corosync+drbd+mysql实现的高可用_MySQL
要求：
一、能够在同一网段内直接通信
二、节点名称，要和uname的结果一样，并保证可以根据节点名称解析到节点的IP地址，配置本地/etc/hosts
三、SSH互信通信
四、保证时间同步
环境准备配置：
test1,192.168.10.55配置
1、配置IP地址
[root@test1~]#vim/etc/sysconfig/network-scripts/ifcfg-eth0
登录后复制
2、配置主机名
[root@test1~]#uname-n [root@test1~]#hostnamemaster1.local#临时生效 [root@test1~]#vim/etc/sysconfig/network#永久生效
登录后复制
3、配置主机名解析
[root@test1~]#vim/etc/hosts 添加： 192.168.10.55master1.local 192.168.10.56master2.local
登录后复制
3.2、测试主机名通信
[root@test1~]#pingmaster1.local [root@test1~]#pingmaster2.local
登录后复制
4、配置SSH互信认证
[root@test1~]#ssh-keygen-trsa-P'' [root@test1~]#ssh-copy-id-i.ssh/[email protected]
登录后复制
5、使用ntp同步时间
在crontab中加入每5分钟执行一次ntpdate命令，用来保证服务器时间是同步的
[root@test1~]#crontab-e */5****/sbin/ntpdate192.168.10.1&>/dev/null
登录后复制
test2,192.168.10.56配置
1、配置IP地址
[root@test2~]#vim/etc/sysconfig/network-scripts/ifcfg-eth0
登录后复制
2、配置主机名
[root@test2~]#uname-n [root@test2~]#hostnametest2.local#临时生效 [root@test2~]#vim/etc/sysconfig/network#永久生效
登录后复制
3、配置主机名解析
[root@test2~]#vim/etc/hosts 添加： 192.168.10.55test1.localtest1 192.168.10.56test2.localtest2
登录后复制
3.2、测试主机名通信
[root@test2~]#pingtest1.local [root@test2~]#pingtest1
登录后复制
4、配置SSH互信认证
[root@test2~]#ssh-keygen-trsa-P'' [root@test2~]#ssh-copy-id-i.ssh/[email protected]
登录后复制
5、使用ntp同步时间
在crontab中加入每5分钟执行一次ntpdate命令，用来保证服务器时间是同步的
[root@test2~]#crontab-e */5****/sbin/ntpdate192.168.10.1&>/dev/null
登录后复制
安装配置heartbeat
CentOS直接yum安装报错，提示找不到可用的软件包
解决办法：
[root@test1src]#wgethttp://mirrors.sohu.com/fedora-epel/6/i386/epel-release-6-8.noarch.rpm [root@test1src]#rpm-ivhepel-release-6-8.noarch.rpm
登录后复制
6.1、安装heartbeat：
[root@test1src]#yuminstallheartbeat
登录后复制
6.2、copy配置文件：
[root@test1src]#cp/usr/share/doc/heartbeat-3.0.4/{ha.cf,authkeys,haresources}/etc/ha.d/登录后复制
6.3、配置认证文件：
[root@test1src]#ddif=/dev/randomcount=1bs=512|md5sum#生成随机数 [root@test1src]#vim/etc/ha.d/authkeys auth1 1md5d0f70c79eeca5293902aiamheartbeat [root@test1src]#chmod600authkeys
登录后复制
test2节点的heartbeat安装和test1一样，此处略过。
6.4、heartbeat主配置文件参数：
[root@test2~]#vim/etc/ha.d/ha.cf #debugfile/var/log/ha-debug#排错日志 logfile#日志位置 keepalive2#多长时间发送一次心跳检测，默认2秒，可以使用ms deadtime30#多长时间检测不到主机就认为挂掉 warntime10#如果没有收到心跳信息，那么在等待多长时间就认为对方挂掉 initdead120#第一个节点起来后，等待其他节点的时间 baud19200#串行线缆的发送速率是多少 auto_failbackon#故障恢复后是否转移回来 ping10.10.10.254#pingnode，万一节点主机不通，要ping哪个主机 ping_groupgroup110.10.10.25410.10.10.253#pingnodegroup，只要组内有一台主机能ping通就可以 respawnhacluster/usr/lib/heartbeat/ipfail#当一个heartbeat服务停止了，会重启对端的heartbeat服务 deadping30#pingnodes多长时间ping不通，就真的故障了 #serialserialportname...#串行设备是什么 serial/dev/ttyS0#Linux serial/dev/cuaa0#FreeBSD serial/dev/cuad0#FreeBSD6.x serial/dev/cua/a#Solaris #Whatinterfacestobroadcastheartbeatsover?#如果使用以太网，定义使用单播、组播还是广播发送心跳信息 bcasteth0#广播 mcasteth0225.0.0.169410#组播 ucasteth0192.168.1.2#单播，只有两个节点的时候才用单播 #定义stonith主机 stonith_host*baytech10.0.0.3myloginmysecretpassword stonith_hostken3rps10/dev/ttyS1kathy0 stonith_hostkathyrps10/dev/ttyS1ken30 #Tellwhatmachinesareinthecluster#告诉集群中有多少个节点， 每一个节点用node和主机名写一行，主机名要和uname-n保持一致 nodeken3 nodekathy 一般只要定义心跳信息的发送方式、和集群中的节点就行。 bcasteth0 nodetest1.local nodetest2.local
登录后复制
6.5、定义haresources资源配置文件：
登录后复制
[root@test2~]#vim/etc/ha.d/haresources #node110.0.0.170Filesystem::/dev/sda1::/data1::ext2#默认用作主节点的主机名，要跟uname-n一样。VIP是多少。 自动挂载哪个设备，到哪个目录下，文件类型是什么。资源类型的参数要用双冒号隔开 #just.linux-ha.org135.9.216.110http#和上面一样，这里使用的资源是在/etc/rc.d/init.d/下面的， 默认先到/etc/ha.d/resource.d/目录下找资源，找不到在到/etc/rc.d/init.d/目录找 master1.localIPaddr::192.168.10.2/24/eth0mysqld master1.localIPaddr::192.168.10.2/24/eth0drbddisk::dataFilesystem::/dev/drbd1::/data::ext3mysqld #使用IPaddr脚本来配置VIP
登录后复制
6.6、拷贝master1.local的配置文件到master2.local上
[root@test1~]#scp-pha.cfharesourcesauthkeysmaster2.local:/etc/ha.d/
登录后复制
7、启动heartbeat
[root@test1~]#serviceheartbeatstart [root@test1~]#sshmaster2.local'serviceheartbeatstart'#一定要在test1上通过ssh的方式启动test2节点的heartbeat
登录后复制
7.1、查看heartbeat启动日志
[root@test1~]#tail-f/var/log/messages Feb1615:12:45test-1heartbeat:[16056]:info:Configurationvalidated.Startingheartbeat3.0.4 Feb1615:12:45test-1heartbeat:[16057]:info:heartbeat:version3.0.4 Feb1615:12:45test-1heartbeat:[16057]:info:Heartbeatgeneration:1455603909 Feb1615:12:45test-1heartbeat:[16057]:info:glib:UDPBroadcastheartbeatstartedonport694(694)interfaceeth0 Feb1615:12:45test-1heartbeat:[16057]:info:glib:UDPBroadcastheartbeatclosedonport694interfaceeth0-Status:1 Feb1615:12:45test-1heartbeat:[16057]:info:glib:pingheartbeatstarted. Feb1615:12:45test-1heartbeat:[16057]:info:G_main_add_TriggerHandler:Addedsignalmanualhandler Feb1615:12:45test-1heartbeat:[16057]:info:G_main_add_TriggerHandler:Addedsignalmanualhandler Feb1615:12:45test-1heartbeat:[16057]:info:G_main_add_SignalHandler:Addedsignalhandlerforsignal17 Feb1615:12:45test-1heartbeat:[16057]:info:Localstatusnowsetto:'up' Feb1615:12:45test-1heartbeat:[16057]:info:Link192.168.10.1:192.168.10.1up. Feb1615:12:45test-1heartbeat:[16057]:info:Statusupdatefornode192.168.10.1:statusping