourceManager[15361]: 2013/01/23_04:09:37 info: Running /etc/ha.d/resource.d/IPaddr 192.168.4.1/16/eth1 start IPaddr[15486]: 2013/01/23_04:09:38 INFO: Using calculated netmask for 192.168.4.1: 255.255.0.0 IPaddr[15486]: 2013/01/23_04:09:38 INFO: eval ifconfig eth1:0 192.168.4.1 netmask 255.255.0.0 broadcast 192.168.255.255 IPaddr[15457]: 2013/01/23_04:09:38 INFO: Success ResourceManager[15361]: 2013/01/23_04:09:38 info: Running /etc/ha.d/resource.d/drbddisk data start Filesystem[15636]: 2013/01/23_04:09:39 INFO: Resource is stopped ResourceManager[15361]: 2013/01/23_04:09:39 info: Running /etc/ha.d/resource.d/Filesystem /dev/drbd1 /data ext3 start Filesystem[15717]: 2013/01/23_04:09:39 INFO: Running start for /dev/drbd1 on /data Filesystem[15706]: 2013/01/23_04:09:39 INFO: Success ResourceManager[15361]: 2013/01/23_04:09:40 info: Running /etc/init.d/mysqld start mach_down[15335]: 2013/01/23_04:09:44 info: /usr/share/heartbeat/mach_down: nice_failback: foreign resources acquired mach_down[15335]: 2013/01/23_04:09:44 info: mach_down takeover complete for node master1. heartbeat[13209]: 2013/01/23_04:09:44 info: mach_down takeover complete. heartbeat[13209]: 2013/01/23_04:10:09 WARN: node master1: is dead heartbeat[13209]: 2013/01/23_04:10:09 info: Dead node master1 gave up resources. heartbeat[13209]: 2013/01/23_04:10:09 info: Link master1:eth2 dead. #说明：当备节点无法检测到主节点的心跳时，自动接管资源，启动VIP地址、drbd服务，自动挂载drbd，启动mysqld服务，备节点接管后，数据依然存在，检测启动的服务如下： [[email protected] ~]# ip addr|grep eth1 3: eth1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast qlen 1000 inet 192.168.4.3/16 brd 192.168.255.255 scope global eth1 inet 192.168.4.1/16 brd 192.168.255.255 scope global secondary eth1:0 [[email protected] ~]# cat /proc/drbd version: 8.3.13 (api:88/proto:86-96) GIT-hash: 83ca112086600faacab2f157bc5a9324f7bd7f77 build by [email protected], 2012-05-07 11:56:36 1: cs:Connected ro:Primary/Secondary ds:UpToDate/UpToDate C r----- ns:3 nr:95 dw:98 dr:10 al:1 bm:0 lo:0 pe:0 ua:0 ap:0 ep:1 wo:b oos:0 [[email protected] ~]# df -h Filesystem Size Used Avail Use% Mounted on /dev/sda3 19G 4.7G 14G 26% / /dev/sda1 190M 18M 163M 10% /boot tmpfs 60M 0 60M 0% /dev/shm /dev/drbd1 471M 40M 408M 9% /data [[email protected] ~]# mysql -uroot -e "show databases like ‘coral‘;" +------------------+ | Database (coral) | +------------------+ | coral | +------------------+ (3)、模拟master1宕机恢复状态 启动的顺序是：先启动VIP--启动drbd资源--挂载drbd分区--启动mysqld服务，日志如下： [[email protected] ~]# /etc/init.d/heartbeat start [[email protected] ~]# tailf /var/log/ha-log heartbeat[27970]: 2013/01/09_17:34:14 info: Version 2 support: no heartbeat[27970]: 2013/01/09_17:34:14 WARN: Logging daemon is disabled --enabling logging daemon is recommended heartbeat[27970]: 2013/01/09_17:34:14 info: ************************** heartbeat[27970]: 2013/01/09_17:34:14 info: Configuration validated. Starting heartbeat 2.1.3 heartbeat[27971]: 2013/01/09_17:34:14 info: heartbeat: version 2.1.3 