详解建立Linux环境软RAID 5教程

1:Raid定义 
RAID,全称Redundant Array of Inexpensive Disks,中文名为廉价磁盘冗余阵列.RAID可分为软RAID和硬RAID,软RAID是通过软件实现多块硬盘冗余的.而硬RAID是一般通过RAID卡来实现RAID的.前者配置简单,管理也比较灵活.对于中小企业来说不失为一最佳选择.硬RAID往往花费比较贵.不过,在性能方面具有一定优势. 
2:RAID分类 
RAID可分为以下几种,做个表格认识下: 
RAID 0 存取速度最快 没有容错 
RAID 1 完全容错 成本高,硬盘使用率低. 
RAID 3 写入性能最好 没有多任务功能 
RAID 4 具备多任务及容错功能 Parity 磁盘驱动器造成性能瓶颈 
RAID 5 具备多任务及容错功能 写入时有overhead 
RAID 0+1 速度快、完全容错 成本高

3:Linux RAID 5实验详解 
假设我有4块硬盘,(没有条件的朋友可以用虚拟机设置出4块硬盘出来).分别为/dev/sda /dev/sdb /dev/sdc /dev/sdd.首先做的就是分区了. 
[root@localhost /]# fdisk /dev/sda 
Device contains neither a valid DOS partition table, nor Sun, SGI or OSF disklabel 
Building a new DOS disklabel. Changes will remain in memory only, 
until you decide to write them. After that, of course, the previous 
content won’t be recoverable. 
Warning: invalid flag 0×0000 of partition table 4 will be corrected by w(rite) 
Command (m for help): n #按n创建新分区 
Command action 
e extended 
p primary partition (1-4) #输入p 选择创建主分区 
p 
Partition number (1-4): 1 #输入 1 创建第一个主分区 
First cylinder (1-130, default 1): #直接回车,选择分区开始柱面这里就从 1 开始 
Using default value 1 
Last cylinder or +size or +sizeM or +sizeK (1-102, default 130): 
Using default value 130 
Command (m for help): w #然后输入w写盘 
The partition table has been altered! 
Calling ioctl() to re-read partition table. 
Syncing disks. 
其它分区照这样做全部分出一个区出来.下面是总分区信息: 
[root@localhost /]# fdisk -l 
Disk /dev/sda: 1073 MB, 1073741824 bytes 
255 heads, 63 sectors/track, 130 cylinders 
Units = cylinders of 16065 * 512 = 8225280 bytes 
Device Boot Start End Blocks Id System 
/dev/sda1 1 130 1044193+ 83 Linux 
Disk /dev/sdb: 1073 MB, 1073741824 bytes 
255 heads, 63 sectors/track, 130 cylinders 
Units = cylinders of 16065 * 512 = 8225280 bytes 
Device Boot Start End Blocks Id System 
/dev/sdb1 1 130 1044193+ 83 Linux 
Disk /dev/sdc: 1073 MB, 1073741824 bytes 
255 heads, 63 sectors/track, 130 cylinders 
Units = cylinders of 16065 * 512 = 8225280 bytes 
Device Boot Start End Blocks Id System 
/dev/sdc1 1 130 1044193+ 83 Linux 
Disk /dev/sdd: 1073 MB, 1073741824 bytes 
255 heads, 63 sectors/track, 130 cylinders 
Units = cylinders of 16065 * 512 = 8225280 bytes 
Device Boot Start End Blocks Id System 
/dev/sdd1 1 130 1044193+ 83 Linux

