恢复丢失的LVM 设备

1288阅读　0评论2012-07-14　ulovko
分类：

前段时间一个朋友慌张的向我求助，他在重装系统的时候，初始化了存储上的两个LVM设备，导致数据丢失。
“案发现场”：
两台服务器做HA，通过光纤连在一个共享存储上，因为其中一台HA节点出现了问题，需要重装系统，在重装系统的过程中，由于没有拔光纤线，在初始化磁盘的时候，误将存储上的两个分区初始化了，但是系统还是正常的装完了，当时他并没有发现什么问题，因为另一台服务器还在正常运行，被初始化的这两个LVM设备在另一台服务器上还能正常访问，但是两个多小时后，服务器崩溃了，等服务器再起来时，这才发现两个LVM设备已经没了，数据也丢了。
问题分析：
初始化LVM时，只是清除了磁盘头部的lvm header信息，磁盘上的数据没有被清除，所以只需要根据备份的逻辑卷的Metadata信息把lvm header重新写到被初始化的磁盘头部就行,好在当时的服务器系统都有做备份，可以找到lvm Metadata信息。
 
当创建vg的时候，系统默认会自动备份逻辑卷Metadata信息到/etc/lvm/backup，当前逻辑卷metadata信息放在/etc/lvm/archive下面,保险起见，最好定期备份/etc/lvm这个目录下的文件到别的地方。
 
 
下面用虚拟机模拟LVM恢复
创建逻辑卷：

点击(此处)折叠或打开

  1. [root@oracle ~]# pvcreate /dev/sdb1
  2. Physical volume "/dev/sdb1" successfully created
  3. [root@oracle ~]# vgcreate lanv /dev/sdb1
  4. Volume group "lanv" successfully created
  5. [root@oracle ~]# lvcreate -n lgl -L 500M lanv
  6. Logical volume "lgl" created
  7. [root@oracle lgl]# pvs
  8. PV VG Fmt Attr PSize PFree
  9. /dev/sdb1 lanv lvm2 a- 1016.00M 516.00M
  10. [root@oracle lgl]# lvs
  11. LV VG Attr LSize Origin Snap% Move Log Copy% Convert
  12. lgl lanv -wi-ao 500.00M
  13. [root@oracle lgl]# vgs
  14. VG #PV #LV #SN Attr VSize VFree
  15. lanv 1 1 0 wz--n- 1016.00M 516.00M
  16. [root@oracle ~]# df -h /lanv/lgl
  17. Filesystem Size Used Avail Use% Mounted on
  18. /dev/mapper/lanv-lgl 485M 11M 449M 3% /lanv/lgl
  19. [root@oracle ~]# cat /etc/fstab
  20. /dev/lanv/lgl /lanv/lgl ext3 defaults 0 0
  21. [root@oracle lgl]# pwd
  22. /lanv/lgl
  23. [root@oracle lgl]# echo "Hello World" >lgl.txt
  24. [root@oracle lgl]# cat lgl.txt
  25. Hello World
删除lvm信息：

点击(此处)折叠或打开

  1. [root@oracle lgl]# fdisk /dev/sdb
  3. Command (m for help): p
  5. Disk /dev/sdb: 1073 MB, 1073741824 bytes
  6. 255 heads, 63 sectors/track, 130 cylinders
  7. Units = cylinders of 16065 * 512 = 8225280 bytes
  9. Device Boot Start End Blocks Id System
  10. /dev/sdb1 1 130 1044193+ 83 Linux
  12. Command (m for help): d
  13. Selected partition 1
  15. Command (m for help): p
  17. Disk /dev/sdb: 1073 MB, 1073741824 bytes
  18. 255 heads, 63 sectors/track, 130 cylinders
  19. Units = cylinders of 16065 * 512 = 8225280 bytes
  21. Device Boot Start End Blocks Id System
  23. Command (m for help): w
  24. The partition table has been altered!
  26. Calling ioctl() to re-read partition table.
  28. WARNING: Re-reading the partition table failed with error 16: Device or resource busy.
  29. The kernel still uses the old table.
  30. The new table will be used at the next reboot.
  31. Syncing disks.

这个时候lvm设备仍然能访问，能正常读写，过一段时间后会报错，系统崩溃。

重启系统报错：




点击(此处)折叠或打开

  1. [root@oracle backup]# fdisk /dev/sdb
  3. Command (m for help): p
  5. Disk /dev/sdb: 1073 MB, 1073741824 bytes
  6. 255 heads, 63 sectors/track, 130 cylinders
  7. Units = cylinders of 16065 * 512 = 8225280 bytes
  9. Device Boot Start End Blocks Id System
  11. Command (m for help): n
  12. Command action
  13. e extended
  14. p primary partition (1-4)
  15. p
  16. Partition number (1-4): 1
  17. First cylinder (1-130, default 1):
  18. Using default value 1
  19. Last cylinder or +size or +sizeM or +sizeK (1-130, default 130):
  20. Using default value 130
  22. Command (m for help): w
  23. The partition table has been altered!
  25. Calling ioctl() to re-read partition table.
  26. Syncing disks.
  28. [root@oracle backup]# fdisk -l
  30. Disk /dev/sda: 32.2 GB, 32212254720 bytes
  31. 255 heads, 63 sectors/track, 3916 cylinders
  32. Units = cylinders of 16065 * 512 = 8225280 bytes
  34. Device Boot Start End Blocks Id System
  35. /dev/sda1 * 1 3655 29358756 83 Linux
  36. /dev/sda2 3656 3916 2096482+ 82 Linux swap / Solaris
  38. Disk /dev/sdb: 1073 MB, 1073741824 bytes
  39. 255 heads, 63 sectors/track, 130 cylinders
  40. Units = cylinders of 16065 * 512 = 8225280 bytes
  42. Device Boot Start End Blocks Id System
  43. /dev/sdb1 1 130 1044193+ 83 Linux

