HA高可用集群部署（ricci+luci+fence) 双机热备

Posted 2020-08-11

主机环境 redhat6.5 6位

实验环境 服务端1 ip172.25.29.1   主机名：server1.example.com   ricci

    服务端2 ip172.25.29.2    主机名：server2.example.com    ricci

    管理端1 ip172.25.29.3    主机名：server3.example.com    luci

    管理端2 ip172.25.29.250     fence_virtd

防火墙状态：关闭

 

1. 安装ricci、luci及创建节点

1.安装、开启ricci（服务端1）

[[email protected]]# vim dvd.repo      #安装之前，修改yum源

#repos on instructor for classroom use

#Main rhel6.5 server

[base]

name=InstructorServer Repository

baseurl=http://172.25.29.250/rhel6.5

gpgcheck=0

 

#HighAvailability rhel6.5

[HighAvailability]

name=InstructorHighAvailability Repository

baseurl=http://172.25.29.250/rhel6.5/HighAvailability

gpgcheck=0

 

#LoadBalancer packages

[LoadBalancer]

name=InstructorLoadBalancer Repository

baseurl=http://172.25.29.250/rhel6.5/LoadBalancer

gpgcheck=0

 

#ResilientStorage

[ResilientStorage]

name=InstructorResilientStorage Repository

baseurl=http://172.25.29.250/rhel6.5/ResilientStorage

gpgcheck=0

 

#ScalableFileSystem

[ScalableFileSystem]

name=InstructorScalableFileSystem Repository

baseurl=http://172.25.29.250/rhel6.5/ScalableFileSystem

gpgcheck=0

[[email protected]]# yum clean all        #清除缓存

Loadedplugins: product-id, subscription-manager

Thissystem is not registered to Red Hat Subscription Management. You can usesubscription-manager to register.

Cleaningrepos: HighAvailability LoadBalancer ResilientStorage

              : ScalableFileSystem base

Cleaningup Everything

 

[[email protected]]# yum install ricci -y    #安装ricci

[[email protected]]# passwd ricci       #修改ricci密码

Changingpassword for user ricci.

Newpassword:

BADPASSWORD: it is based on a dictionary word

BADPASSWORD: is too simple

Retypenew password:

passwd:all authentication tokens updated successfully.

[[email protected]]# /etc/init.d/ricci start      #开启ricci

Startingsystem message bus:                               [  OK  ]

Startingoddjobd:                                         [  OK  ]

generatingSSL certificates...  done

GeneratingNSS database...  done

Startingricci:                                           [  OK  ]

[[email protected]]# chkconfig ricci   on     #开机自动开启

服务端2和服务端1做相同的操作

 

2.安装、开启luci （管理端1）

安装之前，与服务端1一样修改yum源

[[email protected]]# yum install luci -y       #安装luci

[[email protected]]# /etc/init.d/luci start     #开启ruci

Startluci...                                             [  OK  ]

Pointyour web browser to https://server3.example.com:8084 (or equivalent) to accessluci

 

在登陆之前，必须有DNS解析，也就是在/etc/hosts添加

例如: 172.25.29.3         server3.example.com

 

3．创建节点

登陆 https://server3.example.com:8084       #luci开放的是8084端口

安全证书，选I Understand Risks

 

点击Confirm Security Excepton

进入到管理服务器的luci界面，登陆时的密码是安装luci虚拟机的root密码

选择Manage Clusters，之后点击Create创建集群

如图，Cluster Name创建集群的名称，勾选Use the Same Passwordfor All Nodes，指的是所有结点所用的是相同的密码，填写要创建的结点名称和密码，名称是服务端的主机名，密码是上面提到的passwd ricci的修改的密码。勾选Download PackagesReboot和Enable，选择Create Cluster