下一步就是创建RAID了. 
[root@localhost ~]# mdadm –create /dev/md0 –level=5 –raid-devices=3 –spare-devices=1 /dev/sd[a-d]1 #意思是创建RAID设备名为md0, 级别为RAID 5 
mdadm: array /dev/md0 started. 使用3个设备建立RAID,空余一个做备用. 
OK,初步建立了RAID了,我们看下具体情况吧. 
[root@localhost ~]# mdadm –detail /dev/md0 
/dev/md0: 
Version : 00.90.01 
Creation Time : Fri Aug 3 13:53:34 2007 
Raid Level : raid5 
Array Size : 2088192 (2039.25 MiB 2138.31 MB) 
Device Size : 1044096 (1019.63 MiB 1069.15 MB) 
Raid Devices : 3 
Total Devices : 4 
Preferred Minor : 0 
Persistence : Superblock is persistent 
Update Time : Fri Aug 3 13:54:02 2007 
State : clean 
Active Devices : 3 
Working Devices : 4 
Failed Devices : 0 
Spare Devices : 1 
Layout : left-symmetric 
Chunk Size : 64K 
Number Major Minor RaidDevice State 
0 8 1 0 active sync /dev/sda1 
1 8 17 1 active sync /dev/sdb1 
2 8 33 2 active sync /dev/sdc1 
3 8 49 -1 spare /dev/sdd1 
UUID : e62a8ca6:2033f8a1:f333e527:78b0278a 
Events : 0.2 
让RAID开机启动.配置RIAD配置文件吧.默认名字为mdadm.conf,这个文件默认是不存在的,要自己建立.该配置文件存在的主要作用是系统启动的时候能够自动加载软RAID，同时也方便日后管理. 
说明下,mdadm.conf文件主要由以下部分组成:DEVICES选项制定组成RAID所有设备, ARRAY选项指定阵列的设备名、RAID级别、阵列中活动设备的数目以及设备的UUID号. 
[root@localhost ~]# mdadm –detail –scan > /etc/mdadm.conf 
[root@localhost ~]# cat /etc/mdadm.conf 
ARRAY /dev/md0 level=raid5 num-devices=3 UUID=e62a8ca6:2033f8a1:f333e527:78b0278a 
devices=/dev/sda1,/dev/sdb1,/dev/sdc1,/dev/sdd1 
#默认格式是不正确的,需要做以下方式的修改: 
[root@localhost ~]# vi /etc/mdadm.conf 
[root@localhost ~]# cat /etc/mdadm.conf 
devices /dev/sda1,/dev/sdb1,/dev/sdc1,/dev/sdd1 
ARRAY /dev/md0 level=raid5 num-devices=3 UUID=e62a8ca6:2033f8a1:f333e527:78b0278a

将/dev/md0创建文件系统, 
[root@localhost ~]# mkfs.ext3 /dev/md0 
mke2fs 1.35 (28-Feb-2004) 
Filesystem label= 
OS type: Linux 
Block size=4096 (log=2) 
Fragment size=4096 (log=2) 
261120 inodes, 522048 blocks 
26102 blocks (5.00%) reserved for the super user 
First data block=0 
Maximum filesystem blocks=536870912 
16 block groups 
32768 blocks per group, 32768 fragments per group 
16320 inodes per group 
Superblock backups stored on blocks: 
32768, 98304, 163840, 229376, 294912 
Writing inode tables: done 
Creating journal (8192 blocks): done 
Writing superblocks and filesystem accounting information: done 
This filesystem will be automatically checked every 21 mounts or 
180 days, whichever comes first. Use tune2fs -c or -i to override.内容 
挂载/dev/md0到系统中去,我们实验是否可用: 
[root@localhost ~]# cd / 
[root@localhost /]# mkdir mdadm 
[root@localhost /]# mount /dev/md0 /mdadm/ 
[root@localhost /]# cd /mdadm/ 
[root@localhost mdadm]# ls 
lost+found 
[root@localhost mdadm]# cp /etc/services . 
[root@localhost mdadm]# ls 
lost+found services 
好了,如果其中某个硬盘坏了会怎么样呢?系统会自动停止这块硬盘的工作,然后让后备的那块硬盘顶上去工作.我们可以实验下. 
[root@localhost mdadm]# mdadm /dev/md0 –fail /dev/sdc1 
mdadm: set /dev/sdc1 faulty in /dev/md0 
[root@localhost mdadm]# cat /proc/mdstat 
Personalities : [raid5] 
md0 : active raid5 sdc1[3](F) sdd1[2] sdb1[1] sda1[0] # F标签以为此盘为fail. 
2088192 blocks level 5, 64k chunk, algorithm 2 [3/3] [UUU] 
unused devices: 
如果我要移除一块坏的硬盘或添加一块硬盘呢? 
#删除一块硬盘 
[root@localhost mdadm]# mdadm /dev/md0 –remove /dev/sdc1 
mdadm: hot removed /dev/sdc1 
[root@localhost mdadm]# cat /proc/mdstat 
Personalities : [raid5] 
md0 : active raid5 sdd1[2] sdb1[1] sda1[0] 
2088192 blocks level 5, 64k chunk, algorithm 2 [3/3] [UUU] 
unused devices: 
#增加一块硬盘 
[root@localhost mdadm]# mdadm /dev/md0 –add /dev/sdc1 
mdadm: hot added /dev/sdc1 
[root@localhost mdadm]# cat /proc/mdstat 
Personalities : [raid5] 
md0 : active raid5 sdc1[3] sdd1[2] sdb1[1] sda1[0] 
2088192 blocks level 5, 64k chunk, algorithm 2 [3/3] [UUU] 
unused devices: