两硬盘，Linux系统“迁移”及LVM调整

目的:
我的Linux分区不够用了，Linux一直在下电驴。我的另外一个硬盘是windows格式的。我想格掉并把分区分给Linux用。

机器:
台机:无显示器，两IDE硬盘，平时用SSH维护，网络挂时用串口做终端维护。
本本:网络和串口访问台机Linux。


台机开始分区情况:
[root@CentOS5 ~]# fdisk -l
Disk /dev/hda: 120.0 GB, 120034123776 bytes
255 heads, 63 sectors/track, 14593 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot Start End Blocks Id System
/dev/hda1 * 1 651 5229126 b W95 FAT32
/dev/hda2 652 14593 111989115 f W95 Ext'd (LBA)
/dev/hda5 652 13972 107000901 b W95 FAT32
/dev/hda6 13973 14593 4988151 7 HPFS/NTFS

Disk /dev/hdb: 61.4 GB, 61492838400 bytes
255 heads, 63 sectors/track, 7476 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot Start End Blocks Id System
/dev/hdb1 * 1 13 104391 83 Linux
/dev/hdb2 14 7476 59946547+ 8e Linux LVM

[root@CentOS5 ~]# lvm pvs
PV VG Fmt Attr PSize PFree
/dev/hdb2 VolGroup00 lvm2 a- 57.16G 0

[root@CentOS5 ~]# lvm lvs
LV VG Attr LSize Origin Snap% Move Log Copy% Convert
LogVol_opt VolGroup00 -wi-ao 50.16G
LogVol_root VolGroup00 -wi-ao 5.84G
LogVol_swap VolGroup00 -wi-ao 1.16G

CentOS的GRUB装在/hdb1, /hda1上装了个windows 2K. 目前BIOS设置了第一硬盘启动。使用widows 2K的OS loader 来启动grub再启动CentOS。如果改成第二硬盘启动可以直接启动GRUB。

现在我具体想做的步骤:
1.格掉hda成一个/boot分区和一个Linux LVM,比如 hda2成物理LVM Volume Group (vg00).
2.把老Linux系统迁移到新的硬盘。Swap也放到新的硬盘。
A.包括把老的boot分区数据拷贝到新的boot分区
B.在新的vg00中建立新的Logic Volume (lv_root),把老的根分区数据拷进来。
3.当新lv_root分区的系统可以启动的时候,把原来老硬盘空间回收，分配到新的vg中。


具体操作步骤:
正式开始,umount不用的分区。
[root@CentOS5 opt]# fdisk /dev/hda

The number of cylinders for this disk is set to 14593.
There is nothing wrong with that, but this is larger than 1024,
and could in certain setups cause problems with:
1) software that runs at boot time (e.g., old versions of LILO)
2) booting and partitioning software from other OSs
(e.g., DOS FDISK, OS/2 FDISK)

Command (m for help): p

Disk /dev/hda: 120.0 GB, 120034123776 bytes
255 heads, 63 sectors/track, 14593 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot Start End Blocks Id System
/dev/hda1 * 1 651 5229126 b W95 FAT32
/dev/hda2 652 14593 111989115 f W95 Ext'd (LBA)
/dev/hda5 652 13972 107000901 b W95 FAT32
/dev/hda6 13973 14593 4988151 7 HPFS/NTFS

先delete所有原来分区。
Command (m for help): p

Disk /dev/hda: 120.0 GB, 120034123776 bytes
255 heads, 63 sectors/track, 14593 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot Start End Blocks Id System

建立两个分区,一个100M,激活为可启动。剩下的用来建PV。
Disk /dev/hda: 120.0 GB, 120034123776 bytes
255 heads, 63 sectors/track, 14593 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot Start End Blocks Id System
/dev/hda1 * 1 13 104391 83 Linux
/dev/hda2 14 14593 117113850 83 Linux