heartbeat[27971]: 2013/01/09_17:34:14 info: Heartbeat generation: 1351554533 heartbeat[27971]: 2013/01/09_17:34:14 info: glib: UDP multicast heartbeat started for group 225.0.0.7 port 694 interface eth2 (ttl=1 loop=0) heartbeat[27971]: 2013/01/09_17:34:14 info: G_main_add_TriggerHandler: Added signal manual handler heartbeat[27971]: 2013/01/09_17:34:14 info: G_main_add_TriggerHandler: Added signal manual handler heartbeat[27971]: 2013/01/09_17:34:14 info: G_main_add_SignalHandler: Added signal handler for signal 17 heartbeat[27971]: 2013/01/09_17:34:14 info: Local status now set to: ‘up‘ heartbeat[27971]: 2013/01/09_17:34:16 info: Link master2:eth2 up. heartbeat[27971]: 2013/01/09_17:34:16 info: Status update for node master2: status active harc[27978]: 2013/01/09_17:34:16 info: Running /etc/ha.d/rc.d/status status heartbeat[27971]: 2013/01/09_17:34:17 info: Comm_now_up(): updating status to active heartbeat[27971]: 2013/01/09_17:34:17 info: Local status now set to: ‘active‘ heartbeat[27971]: 2013/01/09_17:34:17 info: remote resource transition completed. heartbeat[27971]: 2013/01/09_17:34:17 info: remote resource transition completed. heartbeat[27971]: 2013/01/09_17:34:17 info: Local Resource acquisition completed. (none) heartbeat[27971]: 2013/01/09_17:34:18 info: master2 wants to go standby [foreign] heartbeat[27971]: 2013/01/09_17:34:20 info: standby: acquire [foreign] resources from master2 heartbeat[27997]: 2013/01/09_17:34:20 info: acquire local HA resources (standby). ResourceManager[28010]: 2013/01/09_17:34:20 info: Acquiring resource group: master1 IPaddr::192.168.4.1/16/eth1 drbddisk::data Filesystem::/dev/drbd1::/data::ext3 mysqld IPaddr[28037]: 2013/01/09_17:34:21 INFO: Resource is stopped ResourceManager[28010]: 2013/01/09_17:34:21 info: Running /etc/ha.d/resource.d/IPaddr 192.168.4.1/16/eth1 start IPaddr[28135]: 2013/01/09_17:34:21 INFO: Using calculated netmask for 192.168.4.1: 255.255.0.0 IPaddr[28135]: 2013/01/09_17:34:21 INFO: eval ifconfig eth1:0 192.168.4.1 netmask 255.255.0.0 broadcast 192.168.255.255 IPaddr[28106]: 2013/01/09_17:34:21 INFO: Success ResourceManager[28010]: 2013/01/09_17:34:21 info: Running /etc/ha.d/resource.d/drbddisk data start Filesystem[28286]: 2013/01/09_17:34:21 INFO: Resource is stopped ResourceManager[28010]: 2013/01/09_17:34:21 info: Running /etc/ha.d/resource.d/Filesystem /dev/drbd1 /data ext3 start Filesystem[28367]: 2013/01/09_17:34:21 INFO: Running start for /dev/drbd1 on /data Filesystem[28356]: 2013/01/09_17:34:21 INFO: Success ResourceManager[28010]: 2013/01/09_17:34:22 info: Running /etc/init.d/mysqld start heartbeat[27997]: 2013/01/09_17:34:25 info: local HA resource acquisition completed (standby). heartbeat[27971]: 2013/01/09_17:34:25 info: Standby resource acquisition done [foreign]. heartbeat[27971]: 2013/01/09_17:34:25 info: Initial resource acquisition complete (auto_failback) heartbeat[27971]: 2013/01/09_17:34:25 info: remote resource transition completed. 