正在创建节点，如图

创建完成，如图

创建完成之后，在服务端1和服务端2的/etc/cluster/下会生成cluster.conf文件，查看如下

[[email protected]~]# cd /etc/cluster/

[[email protected]]# ls

cluster.conf  cman-notify.d

[[email protected]]# cat cluster.conf   #查看文件内容

<?xmlversion="1.0"?>

<clusterconfig_version="1" name="wen">        #集群名称

    <clusternodes>

       <clusternodename="server1.example.com" nodeid="1"/>     #结点1

       <clusternodename="server2.example.com" nodeid="2"/>     #结点2

    </clusternodes>

    <cman expected_votes="1"two_node="1"/>

    <fencedevices/>

    <rm/>

</cluster>

 

2.安装fence_virtd、创建fence设备

1.安装、开启fence_virtd（管理端2）

[[email protected]]# yum install fence-virtd* -y

[[email protected]]# fence_virtd -c    #设置

Modulesearch path [/usr/lib64/fence-virt]:   

 

Availablebackends:

    libvirt 0.1

Availablelisteners:

    multicast 1.2

    serial 0.4

 

Listenermodules are responsible for accepting requests

fromfencing clients.

 

Listenermodule [multicast]:        #多播

 

Themulticast listener module is designed for use environments

wherethe guests and hosts may communicate over a network using

multicast.

 

Themulticast address is the address that a client will use to

sendfencing requests to fence_virtd.

 

MulticastIP Address [225.0.0.12]:             #多播ip

 

Usingipv4 as family.

 

MulticastIP Port [1229]:      #多播端口号

 

Settinga preferred interface causes fence_virtd to listen only

onthat interface.  Normally, it listens onall interfaces.

Inenvironments where the virtual machines are using the host

machineas a gateway, this *must* be set (typically to virbr0).

Setto ‘none‘ for no interface.

 

Interface[br0]:   

 

Thekey file is the shared key information which is used to

authenticatefencing requests.  The contents of thisfile must

bedistributed to each physical host and virtual machine within

acluster.

 

KeyFile [/etc/cluster/fence_xvm.key]:     #key文件的路径

 

Backendmodules are responsible for routing requests to

theappropriate hypervisor or management layer.

 

Backendmodule [libvirt]:

 

Configurationcomplete.

 

===Begin Configuration ===

backends{

    libvirt {

       uri = "qemu:///system";

    }

 

}

listeners{

    multicast {

       port = "1229";

       family = "ipv4";

       interface = "br0";

       address = "225.0.0.12";

       key_file ="/etc/cluster/fence_xvm.key";

    }

 

}

fence_virtd{

    module_path ="/usr/lib64/fence-virt";

    backend = "libvirt";

    listener = "multicast";

}

 

===End Configuration ===

Replace/etc/fence_virt.conf with the above [y/N]? y

[[email protected]]# mkdir /etc/cluster     #创建cluster目录

[[email protected]]# cd /etc/cluster/

[[email protected]]# dd if=/dev/urandom of=fence_xvm.key bs=128 count=1    #生成文件

1+0records in

1+0records out

128bytes (128 B) copied, 0.000167107 s, 766 kB/s

[[email protected]]# scp fence_xvm.key [email protected]:/etc/cluster/  #将文件传给服务端1

[email protected]‘spassword:

fence_xvm.key                                 100%  512    0.5KB/s   00:00 

 

测试

[[email protected]]# ls     #查看

cluster.conf  cman-notify.d fence_xvm.key

以同样的方法传送给服务端2

[[email protected]]# systemctl start fence_fence_virtd.service     #开启fence（由于管理端2是7.1的系统，开启时的命令不太一。如果是6.5系统，则用/etc/init.d/fence_virtd start即可）

 

2.创建fence设备

选择Fance Devices

选择Add,如下图，Name指的是添加fence设备的名称，写完之后选择Submit

结果

选择server1.example.com

点击Add Fence Method ,添加Method Name

如图，选择Add Tence Instance

填写Domin，选择Submit

完成之后，服务端2和服务端1的配置相同

 

测试

[[email protected]]# fence_node server2.example.com  #测试server2的结点

fenceserver2.example.com success

[[email protected]]# fence_node server1.example.com   #测试server1的结点

fenceserver1.example.com success

 