Feb1615:12:45test-1heartbeat:[16057]:info:Linktest1.local:eth0up. Feb1615:12:51test-1heartbeat:[16057]:info:Linktest2.local:eth0up. Feb1615:12:51test-1heartbeat:[16057]:info:Statusupdatefornodetest2.local:statusup Feb1615:12:51test-1harc(default)[16068]:info:Running/etc/ha.d//rc.d/statusstatus Feb1615:12:52test-1heartbeat:[16057]:WARN:1lostpacket(s)for[test2.local][3:5] Feb1615:12:52test-1heartbeat:[16057]:info:Nopktsmissingfromtest2.local! Feb1615:12:52test-1heartbeat:[16057]:info:Comm_now_up():updatingstatustoactive Feb1615:12:52test-1heartbeat:[16057]:info:Localstatusnowsetto:'active' Feb1615:12:52test-1heartbeat:[16057]:info:Statusupdatefornodetest2.local:statusactive Feb1615:12:52test-1harc(default)[16086]:info:Running/etc/ha.d//rc.d/statusstatus Feb1615:13:02test-1heartbeat:[16057]:info:localresourcetransitioncompleted. Feb1615:13:02test-1heartbeat:[16057]:info:Initialresourceacquisitioncomplete(T_RESOURCES(us)) Feb1615:13:02test-1heartbeat:[16057]:info:remoteresourcetransitioncompleted. Feb1615:13:02test-1/usr/lib/ocf/resource.d//heartbeat/IPaddr(IPaddr_192.168.10.2)[16138]:INFO:Resourceisstopped Feb1615:13:02test-1heartbeat:[16102]:info:LocalResourceacquisitioncompleted. Feb1615:13:02test-1harc(default)[16219]:info:Running/etc/ha.d//rc.d/ip-request-respip-request-resp Feb1615:13:02test-1ip-request-resp(default)[16219]:receivedip-request-respIPaddr::192.168.10.2/24/eth0OKyes Feb1615:13:02test-1ResourceManager(default)[16238]:info:Acquiringresourcegroup:test1.localIPaddr::192.168.10.2/24/eth0mysqld Feb1615:13:02test-1/usr/lib/ocf/resource.d//heartbeat/IPaddr(IPaddr_192.168.10.2)[16264]:INFO:Resourceisstopped Feb1615:13:03test-1ResourceManager(default)[16238]:info:Running/etc/ha.d/resource.d/IPaddr192.168.10.2/24/eth0start Feb1615:13:03test-1IPaddr(IPaddr_192.168.10.2)[16386]:INFO:Addinginetaddress192.168.10.2/24withbroadcastaddress192.168.10.255todeviceeth0 Feb1615:13:03test-1IPaddr(IPaddr_192.168.10.2)[16386]:INFO:Bringingdeviceeth0up Feb1615:13:03test-1IPaddr(IPaddr_192.168.10.2)[16386]:INFO:/usr/libexec/heartbeat/send_arp-i200-r5-p/var/run/resource-agents/send_arp-192.168.10.2eth0192.168.10.2autonot_usednot_used Feb1615:13:03test-1/usr/lib/ocf/resource.d//heartbeat/IPaddr(IPaddr_192.168.10.2)[16360]:INFO:Success Feb1615:13:03test-1ResourceManager(default)[16238]:info:Running/etc/init.d/mysqldstart Feb1615:13:04test-1ntpd[1605]:Listennormallyon15eth0192.168.10.2UDP123
登录后复制
说明：
1、Link test1.local:eth0 up、Link test2.local:eth0 up #两个节点连接成功并为UP状态。
2、Link 192.168.10.1:192.168.10.1 up #ping节点的IP也已经启动
3、info: Running /etc/init.d/mysqld start #mysql启动成功
4、Listen normally on 15 eth0 192.168.10.2 UDP 123 #VIP启动成功
7.2、查看heartbeat的VIP
[[email protected]]#ipadd|grep"10.2" inet192.168.10.55/24brd192.168.10.255scopeglobaleth0 inet192.168.10.2/24brd192.168.10.255scopeglobalsecondaryeth0
登录后复制
[[email protected]]#ipadd|grep"10.2" inet192.168.10.56/24brd192.168.10.255scopeglobaleth0
登录后复制
注：可以看到现在VIP是在master1.local主机上。而master2.local上没有VIP
8、测试效果
8.1、正常情况下连接mysql
[[email protected]]#mysql-uroot-h'192.168.10.2'-p Enterpassword: WelcometotheMySQLmonitor.Commandsendwith;or\g. YourMySQLconnectionidis2 Serverversion:5.5.44Sourcedistribution Copyright(c)2000,2013,Oracleand/oritsaffiliates.Allrightsreserved. OracleisaregisteredtrademarkofOracleCorporationand/orits affiliates.Othernamesmaybetrademarksoftheirrespective owners. Type'help;'or'\h'forhelp.Type'\c'toclearthecurrentinputstatement. mysql>showvariableslike'server_id'; +---------------+-------+ |Variable_name|Value| +---------------+-------+ |server_id|1| +---------------+-------+ 1rowinset(0.00sec) mysql>
登录后复制
8.2、关闭master1.local上的heartbeat
[[email protected]]#serviceheartbeatstop StoppingHigh-Availabilityservices:Done. [[email protected]]#ipadd|grep"192.168.10.2" inet192.168.10.55/24brd192.168.10.255scopeglobaleth0 [[email protected]]#ipadd|grep"192.168.10.2" inet192.168.10.56/24brd192.168.10.255scopeglobaleth0 inet192.168.10.2/24brd192.168.10.255scopeglobalsecondaryeth0
登录后复制
注：这个时候VIP已经漂移到了master2.local主机上，我们在来看看连接mysql的server_id
[[email protected]]#mysql-uroot-h'192.168.10.2'-p Enterpassword: WelcometotheMySQLmonitor.Commandsendwith;or\g. YourMySQLconnectionidis2 Serverversion:5.5.44Sourcedistribution Copyright(c)2000,2013,Oracleand/oritsaffiliates.Allrightsreserved. OracleisaregisteredtrademarkofOracleCorporationand/orits affiliates.Othernamesmaybetrademarksoftheirrespective owners. Type'help;'or'\h'forhelp.Type'\c'toclearthecurrentinputstatement. mysql>showvariableslike'server_id'; +---------------+-------+ |Variable_name|Value| +---------------+-------+ |server_id|2| +---------------+-------+ 1rowinset(0.00sec) mysql>
登录后复制
注：server_id已经从1变成了2，证明此时访问的是master2.local主机上的mysql服务
测试完毕。下面配置drbd让两台mysql服务器之间使用同一个文件系统，以实现mysql的写高可用。
9、配置DRBD
DRBD：（Distributed Replicated Block Device）分布式复制块设备，是linux内核中的一个模块。DRBD作为磁盘镜像来讲，它一定是主从架构的，它决不允许两个节点同时读写，仅允许一个节点能读写，从节点不能读写和挂载，
但是DRDB有双主的概念，主、从的角色可以切换。DRBD分别将位于两台主机上的硬盘或磁盘分区做成镜像设备，当我们客户端的程序向主节点发起存储请求的时候，这个数据会在底层以TCP/IP协议按位同布一份到备节点，