格式化100M的分区。挂载并拷贝boot内容。
[root@CentOS5 opt]# mkfs.ext3 /dev/hda1
[root@CentOS5 opt]# mkdir /mnt/hda1boot
[root@CentOS5 opt]# mount /dev/hda1 /mnt/hda1boot/
[root@CentOS5 ~]# cp -a /boot/* /mnt/hda1boot/

安装grub到第一硬盘,安装到MBR。
(我在虚拟机中测试时想用grub-install /dev/hda,有出错信息,所以这里用来下面的方法)
grub> root (hd0,0)
Filesystem type is ext2fs, partition type 0x83

grub> setup (hd0)
Checking if "/boot/grub/stage1" exists... no
Checking if "/grub/stage1" exists... yes
Checking if "/grub/stage2" exists... yes
Checking if "/grub/e2fs_stage1_5" exists... yes
Running "embed /grub/e2fs_stage1_5 (hd0)"... 15 sectors are embedded.
succeeded
Running "install /grub/stage1 (hd0) (hd0)1+15 p (hd0,0)/grub/stage2 /grub/grub.conf"... succeeded
Done.

下面准备好LVM。
[root@CentOS5 ~]# pvcreate /dev/hda2
Physical volume "/dev/hda2" successfully created
[root@CentOS5 ~]# pvscan
PV /dev/hdb2 VG VolGroup00 lvm2 [57.16 GB / 0 free]
PV /dev/hda2 lvm2 [111.69 GB]
Total: 2 [168.84 GB] / in use: 1 [57.16 GB] / in no VG: 1 [111.69 GB]
[root@CentOS5 ~]# vgcreate vg00 /dev/hda2
Volume group "vg00" successfully created
[root@CentOS5 ~]# lvcreate -n lv_root -L 6000M vg00
[root@CentOS5 ~]# lvscan
ACTIVE '/dev/VolGroup00/LogVol_root' [5.84 GB] inherit
ACTIVE '/dev/VolGroup00/LogVol_opt' [50.16 GB] inherit
ACTIVE '/dev/VolGroup00/LogVol_swap' [1.16 GB] inherit
ACTIVE '/dev/vg00/lv_root' [5.86 GB] inherit
[root@CentOS5 ~]# mkdir /mnt/lvroot
[root@CentOS5 ~]# mount /dev/vg00/lv_root /mnt/lvroot
开始拷贝/分区下文件和目录:
[root@CentOS5 ~]# cp -a /bin /mnt/lvroot/
[root@CentOS5 ~]# cp -a /etc /mnt/lvroot/
[root@CentOS5 ~]# cp -a /exports /mnt/lvroot/
[root@CentOS5 ~]# cp -a /home /mnt/lvroot/
[root@CentOS5 ~]# cp -a /lib /mnt/lvroot/
[root@CentOS5 ~]# cp -a /net /mnt/lvroot/
[root@CentOS5 ~]# cp -a /root /mnt/lvroot/
[root@CentOS5 ~]# cp -a /sbin /mnt/lvroot/
[root@CentOS5 ~]# cp -a /srv /mnt/lvroot/
[root@CentOS5 ~]# cp -a /tftpboot /mnt/lvroot/
[root@CentOS5 ~]# cp -a /usr /mnt/lvroot/
[root@CentOS5 ~]# cp -a /var /mnt/lvroot/
建立些不拷的目录,同时保证权限一致:
[root@CentOS5 /]# mkdir /mnt/lvroot/boot
[root@CentOS5 /]# mkdir /mnt/lvroot/media
[root@CentOS5 /]# mkdir /mnt/lvroot/misc
[root@CentOS5 /]# mkdir /mnt/lvroot/opt
[root@CentOS5 /]# mkdir /mnt/lvroot/mnt
[root@CentOS5 /]# mkdir /mnt/lvroot/proc
[root@CentOS5 /]# chmod -w /mnt/lvroot/proc/
[root@CentOS5 /]# mkdir /mnt/lvroot/selinux
[root@CentOS5 /]# mkdir /mnt/lvroot/sys
[root@CentOS5 ~]# mkdir /mnt/lvroot/tmp
[root@CentOS5 ~]# chmod 777 /mnt/lvroot/tmp
[root@CentOS5 ~]# chmod +t /mnt/lvroot/tmp

