Heartbeat+DRBD+NFS高可用案例

9.4 部署DRBD 的需求描述

9.4.1业务需求描述

假设两台服务器Rserver-1/Lserver-1, 其实际IP分别为192.168.236.143（Rserver）和192.168.236.192（Lserver）

配置目标：两台服务器分别配置好DRBD服务后、实现在Rserver-1机器上/dev/sdb 分区上写入数据、数据会时时的同步到Lserver-1机器上面、一旦服务器Rserver-1机器宕机或硬盘损坏导致数据不可用、Lserver-1机器上的数据此时是picdata-1机器的一个完整备份、当然、不光是一个完整的备份、还可以瞬间接替坏数据或宕机的Rserver-1机器上数据的异机时时同步、从而达到数据高可用无业务影响的目的

9.4.2 DRBD部署结构图

1、   Drbd服务通过直连或以太网实时互相数据同步、

2、   两台存储服务器互相备份、正常情况下两端各提供一个主分区提供NFS使用

3、   存储服务器之间、存储服务器和交换机之间都是双千兆网卡绑定（bonding）

4、   应用服务器通过NFS访问存储

9.4.3服务主机资源规划

9.4.5 drbd的环境配置

设置hosts文件两台都配置注意这里是主机名也需要改成picadata-1-1 是主机名需要改

例如：hostname picadata-1-1 如果这步没有操作启动服务的时候会出现报错。

echo ‘172.16.1.1 Rserver-1‘>>/etc/hosts

echo ‘172.16.1.2 Lserver-1‘>>/etc/hosts

[[email protected] ~]# tail -2 /etc/hosts

172.16.1.2 Rserver-1

172.16.1.1 Lserver-1

8.3.3配置服务器间心跳连接：

192.168.1.1 和192.168.1.2 两块网卡之间是通过普通网线直连连接的、即不通过交换机、直接把两块网卡连接在一起用于做心跳检测

Master:

ifconfig eth2 192.168.1.1 netmask255.255.255.0

Backup:

ifconfig eth2 192.168.1.2 netmask255.255.255.0

Rserver-1 server 上添加如下主机路由

route add –host 192.168.1.2 dev eth2

####这条命令是：从picdata-1-1server 访问192.168.1.2 走网卡eth2出去、作为心跳线路

echo ‘route add -host 192.168.1.2 deveth2‘ >>/etc/rc.local

##-à加入开机自启动配置里、这样下次启动后就会自动加载这个路由配置。

route -n

Lserver-1 server 上添加如下主机路由

route add –host 192.168.1.2 dev eth2

####这条命令是：从picdata-1-2server 访问192.168.1.2 走网卡eth2出去、作为心跳线路

echo ‘route add -host 192.168.1.1 deveth2‘ >>/etc/rc.local

##-à加入开机自启动配置里、这样下次启动后就会自动加载这个路由配置。

9.5 开始实施部署

9.5.1硬盘进行分区

首先，通过fdisk，mkfs，ext3，tune2fs 等命令、对硬盘进行分区、分区信息如下表

提示：如果生产环境中单个硬盘和raid的硬盘大于2Tfdisk 命令是查看不到的。

在虚拟机中添加两块硬盘。后面查看一下

Rserver-1查看

[[email protected] ~]# fdisk -l

Disk /dev/sda: 21.5 GB, 21474836480 bytes

255 heads, 63 sectors/track, 2610 cylinders

Units = cylinders of 16065 * 512 = 8225280 bytes

Sector size (logical/physical): 512 bytes / 512 bytes

I/O size (minimum/optimal): 512 bytes / 512 bytes

Disk identifier: 0x000486f5

DeviceBoot      Start         End      Blocks  Id  System

/dev/sda1  *           1         64      512000   83 Linux

Partition 1 does not end on cylinder boundary.

/dev/sda2             64        2611    20458496  8e  Linux LVM

Disk /dev/sdb: 21.5 GB, 21474836480 bytes

Lserver-1查看

[[email protected] ~]# fdisk -l

Disk /dev/sda: 21.5 GB, 21474836480 bytes

255 heads, 63 sectors/track, 2610 cylinders

Units = cylinders of 16065 * 512 = 8225280 bytes

Sector size (logical/physical): 512 bytes / 512 bytes

I/O size (minimum/optimal): 512 bytes / 512 bytes

Disk identifier: 0x00087dae

DeviceBoot      Start         End      Blocks  Id  System

/dev/sda1  *           1          64      512000  83  Linux

Partition 1 does not end on cylinder boundary.