所以这能保证只要我们在主节点上存的数据，备节点上在按位一定有一模一样的一份数据。这是在两台主机上实现的，这意味着DRBD是工作在内核模块当中。不像RAID1的镜像是在同一台主机上实现的。
DRBD双主模型的实现：一个节点在数据访问的时候，它一定会将数据、元数据载入内存的，而且它对于某个文件内核中加锁的操作，另一个节点的内核是看不到的，那如果它能将它自己施加的锁通知给另一个节点的内核就可以了。
在这种情况下，我们就只能通过message layer（heartbeat、corosync都可）、pathmaker（把DRBD定义成资源），然后把这两个主机上对应的镜像格式化成集群文件系统（GFS2/OCFS2）。
这就是基于结合分布式文件锁管理器（DLM Distributed Lock Manager）以及集群文件系统所完成的双主模型。DRBD集群只允许有两个节点，要么双主，要么主从。
9.1、DRBD的三种工作模型
A、数据在本地DRBD存储完成后向应用程序返回存储成功的消息，异步模型。效率高，性能好。数据不安全
B、数据在本地DRBD存储完成后并且通过TCP/IP把所有数据发送到从DRBD中，才向本地的应用程序返回存储成功的消息，半同步模型。一般不用。
C、数据在本地DRBD存储完成后，通过TCP/IP把所有数据发送到从DRBD中，从DRBD存储完成后才向应用程序返回成功的消息，同步模型。效率低，性能若，但是数据安全可靠性大，用的最多。
9.2、DRBD的资源
1、资源名称，可以是任意的ascii码不包含空格的字符
2、DRBD设备，在双方节点上，此DRBD设备的设备文件，一般为/dev/drbdN，其主设备号相同都是147，此设备号用来标识不通的设备
3、磁盘配置，在双方节点上，各自提供的存储设备，可以是个分区，可以是任何类型的块设备，也可以是lvm
4、网络配置，双方数据同步时，所使用的网络属性
9.3、安装DRBD
drbd在2.6.33开始，才整合进内核的。
9.3.1、下载drbd
[root@test1~]#wget-O/usr/local/srchttp://oss.linbit.com/drbd/8.4/drbd-8.4.3.tar.gz
登录后复制
9.3.2、安装drbd软件
[root@test1~]#cd/usr/local/src [root@test1src]#tar-zxvfdrbd-8.4.3.tar.gz [root@test1src]#cd/usr/local/src/drbd-8.4.3 [[email protected]]#./configure--prefix=/usr/local/drbd--with-km [[email protected]]#makeKDIR=/usr/src/kernels/2.6.32-573.18.1.el6.x86_64
登录后复制
[[email protected]]#makeinstall [[email protected]]#mkdir-p/usr/local/drbd/var/run/drbd [[email protected]]#cp/usr/local/drbd/etc/rc.d/init.d/drbd/etc/rc.d/init.d/
登录后复制
9.3.3、安装drbd模块
[[email protected]]#cddrbd/ [root@test1drbd]#makeclean [root@test1drbd]#makeKDIR=/usr/src/kernels/2.6.32-573.18.1.el6.x86_64 [root@test1drbd]#cpdrbd.ko/lib/modules/`uname-r`/kernel/lib/ [root@test1drbd]#modprobedrbd [root@test1drbd]#lsmod|grepdrbd
登录后复制
9.3.4、为drbd创建新分区
[root@test1drbd]#fdisk/dev/sdb
WARNING:DOS-compatiblemodeisdeprecated.It'sstronglyrecommendedto
switchoffthemode(command'c')andchangedisplayunitsto
sectors(command'u').
Command(mforhelp):n
Commandaction
eextended
pprimarypartition(1-4)
p
Partitionnumber(1-4):1
Firstcylinder(1-1305,default1):
Usingdefaultvalue1
Lastcylinder,+cylindersor+size{K,M,G}(1-1305,default1305):+9G
Command(mforhelp):w
Thepartitiontablehasbeenaltered!
Callingioctl()tore-readpartitiontable.
AWRNING:Re-readingthepartitiontablefailedwitherror16:Deviceorresourcebusy.
Thekernelstillusestheoldtable.
Thenewtablewillbeusedatthenextreboot.
Syncingdisks.
[root@test1drbd]#partprobe/dev/sdb登录后复制
test2节点的drbd安装和分区配置步骤略过，和test1上一样安装，test2节点的drbd配置文件保证和test1节点一样，使用scp传到test2节点即可
10、配置drbd
10.1、配置drbd的通用配置文件
[[email protected]]#cd/usr/local/drbd/etc/drbd.d [[email protected]]#vimglobal_common.conf global{#global是全局配置 usage-countno;#官方用来统计有多少个客户使用drbd的 #minor-countdialog-refreshdisable-ip-verification } common{#通用配置，用来配置所有资源那些相同属性的。为drbd提供默认属性的 protocolC;#默认使用协议C，即同步模型。 handlers{#处理器段，用来配置drbd的故障切换操作 #TheseareEXAMPLEhandlersonly. #Theymayhavesevereimplications, #likehardresettingthenodeundercertaincircumstances. #Becarefulwhenchosingyourpoison. pri-on-incon-degr"/usr/lib/drbd/notify-pri-on-incon-degr.sh;/usr/lib/drbd/notify-emergency-reboot.sh;echob>/proc/sysrq-trigger;reboot-f";# pri-lost-after-sb"/usr/lib/drbd/notify-pri-lost-after-sb.sh;/usr/lib/drbd/notify-emergency-reboot.sh;echob>/proc/sysrq-trigger;reboot-f";#脑裂之后的操作 local-io-error"/usr/lib/drbd/notify-io-error.sh;/usr/lib/drbd/notify-emergency-shutdown.sh;echoo>/proc/sysrq-trigger;halt-f";#本地i/o错误之后的操作 #fence-peer"/usr/lib/drbd/crm-fence-peer.sh"; #split-brain"/usr/lib/drbd/notify-split-brain.shroot"; #out-of-sync"/usr/lib/drbd/notify-out-of-sync.shroot"; #before-resync-target"/usr/lib/drbd/snapshot-resync-target-lvm.sh-p15---c16k"; #after-resync-target/usr/lib/drbd/unsnapshot-resync-target-lvm.sh; } startup{ #wfc-timeoutdegr-wfc-timeoutoutdated-wfc-timeoutwait-after-sb#设备启动时，两个节点要同步，设置节点的等待时间，超时时间等 } options{ #cpu-maskon-no-data-accessible } disk{ on-io-errordetach;#一旦发生i/o错误，就把磁盘卸载。不继续进行同步 #sizemax-bio-bvecson-io-errorfencingdisk-barrierdisk-flushes #disk-drainmd-flushesresync-rateresync-afteral-extents #c-plan-aheadc-delay-targetc-fill-targetc-max-rate #c-min-ratedisk-timeout } net{#设置网络的buffers/cache大小，初始化时间等 #protocoltimeoutmax-epoch-sizemax-buffersunplug-watermark #connect-intping-intsndbuf-sizercvbuf-sizeko-count #allow-two-primariescram-hmac-algshared-secretafter-sb-0pri #after-sb-1priafter-sb-2prialways-asbprr-conflict #ping-timeoutdata-integrity-algtcp-corkon-congestion #congestion-fillcongestion-extentscsums-algverify-alg #use-rle cram-hmac-alg"sha1";#数据加密使用的算法 shared-secret"mydrbd1fa2jg8";#验证密码 } syncer{ rate200M;#定义数据传输速率 } }
登录后复制
10.2、配置资源文件，资源配置文件的名字要和资源文件中的一样
[[email protected]]#vimmydrbd.res resourcemydrbd{#资源名称，可以是任意的ascii码不包含空格的字符 ontest1.local{#节点1，每个节点必须要能使用名称的方式解析对方节点 device/dev/drbd0;#drbd设备的文件名叫什么 disk/dev/sdb1;#分区设备是哪个 address192.168.10.55:7789;#节点ip和监听的端口 meta-diskinternal;#drbd的meta（原数据）放在什么地方，internal是放在设备内部 } ontest2.local{ device/dev/drbd0; disk/dev/sdb1; address192.168.10.56:7789; meta-diskinternal; } }
登录后复制