建立swap分区。
[root@CentOS5 /]# lvcreate -n lv_swap -L 1200M vg00
[root@CentOS5 /]# mkswap /dev/vg00/lv_swap

开始修改grub和fstab。
/mnt/hda1boot/grub/menu.lst文件中的root和kernel参数中的root=/dev/vg00/lv_root。

1 /dev/vg00/lv_root / ext3 defaults 1 1
2 #LABEL=/boot /boot ext3 defaults 1 2
3 /dev/hda1 /boot ext3 defaults 1 2
4 tmpfs /dev/shm tmpfs defaults 0 0
5 devpts /dev/pts devpts gid=5,mode=620 0 0
6 sysfs /sys sysfs defaults 0 0
7 proc /proc proc defaults 0 0
8 /dev/vg00/lv_swap swap swap defaults 0 0
9
10 #/dev/VolGroup00/LogVol_opt /opt ext3 defaults 1 2

检查下etc下虽开机启动的脚本。
比如我要把rc.d中oracle相关的去掉。
rc.local中的也去掉。

+++++++++++++++++++一切检查就绪,准备重启+++++++++++++++++++++++++++++++==
接上串口线,我的台机没有显示器。:)
哈哈可以的,看到了新的grub菜单。

然后....
Scanning logical volumes
Reading all physical volumes. This may take a while...
Found volume group "VolGroup00" using metadata type lvm2
Found volume group "vg00" using metadata type lvm2
Activating logical volumes
3 logical volume(s) in volume group "VolGroup00" now active
Trying to resume from /dev/VolGroup00/LogVol_swap
No suspend signature on swap, not resuming.
Creating root device.
Mounting root filesystem.
mount: could notKernel panic - not syncing: Attempted to kill init!
find filesystem '/dev/root'
Se

我奇怪的是为什么我grub的kernel root参数设置了新的分区了啊,为什么系统还是从老的volume group "VolGroup00"启动?

问题解决方法:
参考文章:
http://linux.chinaunix.net/bbs/archiver/tid-1030789.html

开始按他的做了，还是不可以。

觉得奇怪它怎么只去active VolGroup00的呢？
后来仔细看了解压的initrd中的init，原来是那里写了启动哪个Volume Group。 手动修改启动vg00.swap root等也对应修改，打包initrd。

后来看了下initrd，如果系统已经启动了，要生成个新的initrd很简单的。
备份原来的initrd,然后用mkinitrd就可以生成个新的了。
(1)生成个现在使用的kernel的initrd。
/sbin/mkinitrd /boot/initrd-`uname -r`.img `uname -r`
(2)生成xen内核的initrd。
/sbin/mkinitrd /root/initrd-2.6.18-92.1.22.el5xen.img 2.6.18-92.1.22.el5xen
当然不是随便的initrd都可以生成的，你需要有对应的kernel modules。
mkinitrd会查看/etc/fstab文件，这也就是老的initrd只回去挂载老的Volume Group的原因。


总结:
这种方法迁移系统基本没什么损失，没有软件需要重装，我/opt的oracle拷贝回来后就可以启动了。
主要是分区数据基本没变化。LVM很好用。

参考文章:
1.LVM howto:http://www.ibiblio.org/pub/Linux/docs/HOWTO/other-formats/pdf/LVM-HOWTO.pdf
2.initrd Fix 参考文章:http://linux.chinaunix.net/bbs/archiver/tid-1030789.html
3.Linux 初始 RAM 磁盘（initrd）概述:http://www.ibm.com/developerworks/cn/linux/l-initrd.html