/dev/sda2             64        2611   20458496   8e  Linux LVM

Disk /dev/sdb: 21.5 GB, 21474836480 bytes

255 heads, 63 sectors/track, 2610 cylinders

Units = cylinders of 16065 * 512 = 8225280 bytes

9.5.2在master和backup中做分区操作（注：两台一样）

因此、我们需要做的就是对/dev/sdb 进行分区、需要分区具体内容见下表

[[email protected] ~]# fdisk /dev/sdb

Device contains neither a valid DOSpartition table, nor Sun, SGI or OSF disklabel

Building a new DOS disklabel with diskidentifier 0x95767900.

Changes will remain in memory only,until you decide to write them.

After that, of course, the previouscontent won‘t be recoverable.

Warning: invalid flag 0x0000 of partitiontable 4 will be corrected by w(rite)

WARNING: DOS-compatible mode isdeprecated. It‘s strongly recommended to

switch off the mode (command ‘c‘) and change display units to

sectors (command ‘u‘).

Command (m for help): n

Command action

e   extended

p   primary partition (1-4)

p   ####新建一个主分区

Partition number (1-4): 1

First cylinder (1-2610, default 1):

Using default value 1

Last cylinder, +cylinders or+size{K,M,G} (1-2610, default 2610): +500M  ####大小为500M

Command (m for help): p

Disk /dev/sdb: 21.5 GB, 21474836480bytes

255 heads, 63 sectors/track, 2610cylinders

Units = cylinders of 16065 * 512 =8225280 bytes

Sector size (logical/physical): 512bytes / 512 bytes

I/O size (minimum/optimal): 512 bytes /512 bytes

Disk identifier: 0x95767900

Device Boot      Start         End      Blocks  Id  System

/dev/sdb1               1          65      522081  83  Linux

Command (m for help): n

Command action

e   extended

p   primary partition (1-4)

p

Partition number (1-4): 2

First cylinder (66-2610, default 66):

Using default value 66

Last cylinder, +cylinders or+size{K,M,G} (66-2610, default 2610): +200M ####新建一个200M

Command (m for help): p

Disk /dev/sdb: 21.5 GB, 21474836480bytes

255 heads, 63 sectors/track, 2610cylinders

Units = cylinders of 16065 * 512 =8225280 bytes

Sector size (logical/physical): 512bytes / 512 bytes

I/O size (minimum/optimal): 512 bytes /512 bytes

Disk identifier: 0x95767900

Device Boot      Start         End      Blocks  Id  System

/dev/sdb1               1          65      522081  83  Linux

/dev/sdb2              66          91      208845  83  Linux

Command (m for help):w  ######表示保存

如果提示

the kernel still uses the old table

The new table will be used at next reboot

上面这句话的意思是内核还不知道你做了分区，需要重启才能让内核知道，可以用如下命令让内核知晓

partprobe

现在查看一下分区的结果

[[email protected] ~]# fdisk -l

Disk /dev/sda: 21.5 GB, 21474836480bytes

255 heads, 63 sectors/track, 2610cylinders

Units = cylinders of 16065 * 512 =8225280 bytes

Sector size (logical/physical): 512bytes / 512 bytes

I/O size (minimum/optimal): 512 bytes /512 bytes

Disk identifier: 0x00087dae

Device Boot      Start         End      Blocks  Id  System

/dev/sda1   *          1          64      512000  83  Linux

Partition 1 does not end on cylinderboundary.

/dev/sda2              64        2611   20458496   8e  Linux LVM

Disk /dev/sdb: 21.5 GB, 21474836480bytes

255 heads, 63 sectors/track, 2610cylinders

Units = cylinders of 16065 * 512 =8225280 bytes

Sector size (logical/physical): 512bytes / 512 bytes

I/O size (minimum/optimal): 512 bytes /512 bytes

Disk identifier: 0x95767900

Device Boot      Start         End      Blocks  Id  System

/dev/sdb1               1          65      522081  83  Linux

/dev/sdb2              66          91      208845  83  Linux

现在对数据分区格式化

[[email protected]Rserver-1~]# mkfs.ext4 /dev/sdb1

[[email protected]Lserver-1~]# mkfs.ext4 /dev/sdb1

[[email protected]Rserver-1~]# tune2fs -c -1 /dev/sdb1

tune2fs 1.41.12(17-May-2010)

Setting maximal mount count to -1   ####设置最大挂载次数为-1

[[email protected]Lserver-1~]# tune2fs -c -1 /dev/sdb1

tune2fs 1.41.12(17-May-2010)