10.3、两个节点的配置文件一样，使用工具把配置文件传到另一个节点
[[email protected]]#scp-r/usr/local/drbd/etc/drbd.*test2.local:/usr/local/drbd/etc/
登录后复制
10.4、在每个节点上初始化已定义的资源
[[email protected]]#drbdadmcreate-mdmydrbd --==Thankyouforparticipatingintheglobalusagesurvey==-- Theserver'sresponseis: Writingmetadata... initializingactivitylog NOTinitializedbitmap Newdrbdmetadatablocksuccessfullycreated. [[email protected]]# [[email protected]]#drbdadmcreate-mdmydrbd --==Thankyouforparticipatingintheglobalusagesurvey==-- Theserver'sresponseis: Writingmetadata... initializingactivitylog NOTinitializedbitmap Newdrbdmetadatablocksuccessfullycreated. [[email protected]]#
登录后复制
10.5、分别启动两个节点的drbd服务
[[email protected]]#servicedrbdstart [[email protected]]#servicedrbdstart
登录后复制
11、测试drbd的同步
11.1、查看drbd的启动状态
[[email protected]]#cat/proc/drbd version:8.4.3(api:1/proto:86-101) GIT-hash:[email protected],2016-02-2310:23:03 0:cs:Connectedro:Secondary/Secondaryds:Inconsistent/InconsistentCr-----#两个节点都是从，将来可以把一个提升为主。Inconsistent处于没有同步的状态 ns:0nr:0dw:0dr:0al:0bm:0lo:0pe:0ua:0ap:0ep:1wo:boos:0
登录后复制
11.2、提升一个节点为主，并覆盖从节点的drbd分区数据。在要提升为主的节点上执行
[[email protected]]#drbdadm----overwrite-data-of-peerprimarymydrbd
登录后复制
11.3、查看主节点同步状态
[[email protected]]#watch-n1cat/proc/drbd Every1.0s:cat/proc/drbdTueFeb2317:10:552016 version:8.4.3(api:1/proto:86-101) GIT-hash:[email protected],2016-02-2310:23:03 0:cs:SyncSourcero:Primary/Secondaryds:UpToDate/InconsistentCr---n- ns:619656nr:0dw:0dr:627840al:0bm:37lo:1pe:8ua:64ap:0ep:1wo:boos:369144 [=============>.......]sync'ed:10.3%(369144/987896)K finish:0:00:12speed:25,632(25,464)K/sec
登录后复制
11.4、查看从节点的状态
[root@test2drbd]#cat/proc/drbd version:8.4.3(api:1/proto:86-101) GIT-hash:[email protected],2016-02-2216:05:34 0:cs:Connectedro:Secondary/Primaryds:UpToDate/UpToDateCr----- ns:4nr:9728024dw:9728028dr:1025al:1bm:577lo:0pe:0ua:0ap:0ep:1wo:foos:0
登录后复制
11.5、在主节点格式化分区并挂在写入数据测试
[root@test1drbd]#mke2fs-j/dev/drbd0 [root@test1drbd]#mkdir/mydrbd [root@test1drbd]#mount/dev/drdb0/mydrbd [root@test1drbd]#cd/mydrbd [root@test1mydrbd]#touchdrbd_test_file [root@test1mydrbd]#ls/mydrbd/ drbd_test_filelost+found
登录后复制
11.6、把主节点降级为从，把从节点提升为主。查看数据是否同步
11.1、主节点操作
11.1.1、卸载分区，注意卸载的时候要退出挂在目录，否则会显示设备忙，不能卸载
[root@test1mydrbd]#cd~ [root@test1~]#umount/mydrbd [root@test1~]#drbdadmsecondarymydrbd
登录后复制
11.1.2、查看现在的drbd状态
[root@test1~]#cat/proc/drbd version:8.4.3(api:1/proto:86-101) GIT-hash:[email protected],2016-02-2216:05:34 0:cs:Connectedro:Secondary/Secondaryds:UpToDate/UpToDateCr----- ns:4nr:9728024dw:9728028dr:1025al:1bm:577lo:0pe:0ua:0ap:0ep:1wo:foos:0
登录后复制
注：可以看到，现在drbd的两个节点的状态都是secondary的，下面把从节点提升为主
11.2、从节点操作
11.2.1、提升操作
[root@test1~]#drdbadmprimarymydrbd
登录后复制
11.2.2、挂在drbd分区
[root@test1~]#mkdir/mydrbd [root@test1~]#mount/dev/drbd0/mydrbd/
登录后复制
11.2.3、查看是否有数据
[root@test2~]#ls/myddrbd/ drbd_test_filelost+found
登录后复制
注：可以看到从节点切换成主后，已经同步了数据。drbd搭建完成。下面结合corosync+mysql配置双主高可用。
12、结合corosync+drbd+mysql实现数据库双主高可用
将drbd配置为corosync双节点高可用集群中的资源，能够实现主从角色的自动切换，注意，要把某一个服务配置为高可用集群的资源，一定不能让这个服务开机自动启动。
12.1、关闭两台节点的drbd开机自启动
12.1.1、主节点操作
[[email protected]]#chkconfigdrbdoff [[email protected]]#chkconfig--list|grepdrbd drbd0:off1:off2:off3:off4:off5:off6:off
登录后复制
12.1.2、从节点操作
[[email protected]]#chkconfigdrbdoff [[email protected]]#chkconfig--list|grepdrbd drbd0:off1:off2:off3:off4:off5:off6:off
登录后复制
12.2、卸载drbd的文件系统并把主节点降级为从节点
12.2.1、从节点操作，注意，这里的从节点刚才提升为主了。现在把他降级
[root@test2drbd]#umount/mydata/ [root@test2drbd]#drbdadmsecondarymydrbd [root@test2drbd]#cat/proc/drbd version:8.4.3(api:1/proto:86-101) GIT-hash:[email protected],2016-02-2216:05:34 0:cs:Connectedro:Secondary/Secondaryds:UpToDate/UpToDateCr----- ns:8nr:9728024dw:9728032dr:1073al:1bm:577lo:0pe:0ua:0ap:0ep:1wo:foos:0
登录后复制
注：确保两个节点都是secondary
12.3、停止两个节点的drbd服务
12.3.1、从节点操作
[root@test2drbd]#servicedrbdstop StoppingallDRBDresources:. [root@test2drbd]#
登录后复制
12.3.2、主节点操作
[[email protected]]#servicedrbdstop StoppingallDRBDresources:. [[email protected]]#
登录后复制
12.4、安装corosync并创建日志目录
12.4.1、主节点操作
[[email protected]]#wget-P/etc/yum.repos.dhttp://download.opensuse.org/repositories/network:/ha-clustering:/Stable/CentOS_CentOS-6/network:ha-clustering:Stable.repo [[email protected]]#yuminstallcorosyncpacemakercrmsh [[email protected]]#mkdir/var/log/cluster
登录后复制
12.4.2、从节点操作
[[email protected]]#wget-P/etc/yum.repos.dhttp://download.opensuse.org/repositories/network:/ha-clustering:/Stable/CentOS_CentOS-6/network:ha-clustering:Stable.repo [[email protected]]#mkdir/var/log/cluster [[email protected]]#yuminstallcorosyncpacemakercrmsh
登录后复制
12.5、corosync配置文件
12.5.1、主节点操作
[[email protected]]#cd/etc/corosync/ [root@test1corosync]#cpcorosync.conf.examplecorosync.conf
登录后复制
12.6、配置主节点配置文件，生成corosync秘钥文件并复制给从节点（包括主配置文件）
12.6.1、主节点操作
[root@test1corosync]#vimcorosync.conf
#Pleasereadthecorosync.conf.5manualpage
compatibility:whitetank
totem{
version:2#secauth:Enablemutualnodeauthentication.Ifyouchooseto
#enablethis("on"),thendoremembertocreateashared
#secretwith"corosync-keygen".