后记，后来在一次演示fdisk时错误的把60G的分区格了，但是还没有重启。内核还是使用老的分区表。这个时候可以通过上面的记录来重新fdisk。照理重启应该是OK的。

我使用了另外的方法：

从以前到操作记录找到原来到分区分法:

原来：
Disk /dev/hdb: 61.4 GB, 61492838400 bytes
255 heads, 63 sectors/track, 7476 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot Start End Blocks Id System
/dev/hdb1 * 1 13 104391 83 Linux
/dev/hdb2 14 7476 59946547+ 8e Linux LVM

现在：
Disk /dev/hdb: 61.4 GB, 61492838400 bytes
255 heads, 63 sectors/track, 7476 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot Start End Blocks Id System
/dev/hdb1 1 2433 19543041 83 Linux


这个应该就只是MBR数据被改了吧， 能不能通过dd命令或其他方法来恢复？
或者我再fdisk /dev/hdb 一下，分成和原来一样到，然后不格式化，可行么？

方法1.
改后，因为操作系统已经移动到/dev/hda磁盘，所以没激活成boot分区。
Disk /dev/hdb: 61.4 GB, 61492838400 bytes
255 heads, 63 sectors/track, 7476 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot Start End Blocks Id System
/dev/hdb1 1 13 104391 83 Linux
/dev/hdb2 14 7476 59946547+ 8e Linux LVM

fdisk w 命令后到结果。
Command (m for help): w
The partition table has been altered!

Calling ioctl() to re-read partition table.

WARNING: Re-reading the partition table failed with error 16: Device or resource busy.
The kernel still uses the old table. //这个哪里看的到吗，在/proc还是？
The new table will be used at the next reboot.
Syncing disks.

目前到分区和原来一样了，这样应该也是算重建了分区表吧？ 还不敢重启，期待有经验到确认。

20090716 update：在使用方法2后，我直接重启来测试这个方法的可行性，结果应该OK的， 然后继续LVM操作。
重启后：
Disk /dev/hdb: 61.4 GB, 61492838400 bytes
255 heads, 63 sectors/track, 7476 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot Start End Blocks Id System
/dev/hdb1 1 13 104391 83 Linux
/dev/hdb2 14 7476 59946547+ 8e Linux LVM

LVM显示：
[root@CentOS~]# pvdisplay
--- Physical volume ---
PV Name /dev/hda2
VG Name vg00
PV Size 111.69 GB / not usable 1018.00 KB
Allocatable yes
PE Size (KByte) 4096
Total PE 28592
Free PE 2831
Allocated PE 25761
PV UUID KCAZwV-ycFK-zP5m-A0eS-yuIm-Cotd-P22gjF

--- Physical volume ---
PV Name /dev/hdb2
VG Name vg00
PV Size 57.17 GB / not usable 1.55 MB
Allocatable yes
PE Size (KByte) 4096
Total PE 14635
Free PE 14635
Allocated PE 0
PV UUID 1Ga7wF-a6Gr-yug2-BwVX-gDIp-1epr-IpHndY


方法2，缩小LV，重新LVM分区操作。
参考文章
http://lfs.linuxsir.org/main/?q=node/42

[root@CentOS ~]# mke2fs -n /dev/vg00/lv_opt
mke2fs 1.39 (29-May-2006)
Filesystem label=
OS type: Linux
Block size=4096 (log=2)
Fragment size=4096 (log=2)
19365888 inodes, 38706176 blocks
1935308 blocks (5.00%) reserved for the super user
First data block=0
Maximum filesystem blocks=0
1182 block groups
32768 blocks per group, 32768 fragments per group
16384 inodes per group
Superblock backups stored on blocks:
32768, 98304, 163840, 229376, 294912, 819200, 884736, 1605632, 2654208,
4096000, 7962624, 11239424, 20480000, 23887872

4096
1G=1024M=1024*1024*1024 b

块大小是4096(4kb) 当前有786432个blocks 现在打算减小一个G的空间。
1G占用的blocks一共是262144个，所以减少后的blocks应该是524288个。

60G = 262144*60
减少后的blocks 38706176-262144*60==22977536

umount /home
resize2fs /dev/vg00/lv_opt 22977536
lvreduce -L-60G /dev/vg00/lv_opt
mount /home

[root@CentOS ~]# resize2fs /dev/vg00/lv_opt 22977536
resize2fs 1.39 (29-May-2006)
Please run 'e2fsck -f /dev/vg00/lv_opt' first.