3.在搭建好的集群上添加服务（双机热备），以apche为例

1.添加服务  这里采用的是双机热备

选择Failover Domains,如图，填写Name，如图选择,前面打勾的三个分别是结点失效之后可以跳到另一个结点、只服务运行指定的结点、当结点失效之跳到另一个结点之后，原先的结点恢复之后，不会跳回原先的结点。下面的Member打勾，是指服务运行server1.example.com和server2.exampe.com结点，后面的Priority值越小，优先级越高，选择Creale

选择Resourcs，点击Add，选择添加IPAddress如图，添加的ip必须是未被占用的ip，24是子网掩码的位数，10指的是等待时间为10秒。选择Submit

以相同的方法添加Script，httpd是服务的名字，/etc/init.d/httpd是服务启动脚本的路径，选择Submit

选择Service Groups,点击Add如图，apache是服务的名字，下面两个勾指分别的是

自动开启服务、运行 ，选择Add Resource

选择172.25.29.100/24之后，如图点击Add Child Resource

选择httpd之后，如图选择Submit，完成

 

2.测试

在测试之前 server1和server2必须安装httpd，并且有测试页。注意：不要开启httpd服务，在访问的时候，会自动开启（如果在访问之前开启了服务，访问的时候会报错）

测试 172.25.29.100

[[email protected]~]# ip addr show   #查看

1:lo: <LOOPBACK,UP,LOWER_UP> mtu 16436 qdisc noqueue state UNKNOWN

    link/loopback 00:00:00:00:00:00 brd00:00:00:00:00:00

    inet 127.0.0.1/8 scope host lo

    inet6 ::1/128 scope host

       valid_lft forever preferred_lft forever

2:eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast stateUP qlen 1000

    link/ether 52:54:00:94:2f:4f brdff:ff:ff:ff:ff:ff

    inet 172.25.29.1/24 brd 172.25.29.255 scopeglobal eth0

    inet 172.25.29.100/24 scope globalsecondary eth0        #自动添加了ip 172.25.29.100

    inet6 fe80::5054:ff:fe94:2f4f/64 scope link

       valid_lft forever preferred_lft forever

[[email protected]~]# clustat      #查看服务

ClusterStatus for wen @ Tue Sep 27 18:12:38 2016

MemberStatus: Quorate

 

 Member Name                             ID   Status

 ------ ----                             ---- ------

 server1.example.com                         1 Online, Local,rgmanager

 server2.example.com                         2 Online, rgmanager

 

 Service Name                   Owner (Last)                   State        

 ------- ----                   ----- ------                   -----        

 service:apache                 server1.example.com            started    #serve1在服务      

[[email protected]~]# /etc/init.d/network stop    #当网络断开之后，fence控制server1自动断电，然后启动；服务转到server2

 

测试

[[email protected]~]# ip addr show      #查看

1:lo: <LOOPBACK,UP,LOWER_UP> mtu 16436 qdisc noqueue state UNKNOWN

    link/loopback 00:00:00:00:00:00 brd00:00:00:00:00:00

    inet 127.0.0.1/8 scope host lo

    inet6 ::1/128 scope host

       valid_lft forever preferred_lft forever

2:eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast stateUP qlen 1000

    link/ether 52:54:00:23:81:98 brdff:ff:ff:ff:ff:ff

    inet172.25.29.2/24 brd 172.25.29.255 scope global eth0

    inet 172.25.29.100/24 scope globalsecondary eth0     #自动添加

    inet6 fe80::5054:ff:fe23:8198/64 scope link

       valid_lft forever preferred_lft forever

 

 [[email protected]~]# clustat       #查看服务

ClusterStatus for wen @ Tue Sep 27 18:22:20 2016

MemberStatus: Quorate

 Member Name                             ID   Status

 ------ ----                             ---- ------

 server1.example.com                         1 Online, Local,rgmanager

 server2.example.com                         2 Online, rgmanager

 Service Name                   Owner (Last)                   State        

 ------- ----                   ----- ------                   ------        

 service:apache                 server2.example.com            started    #server2服务 

本文出自 “不忘初心，方得始终” 博客，请务必保留此出处http://12087746.blog.51cto.com/12077746/1857811