备节点释放资源顺序：停止mysqld服务--卸载drbd1分区--设置drbd为备节点--关闭VIP地址，日志如下： [[email protected] ~]# tailf /var/log/ha-log heartbeat[13209]: 2013/01/23_04:26:53 info: Heartbeat restart on node master1 heartbeat[13209]: 2013/01/23_04:26:53 info: Link master1:eth2 up. heartbeat[13209]: 2013/01/23_04:26:53 info: Status update for node master1: status init heartbeat[13209]: 2013/01/23_04:26:53 info: Status update for node master1: status up harc[16151]: 2013/01/23_04:26:53 info: Running /etc/ha.d/rc.d/status status harc[16167]: 2013/01/23_04:26:53 info: Running /etc/ha.d/rc.d/status status heartbeat[13209]: 2013/01/23_04:26:53 info: all clients are now paused heartbeat[13209]: 2013/01/23_04:26:55 info: Status update for node master1: status active harc[16183]: 2013/01/23_04:26:55 info: Running /etc/ha.d/rc.d/status status heartbeat[13209]: 2013/01/23_04:26:55 info: all clients are now resumed heartbeat[13209]: 2013/01/23_04:26:55 info: remote resource transition completed. heartbeat[13209]: 2013/01/23_04:26:55 info: master2 wants to go standby [foreign] heartbeat[13209]: 2013/01/23_04:26:55 info: standby: master1 can take our foreign resources heartbeat[16199]: 2013/01/23_04:26:55 info: give up foreign HA resources (standby). ResourceManager[16212]: 2013/01/23_04:26:55 info: Releasing resource group: master1 IPaddr::192.168.4.1/16/eth1 drbddisk::data Filesystem::/dev/drbd1::/data::ext3 mysqld ResourceManager[16212]: 2013/01/23_04:26:55 info: Running /etc/init.d/mysqld stop ResourceManager[16212]: 2013/01/23_04:26:57 info: Running /etc/ha.d/resource.d/Filesystem /dev/drbd1 /data ext3 stop Filesystem[16297]: 2013/01/23_04:26:57 INFO: Running stop for /dev/drbd1 on /data Filesystem[16297]: 2013/01/23_04:26:57 INFO: Trying to unmount /data Filesystem[16297]: 2013/01/23_04:26:57 INFO: unmounted /data successfully Filesystem[16286]: 2013/01/23_04:26:57 INFO: Success ResourceManager[16212]: 2013/01/23_04:26:57 info: Running /etc/ha.d/resource.d/drbddisk data stop ResourceManager[16212]: 2013/01/23_04:26:57 info: Running /etc/ha.d/resource.d/IPaddr 192.168.4.1/16/eth1 stop IPaddr[16445]: 2013/01/23_04:26:58 INFO: ifconfig eth1:0 down IPaddr[16416]: 2013/01/23_04:26:58 INFO: Success heartbeat[16199]: 2013/01/23_04:26:58 info: foreign HA resource release completed (standby). heartbeat[13209]: 2013/01/23_04:26:58 info: Local standby process completed [foreign]. heartbeat[13209]: 2013/01/23_04:27:02 WARN: 1 lost packet(s) for [master1] [15:17] heartbeat[13209]: 2013/01/23_04:27:02 info: remote resource transition completed. heartbeat[13209]: 2013/01/23_04:27:02 info: No pkts missing from master1! heartbeat[13209]: 2013/01/23_04:27:02 info: Other node completed standby takeover of foreign resources. |
|
6、从库同VIP同步 |