Setting maximal mount count to -1  ####设置最大挂载次数为-1

9.6、安装前准备：(Rserver-1,Lserver-1)
1、关闭iptables和SELINUX，避免安装过程中报错。

# service iptables stop

# chkconfig iptables off

# setenforce 0

# vi /etc/selinux/config

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

SELINUX=disabled

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

9.6.1时间同步：

ntpdate -u asia.pool.ntp.org

9.6.2 DRBD的安装配置：

# yum install gcc gcc-c++ make glibcflex kernel-develkernel-headers  这两个的安装包一定要和uname –r 的版本一定是需要一样的。不然后面不能把drbd 加入到内核当中。可以用本地yum的方式安装。

9.6.3安装DRBD：(Rserver-1主,Lserver-1备)

# wget http://oss.linbit.com/drbd/8.4/drbd-8.4.2.tar.gz

# tar zxvf drbd-8.4.3.tar.gz

# cd drbd-8.4.3

# ./configure --prefix=/usr/local/drbd--with-km --with-heartbeat --sysconfdir=/etc/

# make KDIR=/usr/src/kernels/2.6.32-504.16.2.el6.x86_64/

# make install

# mkdir -p /usr/local/drbd/var/run/drbd

# chkconfig --add drbd

# chkconfig drbd on

2、加载DRBD模块:(Rserver-1主,Lserver-1备)

# modprobe drbd

查看DRBD模块是否加载到内核：

# lsmod |grep drbd

drbd 310172 4

libcrc32c 1246 1 drbd

3、参数配置：(Rserver-1主,Lserver-1备)

vi /etc/drbd.conf

清空文件内容，并添加如下配置：

resource r0{

protocol C;

startup { wfc-timeout 0;degr-wfc-timeout 120;}

disk { on-io-error detach;}

net{

timeout 60;

connect-int 10;

ping-int 10;

max-buffers 2048;

max-epoch-size 2048;

}

syncer { rate 200M;}

on Rserver-1{     #######on 后面是主机名

device /dev/drbd0;        #####指定的是一个drbd一个盘

disk /dev/sdb1;         #####本地磁盘。就是上面分区好的硬盘

address 172.16.1.1:7788;       ######内网IP

meta-disk internal;

}

on Lserver-1{

device /dev/drbd0;

disk /dev/sdb1;

address 172.16.1.2:7788;

meta-disk internal;

}

}

注：请修改上面配置中的主机名、IP、和disk为自己的具体配置

4，创建DRBD设备并激活r0资源：(Rserver-1主,Lserver-1备)

# mknod /dev/drbd0 b 147 0

# drbdadm create-md r0

等待片刻，显示success表示drbd块创建成功

Writing meta data...

initializing activity log

NOT initializing bitmap

New drbd meta data block successfullycreated.

--== Creating metadata ==--

As with nodes, we count the total numberof devices mirrored by DRBD

at http://usage.drbd.org.

The counter works anonymously. Itcreates a random number to identify

the device and sends that random number,along with the kernel and

DRBD version, to usage.drbd.org.

http://usage.drbd.org/cgi-bin/insert_usage.pl?

nu=716310175600466686&ru=15741444353112217792&rs=1085704704

* If you wish to opt out entirely,simply enter ‘no‘.

* To continue, just press [RETURN]

success

再次输入该命令：

# drbdadm create-md r0

成功激活r0

[need to type ‘yes‘ to confirm] yes

Writing meta data...

initializing activity log

NOT initializing bitmap

New drbd meta data block successfullycreated.

5、启动DRBD服务：(Rserver-1主,Lserver-1备)

servicedrbd start

注：需要主从共同启动方能生效

6、查看状态：(Rserver-1主,Lserver-1备)

# service drbd status

drbd driver loaded OK; device status:

version: 8.4.3(api:1/proto:86-101)

GIT-hash:89a294209144b68adb3ee85a73221f964d3ee515 build by [email protected],2015-05-12 21:05:41

m:res cs         ro                   ds                         p  mounted    fstype

0:r0  Connected Secondary/Secondary Inconsistent/Inconsistent  C

这里ro:Secondary/Secondary表示两台主机的状态都是备机状态，ds是磁盘状态，显示的状态内容为“Inconsistent不一致”，这是因为DRBD无法判断哪一方为主机，应以哪一方的磁盘数据作为标准。
7、将drbd1.example.com主机配置为主节点(Rserver-1)

# drbdsetup /dev/drbd0 primary --force

8、分别查看主从DRBD状态：(Rserver-1)主

# service drbd status

drbd driver loaded OK; device status:

version: 8.4.3(api:1/proto:86-101)