secauth:on
threads:2
#interface:defineatleastoneinterfacetocommunicate
#over.Ifyoudefinemorethanoneinterfacestanza,youmust
#alsosetrrp_mode.
interface{
#Ringsmustbeconsecutivelynumbered,startingat0.
ringnumber:0
#Thisisnormallythe*network*addressofthe
#interfacetobindto.Thisensuresthatyoucanuse
#identicalinstancesofthisconfigurationfile
#acrossallyourclusternodes,withouthavingto
#modifythisoption.
bindnetaddr:192.168.10.0
#However,ifyouhavemultiplephysicalnetwork
#interfacesconfiguredforthesamesubnet,thenthe
#networkaddressaloneisnotsufficienttoidentify
#theinterfaceCorosyncshouldbindto.Inthatcase,
#configurethe*host*addressoftheinterface
#instead:
bindnetaddr:192.168.10.0
#Whenselectingamulticastaddress,considerRFC
#2365(which,amongotherthings,specifiesthat
#239.255.x.xaddressesarelefttothediscretionof
#thenetworkadministrator).Donotreusemulticast
#addressesacrossmultipleCorosyncclusterssharing
#thesamenetwork.
mcastaddr:239.212.16.19
#CorosyncusestheportyouspecifyhereforUDP
#messaging,andalsotheimmediatelypreceding
#port.Thusifyousetthisto5405,Corosyncsends
#messagesoverUDPports5405and5404.
mcastport:5405
#Time-to-liveforclustercommunicationpackets.The
#numberofhops(routers)thatthisringwillallow
#itselftopass.Notethatmulticastroutingmustbe
#specificallyenabledonmostnetworkrouters.
ttl:1#每一个数据报文不允许经过路由
}
}
logging{
#Logthesourcefileandlinewheremessagesarebeing
#generated.Whenindoubt,leaveoff.Potentiallyusefulfor
#debugging.
fileline:off
#Logtostandarderror.Whenindoubt,settono.Usefulwhen
#runningintheforeground(wheninvoking"corosync-f")
to_stderr:no
#Logtoalogfile.Whensetto"no",the"logfile"option
#mustnotbeset.
to_logfile:yes
logfile:/var/log/cluster/corosync.log
#Logtothesystemlogdaemon.Whenindoubt,settoyes.
to_syslog:no
#Logdebugmessages(veryverbose).Whenindoubt,leaveoff.
debug:off
#Logmessageswithtimestamps.Whenindoubt,settoon
#(unlessyouareonlyloggingtosyslog,wheredouble
#timestampscanbeannoying).
timestamp:on
logger_subsys{
subsys:AMF
debug:off
}
}
service{
ver:0
name:pacemaker
}
aisexec{
user:root
group:root
}
[root@test1corosync]#corosync-keygen
[root@test1corosync]#scp-pauthkeycorosync.conftest2.local:/etc/corosync/登录后复制
12.7、启动corosync
12.7.1、主节点操作（注意：两个corosync的启动操作都要在主节点上进行）
[root@test1corosync]#servicecorosyncstart StartingCorosyncClusterEngine(corosync):[OK] [root@test1corosync]#sshtest2.local'servicecorosyncstart' [email protected]'spassword: StartingCorosyncClusterEngine(corosync):[OK]
登录后复制
12.8、查看集群状态
12.8.1、问题：安装之后系统没有crm命令，不能使用crm的交互式模式
[root@test1corosync]#crmstatus -bash:crm:commandnotfound
登录后复制
解决办法：安装ha-cluster的yum源，在安装crmsh软件包
[root@test1corosync]#wget-P/etc/yum.repos.d/http://download.opensuse.org/repositories/network:/ha-clustering:/Stable/CentOS_CentOS-6/network:ha-clustering:Stable.repo
登录后复制
参考文档：http://www.dwhd.org/20150530_014731.html 中的第二，安装crmsh步骤
http://serverfault.com/questions/487309/crm-commandcluster-managment-for-pacemaker-not-found-in-latest-centos-6
安装完成后就可以使用crm命令行模式了
12.8.2、查看集群节点状态
[root@test1corosync]#crmstatus Lastupdated:WedFeb2413:47:172016 Lastchange:WedFeb2411:26:062016 Stack:classicopenais(withplugin) CurrentDC:test2.local-partitionwithquorum Version:1.1.11-97629de 2Nodesconfigured,2expectedvotes 0Resourcesconfigured Online:[test1.localtest2.local] Fulllistofresources:
登录后复制
注：显示有2个nodes已配置并处于在线状态
12.8.3、查看pacemaker是否启动
[root@test1corosync]#greppcmk_startup/var/log/cluster/corosync.log Feb2411:05:15corosync[pcmk]info:pcmk_startup:CRM:Initialized Feb2411:05:15corosync[pcmk]Logging:Initializedpcmk_startup
登录后复制
12.8.4、检查集群引擎是否启动
[root@test1corosync]#grep-e"CorosyncClusterEngine"-e"configurationfile"/var/log/cluster/corosync.log
Feb2411:04:16corosync[MAIN]CorosyncClusterEngine('1.4.7'):startedandreadytoprovideservice.
Feb2411:04:16corosync[MAIN]Successfullyreadmainconfigurationfile'/etc/corosync/corosync.conf'.登录后复制
注：经过上面的步骤，可以确认corosync的服务已经启动并没有问题了
12.9、配置corosync的属性
12.9.1、禁用STONITH设备以保证verify不会出错
[root@test1corosync]#crmconfigure crm(live)configure#propertystonith-enabled=false crm(live)configure#verify crm(live)configure#commit
登录后复制
12.9.2、配置当不具备法定票数的时候不能关闭服务
crm(live)configure#propertyno-quorum-policy=ignore crm(live)configure#verify crm(live)configure#commit
登录后复制
12.9.3、配置资源默认粘性值
crm(live)configure#rsc_defaultsresource-stickiness=100 crm(live)configure#verify crm(live)configure#commit
登录后复制
12.9.3、查看当前的配置
crm(live)configure#show nodetest1.local nodetest2.local propertycib-bootstrap-options:\ dc-version=1.1.11-97629de\ cluster-infrastructure="classicopenais(withplugin)"\ expected-quorum-votes=2\ stonith-enabled=false\ no-quorum-policy=ignore rsc_defaultsrsc-options:\ resource-stickiness=100
登录后复制
12.10、配置集群资源
12.10.1、定义drbd的资源
crm(live)configure#primitivemysqldrbdocf:linbit:drbdparamsdrbd_resource=mydrbdopstarttimeout=240opstoptimeout=100opmonitorrole=Masterinterval=10timeout=20opmonitorrole=Slaveinterval=20timeout=20
登录后复制
注：ocf:资源代理，这里用linbit代理drbd。drbd_resource=mydrbd：drbd的资源名字。starttimeout：启动超时时间。stoptimeout：停止超时时间。monitorrole=Master：定义主节点的监控时间，interval：监控间隔，timeout：超时时间。monitorrole=Slave：定义从节点的监控时间，interval：监控间隔，timeout：超时时间
登录后复制
crm(live)configure#show#查看配置 nodetest1.local nodetest2.local primitivemydrbdocf:linbit:drbd\ paramsdrbd_resource=mydrbd\ opstarttimeout=240interval=0\ opstoptimeout=100interval=0\ opmonitorrole=Masterinterval=10stimeout=20s\ opmonitorrole=Slaveinterval=20stimeout=20s propertycib-bootstrap-options:\ dc-version=1.1.11-97629de\ cluster-infrastructure="classicopenais(withplugin)"\ expected-quorum-votes=2\ stonith-enabled=false\ no-quorum-policy=ignore rsc_defaultsrsc-options:\ resource-stickiness=100 crm(live)configure#verify#验证配置