[root@CentOS ~]# e2fsck -f /dev/vg00/lv_opt
e2fsck 1.39 (29-May-2006)
Pass 1: Checking inodes, blocks, and sizes
Pass 2: Checking directory structure
Pass 3: Checking directory connectivity
Pass 4: Checking reference counts
Pass 5: Checking group summary information
/dev/vg00/lv_opt: 32122/19365888 files (6.1% non-contiguous), 19459972/38706176 blocks
[root@CentOS ~]# resize2fs /dev/vg00/lv_opt 22977536
resize2fs 1.39 (29-May-2006)
Resizing the filesystem on /dev/vg00/lv_opt to 22977536 (4k) blocks.

The filesystem on /dev/vg00/lv_opt is now 22977536 blocks long.

[root@CentOS ~]# lvreduce -L-60G /dev/vg00/lv_opt
WARNING: Reducing active logical volume to 87.65 GB
THIS MAY DESTROY YOUR DATA (filesystem etc.)
Do you really want to reduce lv_opt? [y/n]: y
Reducing logical volume lv_opt to 87.65 GB
Logical volume lv_opt successfully resized

[root@CentOS opt]# pvs
PV VG Fmt Attr PSize PFree
/dev/hda2 vg00 lvm2 a- 111.69G 17.00G
/dev/hdb2 vg00 lvm2 a- 57.17G 51.22G
[root@CentOS opt]# pvmove /dev/hdb2
/dev/hdb2: Moved: 4.9%
/dev/hdb2: Moved: 9.5%
/dev/hdb2: Moved: 14.1%
/dev/hdb2: Moved: 18.5%
/dev/hdb2: Moved: 22.9%
/dev/hdb2: Moved: 27.5%
/dev/hdb2: Moved: 31.9%
/dev/hdb2: Moved: 36.4%
/dev/hdb2: Moved: 40.9%
/dev/hdb2: Moved: 45.4%
/dev/hdb2: Moved: 49.8%
/dev/hdb2: Moved: 54.2%
/dev/hdb2: Moved: 58.7%
/dev/hdb2: Moved: 63.1%
/dev/hdb2: Moved: 67.6%
/dev/hdb2: Moved: 72.1%
/dev/hdb2: Moved: 76.6%
/dev/hdb2: Moved: 81.1%
/dev/hdb2: Moved: 85.7%
/dev/hdb2: Moved: 90.1%
/dev/hdb2: Moved: 94.7%
/dev/hdb2: Moved: 99.3%
/dev/hdb2: Moved: 100.0%
[root@CentOS opt]# pvs
PV VG Fmt Attr PSize PFree
/dev/hda2 vg00 lvm2 a- 111.69G 11.06G
/dev/hdb2 vg00 lvm2 a- 57.17G 57.17G