(1)、master配置 1)、设置server-id值并开启Binlog参数 [[email protected] ~]# vim /etc/my.cnf log-bin=/usr/local/mysql/mysql-bin server-id = 3 [[email protected] ~]# /etc/init.d/mysqld restart #注意：只有master1有重启操作，master2无需重启操作，因为备节点的mysql是未启动状态，备节点只有heartbeat才能启动mysql 2)、授权并建立同步账户rep [[email protected] ~]# mysql -uroot -p mysql> GRANT REPLICATION SLAVE ON *.* TO ‘rep‘@‘192.168.4.%‘ IDENTIFIED BY ‘rep‘; (2)、slave配置 1)、设置server-id值并关闭binlog设置 [[email protected] ~]# vim /etc/my.cnf #log-bin=mysql-bin server-id = 4 [[email protected] ~]# /etc/init.d/mysqld restart #说明：从库无需开启binlog日志功能，除非有需求做级联复制架构或对mysql增量备份操作才开启 2)、配置同步参数 [[email protected] ~]# mysql -uroot CHANGE MASTER TO MASTER_HOST=‘192.168.4.1‘, MASTER_PORT=3306, MASTER_USER=‘rep‘, MASTER_PASSWORD=‘rep‘, MASTER_LOG_FILE=‘mysql-bin.000001‘, MASTER_LOG_POS=0; 3)、检查是否主从同步 [[email protected] ~]# mysql -uroot mysql> show slave status\G ... Slave_IO_Running: Yes Slave_SQL_Running: Yes ... (3)、模拟高可用宕机切换是否影响从库同步 1)、主从正常状态 [[email protected] ~]# mysql -uroot mysql> create database coral1; Query OK, 1 row affected (0.02 sec) [[email protected] ~]# mysql -uroot -e "show slave status\G"|egrep "Slave_IO_Running|Slave_SQL_Running" Slave_IO_Running: Yes Slave_SQL_Running: Yes [[email protected] ~]# mysql -uroot -e "show databases like ‘coral%‘;" +-------------------+ | Database (coral%) | +-------------------+ | coral1 | +-------------------+ 2)、模拟高可用主节点宕机 [[email protected] ~]# /etc/init.d/heartbeat stop #说明：模拟主节点宕机 [[email protected] ~]# ip addr|grep eth1 3: eth1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast qlen 1000 inet 192.168.4.3/16 brd 192.168.255.255 scope global eth1 inet 192.168.4.1/16 brd 192.168.255.255 scope global secondary eth1:0 [[email protected] ~]# mysql -uroot mysql> create database coral2; Query OK, 1 row affected (0.08 sec) #说明：VIP地址已经漂移到master2上面 [[email protected] ~]# mysql -uroot -e "show slave status\G"|egrep "Slave_IO_Running|Slave_SQL_Running" Slave_IO_Running: Yes Slave_SQL_Running: Yes [[email protected] ~]# mysql -uroot -e "show databases like ‘coral%‘" +-------------------+ | Database (coral%) | +-------------------+ | coral1 | | coral2 | +-------------------+ #注意：高可用主备节点切换过程中，会有一段时间从库才能连接上，大于在60秒内 #说明：此时主从同步是正常的 3)、模拟高可用主节点宕机恢复 [[email protected] ~]# /etc/init.d/heartbeat start [[email protected] ~]# mysql -uroot mysql> create database coral3; [[email protected] ~]# mysql -uroot -e "show slave status\G"|egrep "Slave_IO_Running|Slave_SQL_Running" Slave_IO_Running: Yes Slave_SQL_Running: Yes [[email protected] ~]# mysql -uroot -e "show databases like ‘coral%‘" +-------------------+ | Database (coral%) | +-------------------+ | coral1 | | coral2 | | coral3 | +-------------------+ #说明：高可用主节点故障恢复后也不影响主从库的同步 |
|
7、高可用脑裂问题及解决方案 |
(1)、导致裂脑发生的原因 1、高可用服务器之间心跳链路故障，导致无法相互检查心跳 2、高可用服务器上开启了防火墙，阻挡了心跳检测 3、高可用服务器上网卡地址等信息配置不正常，导致发送心跳失败 4、其他服务配置不当等原因，如心跳方式不同，心跳广播冲突，软件BUG等 (2)、防止裂脑一些方案 1、加冗余线路 2、检测到裂脑时，强行关闭心跳检测(远程关闭主节点，控制电源的电路fence) 3、做好脑裂的监控报警 4、报警后，备节点在接管时设置比较长的时间去接管，给运维人员足够的时间去处理(人为处理) 5、启动磁盘锁，正在服务的一方锁住磁盘，裂脑发生时，让对方完全抢不走"共享磁盘资源" 磁盘锁存在的问题： 使用锁磁盘会有死锁的问题，如果占用共享磁盘的一方不主动"解锁"另一方就永远得不到共享磁盘，假如服务器节点突然死机或崩溃，就不可能执行解锁命令，备节点也就无法接管资源和服务了，有人在HA中设计了智能锁，正在提供服务的一方只在发现心跳全部断开时才会启用磁盘锁，平时就不上锁 |