登录后复制
12.10.2、定义drbd的主从资源
crm(live)configure#msms_mysqldrbdmysqldrbdmetamaster-max=1master-node-max=1clone-max=2clone-node-max=1notify=true 注：msms_mydrbdmydrbd：把mydrbd做成主从，这个资源名称叫ms_mydrbd。meta：定义元数据属性。master-max=1：最多有1个主资源，master-node-max=1：最多有1个主节点，clone-max=2：最多有2个克隆，clone-node-max=1：每个节点上，可以启动几个克隆资源。
登录后复制
crm(live)configure#verify crm(live)configure#commit crm(live)configure#show nodetest1.local nodetest2.local primitivemydrbdocf:linbit:drbd\ paramsdrbd_resource=mydrbd\ opstarttimeout=240interval=0\ opstoptimeout=100interval=0\ opmonitorrole=Masterinterval=10stimeout=20s\ opmonitorrole=Slaveinterval=20stimeout=20s msms_mydrbdmydrbd\ metamaster-max=1master-node-max=1clone-max=2clone-node-max=1 propertycib-bootstrap-options:\ dc-version=1.1.11-97629de\ cluster-infrastructure="classicopenais(withplugin)"\ expected-quorum-votes=2\ stonith-enabled=false\ no-quorum-policy=ignore rsc_defaultsrsc-options:\ resource-stickiness=100
登录后复制
注：可以看到现在配置有一个primitive mydrbd的资源，一个ms的主从类型。
12.10.3、查看集群节点状态
crm(live)configure#cd crm(live)#status Lastupdated:ThuFeb2514:45:522016 Lastchange:ThuFeb2514:44:442016 Stack:classicopenais(withplugin) CurrentDC:test1.local-partitionwithquorum Version:1.1.11-97629de 2Nodesconfigured,2expectedvotes 2Resourcesconfigured Online:[test1.localtest2.local] Fulllistofresources: Master/SlaveSet:ms_mysqldrbd[mysqldrbd] Masters:[test1.local] Slaves:[test2.local]
登录后复制
注：正常状态，已经可以看到主从状态，test1是master。
错误状态：
Master/SlaveSet:ms_mydrbd[mydrbd] mydrbd(ocf::linbit:drbd):FAILEDtest2.local(unmanaged) mydrbd(ocf::linbit:drbd):FAILEDtest1.local(unmanaged) Failedactions: mydrbd_stop_0ontest2.local'notconfigured'(6):call=87,status=complete,last-rc-change='ThuFeb2514:17:342016',queued=0ms,exec=34ms mydrbd_stop_0ontest2.local'notconfigured'(6):call=87,status=complete,last-rc-change='ThuFeb2514:17:342016',queued=0ms,exec=34ms mydrbd_stop_0ontest1.local'notconfigured'(6):call=72,status=complete,last-rc-change='ThuFeb2514:17:342016',queued=0ms,exec=34ms mydrbd_stop_0ontest1.local'notconfigured'(6):call=72,status=complete,last-rc-change='ThuFeb2514:17:342016',queued=0ms,exec=34ms解决办法：定义资源的时候要注意，drbd_resource=mydrbd的名字是drbd资源的名字且主从资源的名称不能和drbd资源的名称一样，还有各种超时设置中不要加s。测试了很久，才找到这个问题。。。。如有不同看法，请各位大神赐教，谢谢。
登录后复制
12.10.4、验证主从的切换
[root@test1~]#crmnodestandbytest1.local#将主节点下线 [root@test1~]#crmstatus注：查看状态，显示主节点已经不在线，而test2成为了master Lastupdated:ThuFeb2514:51:582016 Lastchange:ThuFeb2514:51:442016 Stack:classicopenais(withplugin) CurrentDC:test1.local-partitionwithquorum Version:1.1.11-97629de 2Nodesconfigured,2expectedvotes 2Resourcesconfigured Nodetest1.local:standby Online:[test2.local] Fulllistofresources: Master/SlaveSet:ms_mysqldrbd[mysqldrbd] Masters:[test2.local] Stopped:[test1.local] [root@test1~]#crmnodeonlinetest1.local#重新将test1上线 [root@test1~]#crmstatus#查看状态，显示test依旧为master，而test1成为了slave Lastupdated:ThuFeb2514:52:552016 Lastchange:ThuFeb2514:52:392016 Stack:classicopenais(withplugin) CurrentDC:test1.local-partitionwithquorum Version:1.1.11-97629de 2Nodesconfigured,2expectedvotes 2Resourcesconfigured Online:[test1.localtest2.local] Fulllistofresources: Master/SlaveSet:ms_mysqldrbd[mysqldrbd] Masters:[test2.local] Slaves:[test1.local]
登录后复制
12.10.5、定义文件系统资源
[root@test1~]#crm crm(live)#configure crm(live)configure#primitivemystoreocf:heartbeat:Filesystemparamsdevice=/dev/drbd0directory=/mydrbdfstype=ext3opstarttimeout=60opstoptimeout=60注：这里的语句表示，primitivemystore：定义一个资源mystore，使用heartbeat的文件系统，params参数定义：drbd的设备名，挂在目录，文件系统类型，和启动停止超时时间 crm(live)configure#verify
登录后复制
12.10.6、定义排列约束以确保Filesystem和主节点在一起。
crm(live)configure#colocationmystore_withms_mysqldrbdinf:mystorems_mysqldrbd:Master crm(live)configure#verify
登录后复制
12.10.7、定义Order约束，以确保主从资源必须要先成为主节点以后才能挂在文件系统
crm(live)configure#ordermystore_after_ms_mysqldrbdmandatory:ms_mysqldrbd:promotemystore:start注：这里的语句表示，mystore_after_ms_mysqldrbdmandatory:，mystore在ms_mysqldrbd之后启动，mandatory(代表强制的)，先启动ms_mysqldrbd，promote(角色切换成功后)，在启动mystore:start crm(live)configure#verify crm(live)configure#commit crm(live)configure#cd crm(live)#status#查看状态，可以看到文件系统已经自动挂在到主节点test1.local上了。 Lastupdated:ThuFeb2515:29:392016 Lastchange:ThuFeb2515:29:362016 Stack:classicopenais(withplugin) CurrentDC:test1.local-partitionwithquorum Version:1.1.11-97629de 2Nodesconfigured,2expectedvotes 3Resourcesconfigured Online:[test1.localtest2.local] Fulllistofresources: Master/SlaveSet:ms_mysqldrbd[mysqldrbd] Masters:[test1.local] Slaves:[test2.local] mystore(ocf::heartbeat:Filesystem):Startedtest1.local [root@test1~]#ls/mydrbd/#已经有之前创建的文件了。 inittablost+found
登录后复制
12.10.8、切换主从节点，验证文件系统是否会自动挂载
[root@test1~]#crmnodestandbytest1.local [root@test1~]#crmnodeonlinetest1.local [root@test1~]#crmstatus Lastupdated:ThuFeb2515:32:392016 Lastchange:ThuFeb2515:32:362016 Stack:classicopenais(withplugin) CurrentDC:test1.local-partitionwithquorum Version:1.1.11-97629de 2Nodesconfigured,2expectedvotes 3Resourcesconfigured Online:[test1.localtest2.local] Fulllistofresources: Master/SlaveSet:ms_mysqldrbd[mysqldrbd] Masters:[test2.local] Slaves:[test1.local] mystore(ocf::heartbeat:Filesystem):Startedtest2.local [root@test2~]#ls/mydrbd/ inittablost+found
登录后复制
12.9.12、配置mysql
12.9.12.1、关闭mysqld的开机自启动
[root@test1~]#chkconfigmysqldoff [root@test2~]#chkconfigmysqldoff#一定要记住，只要是高可用集群中的资源的服务，一定不能开机自启动
登录后复制
12.9.12.2、配置主节点1的mysql服务（mysql安装就不写了，直接进入mysql的配置）