这步后reboot测试方法1，结果OK

[root@CentOS log]# pvdisplay /dev/hdb2
--- Physical volume ---
PV Name /dev/hdb2
VG Name vg00
PV Size 57.17 GB / not usable 1.55 MB
Allocatable yes
PE Size (KByte) 4096
Total PE 14635
Free PE 14635
Allocated PE 0
PV UUID 1Ga7wF-a6Gr-yug2-BwVX-gDIp-1epr-IpHndY
确认PE都没用,Total PE == Free PE
[root@CentOS log]# vgreduce vg00 /dev/hdb2
Removed "/dev/hdb2" from volume group "vg00"
[root@CentOS log]# pvs
PV VG Fmt Attr PSize PFree
/dev/hda2 vg00 lvm2 a- 111.69G 11.06G
/dev/hdb2 lvm2 -- 57.17G 57.17G
[root@CentOS log]# pvremove /dev/hdb2
Labels on physical volume "/dev/hdb2" successfully wiped
[root@CentOS log]# pvs
PV VG Fmt Attr PSize PFree
/dev/hda2 vg00 lvm2 a- 111.69G 11.06G
fdisk把/dev/hdb这块硬盘分成一个区，分区格式LVM
Disk /dev/hdb: 61.4 GB, 61492838400 bytes
255 heads, 63 sectors/track, 7476 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot Start End Blocks Id System
/dev/hdb1 1 7476 60050938+ 8e Linux LVM
这次得小心操作啊

[root@CentOS log]# pvscan
PV /dev/hda2 VG vg00 lvm2 [111.69 GB / 11.06 GB free]
PV /dev/hdb1 lvm2 [57.27 GB]
Total: 2 [168.96 GB] / in use: 1 [111.69 GB] / in no VG: 1 [57.27 GB]
[root@CentOS log]# pvcreate /dev/hdb1
Physical volume "/dev/hdb1" successfully created
[root@CentOS log]# pvscan
PV /dev/hda2 VG vg00 lvm2 [111.69 GB / 11.06 GB free]
PV /dev/hdb1 VG vg00 lvm2 [57.27 GB / 57.27 GB free]
Total: 2 [168.95 GB] / in use: 2 [168.95 GB] / in no VG: 0 [0 ]
[root@CentOS log]# vgs
VG #PV #LV #SN Attr VSize VFree
vg00 2 3 0 wz--n- 168.95G 68.32G
OK,操作结束，空余空间又可以分配给/opt了。
[root@CentOS ~]# lvextend -L +50G /dev/vg00/lv_opt
Extending logical volume lv_opt to 137.65 GB
Logical volume lv_opt successfully resized
[root@CentOS ~]# lvscan
ACTIVE '/dev/vg00/lv_root' [11.80 GB] inherit
ACTIVE '/dev/vg00/lv_swap' [1.17 GB] inherit
ACTIVE '/dev/vg00/lv_opt' [137.65 GB] inherit
但这个时候如果mount空间用df看到还是没加的。

[root@CentOS ~]# resize2fs /dev/vg00/lv_opt
resize2fs 1.39 (29-May-2006)
Please run 'e2fsck -f /dev/vg00/lv_opt' first.

[root@CentOS ~]# e2fsck -f /dev/vg00/lv_opt
e2fsck 1.39 (29-May-2006)
Pass 1: Checking inodes, blocks, and sizes
Pass 2: Checking directory structure
Pass 3: Checking directory connectivity
Pass 4: Checking reference counts
Pass 5: Checking group summary information
/dev/vg00/lv_opt: 32122/11501568 files (6.1% non-contiguous), 19212227/22977536 blocks
[root@CentOS ~]# resize2fs /dev/vg00/lv_opt
resize2fs 1.39 (29-May-2006)
Resizing the filesystem on /dev/vg00/lv_opt to 36084736 (4k) blocks.
The filesystem on /dev/vg00/lv_opt is now 36084736 blocks long.

[root@CentOS ~]# mount /opt/
[root@CentOS ~]# df -m
Filesystem 1M-blocks Used Available Use% Mounted on
/dev/mapper/vg00-lv_root
11898 4936 6359 44% /
/dev/hda1 99 20 75 21% /boot
tmpfs 328 0 328 0% /dev/shm
/dev/mapper/vg00-lv_opt
138743 73642 62282 55% /opt
这次误操作的损失:
因为来不及借移动硬盘，上面的很多数据被我删光。主要是些视频教程，还有已经装好的oracle,Webpshere,xenos.
虽然这些都可以再下过来装上，不过时间浪费了不少。