GIT-hash:89a294209144b68adb3ee85a73221f964d3ee515 build by [email protected], 2017-05-1813:40:26

m:res cs         ro                 ds                 p  mounted    fstype

0:r0  Connected  Primary/Secondary  UpToDate/UpToDate  C

（Lserver-1)备

# service drbd status

drbd driver loaded OK; device status:

version: 8.4.3(api:1/proto:86-101)

GIT-hash:89a294209144b68adb3ee85a73221f964d3ee515 build by [email protected], 2017-05-1813:38:57

m:res cs         ro                 ds                 p  mounted    fstype

0:r0  Connected  Secondary/Primary  UpToDate/UpToDate  C

ro在主从服务器上分别显示 Primary/Secondary和Secondary/Primary
ds显示UpToDate/UpToDate
表示主从配置成功。
9、挂载DRBD：(Rserver-1)主
从刚才的状态上看到mounted和fstype参数为空，所以我们这步开始挂载DRBD到系统目录/store

# mkfs.ext4 /dev/drbd0

# mkdir /data

# mount /dev/drbd0 /data

注：Secondary节点上不允许对DRBD设备进行任何操作，包括挂载；所有的读写操作只能在Primary节点上进行，只有当Primary节点挂掉时，Secondary节点才能提升为Primary节点，并自动挂载DRBD继续工作。
成功挂载后的DRBD状态：（Rserver-1主）

[[email protected]~]# service drbd status

drbddriver loaded OK; device status:

version:8.4.2 (api:1/proto:86-101)

GIT-hash:7ad5f850d711223713d6dcadc3dd48860321070c build by [email protected], 2017-05-1813:40:26

m:res  cs        ro                 ds                 p  mounted fstype

0:r0   Connected Primary/Secondary  UpToDate/UpToDate  C /data    ext4

9.7配置heartbeat服务

yum install heartbeat –y

9.7.1配置ha.cf

cd/usr/share/doc/heartbeat-3.0.4

ll|grep ha.cfauthkeys haresources

8.4.2.1配置ha.cf 文件

debugfile/var/log/ha-debug

logfile/var/log/ha-log

logfacility     local0

####-à以上三行为日志的配置、在你配置时一般不需要改动、

keepalive 2

deadtime 30

warntime 10

initdead 120

###-à以上四行为一些基础的参数、在你配置时一般不需要改动

#       serial serialportname ...

mcast eth2225.0.0.219 694 1 0

##-à此行表示使用多播的方式、需要改动的仅有eth2 改成你的心跳线的网卡

auto_failbackon

node    Rserver-1 ##-à两台存储server的主机名

node    Lserver-1  ##-à两台存储server的主机名

crm     no

9.7.2配置authkeys

auth 3

#1 crc

#2 sha1 HI!

3 md5 Hello!

authkey文件必须为600 权限。Authkey文件中已经说明了需要配置600 权限

#       Authentication file.  Must be mode 600

9.7.3配置haresources

添加一行文件

Rserver-1IPaddr::172.16.1.10/24/eth1 drbddisk::r0 Filesystem::/dev/drbd0::/data::ext4 killnfsd

注：该文件内IPaddr,Filesystem等脚本存放路径在/etc/ha.d/resource.d/下，也可在该目录下存放服务启动脚本（例如：mysql,www），将相同脚本名称添加到/etc/ha.d/haresources内容中，从而跟随heartbeat启动而启动该脚本。
IPaddr::192.168.0.190/24/eth0：用IPaddr脚本配置对外服务的浮动虚拟IP
drbddisk::r0：用drbddisk脚本实现DRBD主从节点资源组的挂载和卸载
Filesystem::/dev/drbd0::/store::ext4：用Filesystem脚本实现磁盘挂载和卸载

Killnfsd这个为控制nfs启动的脚本

9.7.4、编辑脚本文件killnfsd,用来重启NFS服务：(Rserver-1,Lserver-1)

# vi/etc/ha.d/resource.d/killnfsd

killall -9nfsd; /etc/init.d/nfs restart;exit 0

赋予755执行权限：

# chmod 755/etc/ha.d/resource.d/killnfsd

9.7.5、启动HeartBeat服务
在两个节点上启动HeartBeat服务，先启动（Rserver-1）：(Rserver-1,Lserver-1)

# serviceheartbeat start

# chkconfigheartbeat on

现在从其他机器能够ping通虚IP 172.16.1.10，表示配置成功

9.7.6、配置NFS: (Rserver-1,Lserver-1)

编辑exports配置文件，添加以下配置：

# vi/etc/exports

/data        *(rw,no_root_squash)