[root@test1mysql]#mkdir/mydrbd/data [root@test1mysql]#chown-Rmysql.mysql/mydrbd/data/ [root@test1mysql]#./scripts/mysql_install_db--user=mysql--datadir=/mydrbd/data/--basedir=/usr/local/mysql/ InstallingMySQLsystemtables... 16022516:07:12[Note]/usr/local/mysql//bin/mysqld(mysqld5.5.44)startingasprocess18694... OK Fillinghelptables... 16022516:07:18[Note]/usr/local/mysql//bin/mysqld(mysqld5.5.44)startingasprocess18701... OK Tostartmysqldatboottimeyouhavetocopy support-files/mysql.servertotherightplaceforyoursystem PLEASEREMEMBERTOSETAPASSWORDFORTHEMySQLrootUSER! Todoso,starttheserver,thenissuethefollowingcommands: /usr/local/mysql//bin/mysqladmin-urootpassword'new-password' /usr/local/mysql//bin/mysqladmin-uroot-htest1.localpassword'new-password' Alternativelyyoucanrun: /usr/local/mysql//bin/mysql_secure_installation whichwillalsogiveyoutheoptionofremovingthetest databasesandanonymoususercreatedbydefault.Thisis stronglyrecommendedforproductionservers. Seethemanualformoreinstructions. YoucanstarttheMySQLdaemonwith: cd/usr/local/mysql/;/usr/local/mysql//bin/mysqld_safe& YoucantesttheMySQLdaemonwithmysql-test-run.pl cd/usr/local/mysql//mysql-test;perlmysql-test-run.pl Pleasereportanyproblemsat [root@test1mysql]#servicemysqldstart StartingMySQL.....[OK] [root@test1mysql]#mysql-uroot WelcometotheMySQLmonitor.Commandsendwith;or\g. YourMySQLconnectionidis2 Serverversion:5.5.44Sourcedistribution Copyright(c)2000,2013,Oracleand/oritsaffiliates.Allrightsreserved. OracleisaregisteredtrademarkofOracleCorporationand/orits affiliates.Othernamesmaybetrademarksoftheirrespective owners. Type'help;'or'\h'forhelp.Type'\c'toclearthecurrentinputstatement. mysql>showdatabases; +--------------------+ |Database| +--------------------+ |information_schema| |mysql| |performance_schema| |test| +--------------------+ 4rowsinset(0.06sec) mysql>createdatabasedrbd_mysql; QueryOK,1rowaffected(0.00sec) mysql>showdatabases; +--------------------+ |Database| +--------------------+ |information_schema| |drbd_mysql| |mysql| |performance_schema| |test| +--------------------+ 5rowsinset(0.00sec) mysql>
登录后复制
12.9.12.3、配置从节点的mysql
注：因为刚才已经在test1.local上的共享存储上初始化了mysql的data目录，在test2.local上就不用重复初始化了。
[root@test1mysql]#crmstatus Lastupdated:ThuFeb2516:14:142016 Lastchange:ThuFeb2515:35:162016 Stack:classicopenais(withplugin) CurrentDC:test1.local-partitionwithquorum Version:1.1.11-97629de 2Nodesconfigured,2expectedvotes 3Resourcesconfigured Online:[test1.localtest2.local] Fulllistofresources: Master/SlaveSet:ms_mysqldrbd[mysqldrbd] Masters:[test1.local] Slaves:[test2.local] mystore(ocf::heartbeat:Filesystem):Startedtest1.local
登录后复制
1、先让test2.local成为master才能继续操作
[root@test1mysql]#crmnodestandbytest1.local [root@test1mysql]#crmnodeonlinetest1.local [root@test1mysql]#crmstatus Lastupdated:ThuFeb2516:14:462016 Lastchange:ThuFeb2516:14:302016 Stack:classicopenais(withplugin) CurrentDC:test1.local-partitionwithquorum Version:1.1.11-97629de 2Nodesconfigured,2expectedvotes 3Resourcesconfigured Nodetest1.local:standby Online:[test2.local] Fulllistofresources: Master/SlaveSet:ms_mysqldrbd[mysqldrbd] Masters:[test2.local] Stopped:[test1.local] mystore(ocf::heartbeat:Filesystem):Startedtest2.local#要确保让test2.local成为master节点
登录后复制
2、test2.local的mysql配置
[root@test2~]#vim/etc/my.cnf 添加: datadir=/mydrbd/data [root@test2~]#servicemysqldstart StartingMySQL.[OK] [root@test2~]#mysql-uroot WelcometotheMySQLmonitor.Commandsendwith;or\g. YourMySQLconnectionidis1 Serverversion:5.5.44Sourcedistribution Copyright(c)2000,2013,Oracleand/oritsaffiliates.Allrightsreserved. OracleisaregisteredtrademarkofOracleCorporationand/orits affiliates.Othernamesmaybetrademarksoftheirrespective owners. Type'help;'or'\h'forhelp.Type'\c'toclearthecurrentinputstatement. mysql>showdatabases; +--------------------+ |Database| +--------------------+ |information_schema| |drbd_mysql| |mysql| |performance_schema| |test| +--------------------+ 5rowsinset(0.10sec) mysql>
登录后复制
12.9.13、定义mysql资源
1、停止mysql
[root@test2~]#servicemysqldstop ShuttingdownMySQL.[OK]
登录后复制
2、定义mysql资源
[root@test1mysql]#crmconfigure crm(live)configure#primitivemysqldlsb:mysqld crm(live)configure#verify
登录后复制
3、定义mysql和主节点约束
crm(live)configure#colocationmysqld_with_mystoreinf:mysqldmystore#注：因为mystore一定和主节点在一起，那么我们就定义mysql和mystore的约束。 crm(live)configure#verify
登录后复制
4、定义mysql和mystore启动次序约束
crm(live)configure#ordermysqld_after_mystoremandatory:mystoremysqld#一定要弄清楚启动的先后次序，mysql是在mystore之后启动的。 crm(live)configure#verify crm(live)configure#commit crm(live)configure#cd crm(live)#status Lastupdated:ThuFeb2516:44:292016 Lastchange:ThuFeb2516:42:162016 Stack:classicopenais(withplugin) CurrentDC:test1.local-partitionwithquorum Version:1.1.11-97629de 2Nodesconfigured,2expectedvotes 4Resourcesconfigured Online:[test1.localtest2.local] Fulllistofresources: Master/SlaveSet:ms_mysqldrbd[mysqldrbd] Masters:[test2.local] Slaves:[test1.local] mystore(ocf::heartbeat:Filesystem):Startedtest2.local mysqld(lsb:mysqld):Startedtest2.local注：现在主节点在test2.local上。
登录后复制
5、验证test2.local上的mysql登录是否正常和角色切换后是否正常