现在可以用vgcfgrestore 命令根据备份的Metadata信息恢复lvm

先查看下备份的Metadata信息


 

点击(此处)折叠或打开

  1. [root@oracle backup]# more lanv
  2. # Generated by LVM2 version 2.02.46-RHEL5 (2009-06-18): Sun Jul 8 19:44:21 2012
  4. contents = "Text Format Volume Group"
  5. version = 1
  7. description = "Created *after* executing 'vgs'"
  9. creation_host = "oracle" # Linux oracle 2.6.18-164.el5 #1 SMP Tue Aug 18 15:51:48 EDT 2009 x86_64
  10. creation_time = 1341747861 # Sun Jul 8 19:44:21 2012
  12. lanv {
  13. id = "hlt8bK-xsoA-Z3GJ-iosl-7cwk-e5b1-TUx3OB"
  14. seqno = 4
  15. status = ["RESIZEABLE", "READ", "WRITE"]
  16. flags = []
  17. extent_size = 8192 # 4 Megabytes
  18. max_lv = 0
  19. max_pv = 0
  21. physical_volumes {
  23. pv0 {
  24. id = "qm6Oo3-92NX-E0ca-e0GV-2eA6-q2Jm-0CzGre"
  25. device = "/dev/sdb" # Hint only
  27. status = ["ALLOCATABLE"]
  28. flags = []
  29. dev_size = 2088387 # 1019.72 Megabytes
  30. pe_start = 384
  31. pe_count = 254 # 1016 Megabytes
  32. }
  33. }
  35. logical_volumes {
  37. lgl {
  38. id = "ifkQpI-bf7f-6SvS-8V9J-HVmO-wAZj-0jWY4B"
  39. status = ["READ", "WRITE", "VISIBLE"]
  40. flags = []
  41. segment_count = 1
  43. segment1 {
  44. start_extent = 0
  45. extent_count = 125 # 500 Megabytes
  47. type = "striped"
  48. stripe_count = 1 # linear
  50. stripes = [
  51. "pv0", 0
  52. ]
  53. }
  54. }
  55. }
  56. }
vgcfgrestore 命令默认从/etc/lvm/backup查找Metadata信息来恢复lvm

点击(此处)折叠或打开

  1. [root@oracle ~]# vgcfgrestore lanv
  2. Couldn't find device with uuid 'qm6Oo3-92NX-E0ca-e0GV-2eA6-q2Jm-0CzGre'.
  3. Cannot restore Volume Group lanv with 1 PVs marked as missing.
  4. Restore failed.
根据错误提示，重新创建pv

点击(此处)折叠或打开

  1. [root@oracle backup]# pvcreate --uuid qm6Oo3-92NX-E0ca-e0GV-2eA6-q2Jm-0CzGre /dev/sdb1
  2. Can't initialize physical volume "/dev/sdb1" of volume group "lanv" without -ff
  3. [root@oracle backup]# pvcreate --uuid qm6Oo3-92NX-E0ca-e0GV-2eA6-q2Jm-0CzGre /dev/sdb1 -ff
  4. Really INITIALIZE physical volume "/dev/sdb1" of volume group "lanv" [y/n]? y
  5. WARNING: Forcing physical volume creation on /dev/sdb1 of volume group "lanv"
  6. Physical volume "/dev/sdb1" successfully created

点击(此处)折叠或打开

  1. [root@oracle backup]# pvs
  2. PV VG Fmt Attr PSize PFree
  3. /dev/sdb1 lvm2 -- 1019.72M 1019.72M
现在再用vgcfgrestore 命令恢复

点击(此处)折叠或打开

  1. [root@oracle backup]# vgcfgrestore lanv
  2. Restored volume group lanv
  3. [root@oracle backup]# vgs
  4. VG #PV #LV #SN Attr VSize VFree
  5. lanv 1 1 0 wz--n- 1016.00M 516.00M
  6. [root@oracle backup]# lvs
  7. LV VG Attr LSize Origin Snap% Move Log Copy% Convert
  8. lgl lanv -wi--- 500.00M
  9. [root@oracle backup]# vgchange -ay lanv
  10. 1 logical volume(s) in volume group "lanv" now active
查看恢复结果

点击(此处)折叠或打开

  1. [root@oracle backup]# lvs
  2. LV VG Attr LSize Origin Snap% Move Log Copy% Convert
  3. lgl lanv -wi-a- 500.00M
  4. [root@oracle backup]# mount /dev/lanv/lgl /lanv/lgl/
  5. [root@oracle backup]# df -h /lanv/lgl/
  6. Filesystem Size Used Avail Use% Mounted on
  7. /dev/mapper/lanv-lgl 485M 11M 449M 3% /lanv/lgl
  8. [root@oracle backup]# cd /lanv/lgl/
  9. [root@oracle lgl]# ls
  10. lgl.txt
  11. [root@oracle lgl]# cat lgl.txt
  12. Hello World
到此lvm已经恢复，里面的数据完好无损

恢复步奏其实不难，就两条命令，但是在生产环境应该尽量避免发生这种情况，比如在装系统时，应该先把外部存储设备都拔了，免得不小心删除了数据。





 
上一篇：linux配置网络rescue mode
下一篇：工作效率與洗澡的順序：為什麼要先洗身體→洗頭→洗臉！