9.7.7重启NFS服务：

# servicerpcbind restart

# service nfsrestart

# chkconfigrpcbind on

# chkconfig nfsoff

注：这里设置NFS开机不要自动运行，因为/etc/ha.d/resource.d/killnfsd 该脚本会控制NFS的启动。

9.8、测试高可用
9.8.1、正常热备切换
在客户端挂载NFS共享目录

# mount -t nfs 172.16.1.10:/store/tmp

模拟将主节点的heartbeat Rserver-1主节点服务停止

，则备节点Lserver-1备节点会立即无缝接管；

测试客户端挂载的NFS共享读写正常。

此时备机（Lserver-1备）上的DRBD状态：

如果备上面的状态成为primary 就表示已经切换成功。

9.8.2、异常宕机切换

首先把服务和IP全部切换回主上去。后面直接关闭主的电源

[[email protected]]# /etc/init.d/heartbeat start

Starting High-Availabilityservices: INFO:  Resource is stopped

Done.

[[email protected]]# ip addr list

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 state UP qlen1000

link/ether 00:0c:29:20:dc:da brdff:ff:ff:ff:ff:ff

inet 192.168.236.143/24 brd 192.168.236.255scope global eth0

inet 192.168.236.10/24 brd 192.168.236.255scope global secondary eth0

inet6 fe80::20c:29ff:fe20:dcda/64 scopelink

valid_lft forever preferred_lft forever

3: eth1:<BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen1000

link/ether 00:0c:29:20:dc:e4 brdff:ff:ff:ff:ff:ff

inet 172.16.1.1/24 brd 172.16.1.255 scopeglobal eth1

inet 172.16.1.10/24 brd 172.16.1.255 scopeglobal secondary eth1

inet6 fe80::20c:29ff:fe20:dce4/64 scopelink

valid_lft forever preferred_lft forever

4: eth2:<BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen1000

link/ether 00:0c:29:20:dc:ee brdff:ff:ff:ff:ff:ff

inet 192.168.1.1/24 brd 192.168.1.255 scopeglobal eth2

inet6 fe80::20c:29ff:fe20:dcee/64 scopelink

valid_lft forever preferred_lft forever

5: pan0:<BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN

link/ether 6e:5d:75:f7:48:77 brdff:ff:ff:ff:ff:ff

[[email protected]]# df

Filesystem                      1K-blocks    Used Available Use% Mounted on

/dev/mapper/vg_rserver1-lv_root  18650424 4093320  13609700 24% /

tmpfs                              372156      76   372080   1% /dev/shm

/dev/sda1                          495844   34853   435391   8% /boot

/dev/sr0                          4363088 4363088         0 100% /media/CentOS_6.5_Final

/dev/drbd0                         505552   10521   468930   3% /data

[[email protected]]#

已经切换成功。现在测试一下直接宕机看看能不能转换

已经关闭了电源。查看一下备的情况吧。

[[email protected]]# ip addr list

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 00:0c:29:4d:f6:92 brdff:ff:ff:ff:ff:ff

    inet 192.168.236.192/24 brd 192.168.236.255scope global eth0

    inet6 fe80::20c:29ff:fe4d:f692/64 scopelink

       valid_lft forever preferred_lft forever

3:eth1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast stateUP qlen 1000

    link/ether 00:0c:29:4d:f6:9c brdff:ff:ff:ff:ff:ff

    inet 172.16.1.2/24 brd 172.16.1.255 scopeglobal eth1

    inet 172.16.1.10/24 brd 172.16.1.255 scopeglobal secondary eth1

    inet6 fe80::20c:29ff:fe4d:f69c/64 scopelink

       valid_lft forever preferred_lft forever

4:eth2: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast stateUP qlen 1000

    link/ether 00:0c:29:4d:f6:a6 brdff:ff:ff:ff:ff:ff

    inet 192.168.1.2/24 brd 192.168.1.255 scopeglobal eth2

    inet6 fe80::20c:29ff:fe4d:f6a6/64 scopelink

       valid_lft forever preferred_lft forever

5:pan0: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN

    link/ether 92:be:67:20:6e:b6 brdff:ff:ff:ff:ff:ff

[[email protected]]# df

Filesystem                      1K-blocks    Used Available Use% Mounted on

/dev/mapper/vg_lserver1-lv_root  18650424 3966516  13736504 23% /

tmpfs                              372156     224   371932   1% /dev/shm

/dev/sda1                          495844   34856   435388   8% /boot

/dev/sr0                          4363088 4363088         0 100% /media/CentOS_6.5_Final

/dev/drbd0                        505552   10521   468930   3% /data

[[email protected]]#

客户端检查一下

如上图显示heartbeat+DRBD+NFS已经搭建成功。