[root@test2~]#mysql-uroot WelcometotheMySQLmonitor.Commandsendwith;or\g. YourMySQLconnectionidis2 Serverversion:5.5.44Sourcedistribution Copyright(c)2000,2013,Oracleand/oritsaffiliates.Allrightsreserved. OracleisaregisteredtrademarkofOracleCorporationand/orits affiliates.Othernamesmaybetrademarksoftheirrespective owners. Type'help;'or'\h'forhelp.Type'\c'toclearthecurrentinputstatement. mysql>showdatabases;#test2.local上已经启动好了mysql并自动挂在了drbd资源 +--------------------+ |Database| +--------------------+ |information_schema| |drbd_mysql| |mysql| |performance_schema| |test| +--------------------+ 5rowsinset(0.07sec) mysql>createdatabasemydb;#创建一个数据库，然后切换到test1.local节点上，看是否正常 QueryOK,1rowaffected(0.00sec) mysql>showdatabases; +--------------------+ |Database| +--------------------+ |information_schema| |drbd_mysql| |mydb| |mysql| |performance_schema| |test| +--------------------+ 6rowsinset(0.00sec) mysql>exit [root@test1mysql]#crmnodestandbytest2.local#将test2.local主节点standby，让test1.local自动成为master [root@test1mysql]#crmnodeonlinetest2.local [root@test1mysql]#crmstatus Lastupdated:ThuFeb2516:53:242016 Lastchange:ThuFeb2516:53:192016 Stack:classicopenais(withplugin) CurrentDC:test1.local-partitionwithquorum Version:1.1.11-97629de 2Nodesconfigured,2expectedvotes 4Resourcesconfigured Online:[test1.localtest2.local] Fulllistofresources: Master/SlaveSet:ms_mysqldrbd[mysqldrbd] Masters:[test1.local] Slaves:[test2.local] mystore(ocf::heartbeat:Filesystem):Startedtest1.local#test1.local已经成为master mysqld(lsb:mysqld):Startedtest1.local [root@test1mysql]#mysql-uroot#在test1.local上登录mysql WelcometotheMySQLmonitor.Commandsendwith;or\g. YourMySQLconnectionidis1 Serverversion:5.5.44Sourcedistribution Copyright(c)2000,2013,Oracleand/oritsaffiliates.Allrightsreserved. OracleisaregisteredtrademarkofOracleCorporationand/orits affiliates.Othernamesmaybetrademarksoftheirrespective owners. Type'help;'or'\h'forhelp.Type'\c'toclearthecurrentinputstatement. mysql>showdatabases; +--------------------+ |Database| +--------------------+ |information_schema| |drbd_mysql| |mydb|#已经有刚才在test2.local上创建的mydb数据库。目前，一切正常 |mysql| |performance_schema| |test| +--------------------+ 6rowsinset(0.15sec) mysql>
登录后复制
12.9.14、定义VIP资源及约束关系
crm(live)configure#primitivemyipocf:heartbeat:IPaddrparamsip=192.168.10.3nic=eth0cidr_netmask=24#这里出了一个错误，浪费半天时间，是因为子网掩码写的255.255.255.0，应该写成24，忘记命令的时候，要多使用table和help。 crm(live)configure#verify crm(live)configure#colocationmyip_with_ms_mysqldrbdinf:ms_mysqldrbd:Mastermyip#定义vip和ms_mysqldrbd的约束关系 crm(live)configure#verify crm(live)configure#commit crm(live)configure#cd crm(live)#status#查看状态 Lastupdated:FriFeb2610:05:162016 Lastchange:FriFeb2610:05:122016 Stack:classicopenais(withplugin) CurrentDC:test1.local-partitionwithquorum Version:1.1.11-97629de 2Nodesconfigured,2expectedvotes 5Resourcesconfigured Online:[test1.localtest2.local] Fulllistofresources: Master/SlaveSet:ms_mysqldrbd[mysqldrbd] Masters:[test1.local] Slaves:[test2.local] mystore(ocf::heartbeat:Filesystem):Startedtest1.local mysqld(lsb:mysqld):Startedtest1.local myip(ocf::heartbeat:IPaddr):Startedtest1.local#vip已经启动在test1.local节点上了。
登录后复制
12.9.15、测试连接VIP
[root@test1~]#ipaddr#查看test1.local上是否绑定VIP 1:lo:mtu65536qdiscnoqueuestateUNKNOWN link/loopback00:00:00:00:00:00brd00:00:00:00:00:00 inet127.0.0.1/8scopehostlo inet6::1/128scopehost valid_lftforeverpreferred_lftforever 2:eth0:mtu1500qdiscpfifo_faststateUPqlen1000 link/ether00:0c:29:34:7d:9fbrdff:ff:ff:ff:ff:ff inet192.168.10.55/24brd192.168.10.255scopeglobaleth0 inet192.168.10.3/24brd192.168.10.255scopeglobalsecondaryeth0#VIP已经绑定到test1.local上的eth0接口上 inet6fe80::20c:29ff:fe34:7d9f/64scopelink valid_lftforeverpreferred_lftforever [root@test-3~]#mysql-uroot-h192.168.10.3-p#使用VIP连接mysql，这里要给连接的客户端授权，不然不能登录 Enterpassword: WelcometotheMySQLmonitor.Commandsendwith;or\g. YourMySQLconnectionidis6 Serverversion:5.5.44Sourcedistribution Copyright(c)2000,2013,Oracleand/oritsaffiliates.Allrightsreserved. OracleisaregisteredtrademarkofOracleCorporationand/orits affiliates.Othernamesmaybetrademarksoftheirrespective owners. Type'help;'or'\h'forhelp.Type'\c'toclearthecurrentinputstatement. mysql>showdatabases;#下面有我们刚才创建的两个库 +--------------------+ |Database| +--------------------+ |information_schema| |drbd_mysql| |mydb| |mysql| |performance_schema| |test| +--------------------+ 6rowsinset(0.08sec)
登录后复制
模拟test1.local故障：
[root@test1mysql]#crmnodestandbytest1.local#将test1.local降级为slave，看vip是否会自动切换 [root@test1mysql]#crmnodeonlinetest1.local [root@test1mysql]#crmstatus Lastupdated:FriFeb2610:20:382016 Lastchange:FriFeb2610:20:352016 Stack:classicopenais(withplugin) CurrentDC:test1.local-partitionwithquorum Version:1.1.11-97629de 2Nodesconfigured,2expectedvotes 5Resourcesconfigured Online:[test1.localtest2.local] Fulllistofresources: Master/SlaveSet:ms_mysqldrbd[mysqldrbd] Masters:[test2.local] Slaves:[test1.local] mystore(ocf::heartbeat:Filesystem):Startedtest2.local mysqld(lsb:mysqld):Startedtest2.local myip(ocf::heartbeat:IPaddr):Startedtest2.local#vip已经在test2.local节点上了。
登录后复制
查看test2的IP信息:
[root@test2~]#ipaddr#在test2.local上查看VIP绑定信息 1:lo:mtu65536qdiscnoqueuestateUNKNOWN link/loopback00:00:00:00:00:00brd00:00:00:00:00:00 inet127.0.0.1/8scopehostlo inet6::1/128scopehost valid_lftforeverpreferred_lftforever 2:eth0:mtu1500qdiscpfifo_faststateUPqlen1000 link/ether00:0c:29:fd:7f:e5brdff:ff:ff:ff:ff:ff inet192.168.10.56/24brd192.168.10.255scopeglobaleth0 inet192.168.10.3/24brd192.168.10.255scopeglobalsecondaryeth0#VIP已经在test2.local上的eth0接口上。 inet6fe80::20c:29ff:fefd:7fe5/64scopelink valid_lftforeverpreferred_lftforever
登录后复制
测试连接MySQL:
[root@test-3~]#mysql-uroot-h192.168.10.3-p#连接mysql Enterpassword: WelcometotheMySQLmonitor.Commandsendwith;or\g. YourMySQLconnectionidis1 Serverversion:5.5.44Sourcedistribution Copyright(c)2000,2013,Oracleand/oritsaffiliates.Allrightsreserved. OracleisaregisteredtrademarkofOracleCorporationand/orits affiliates.Othernamesmaybetrademarksoftheirrespective owners. Type'help;'or'\h'forhelp.Type'\c'toclearthecurrentinputstatement. mysql>showdatabases;#一切正常。 +--------------------+ |Database| +--------------------+ |information_schema| |drbd_mysql| |mydb| |mysql| |performance_schema| |test| +--------------------+ 6rowsinset(0.06sec)
登录后复制
到此，corosync+drbd+mysql已经配置完毕，文档不够详细，没有corosync、heartbeat、drbd、mysql结合的原理讲解，没有发生脑裂后的处理办法和预防方法。以后有时间在加上吧。
以上就是corosync+drbd+mysql实现的高可用_MySQL的内容，更多相关内容请关注Work网（www.php.cn）！