Hack and / - When Disaster Strikes: Restoring a Master Boot Record
I had at least used Linux long enough that after I made my mistake, I realized my actual data was still there and that there must be some way to restore the partition table. This was when I first came across the wonderful tool called gpart.
gpart is short for Guess Partition, and that is exactly what it does. When you run the gpart command, it scans through a disk looking for signs of partitions. If it finds what appears to be the beginning of a Windows FAT32 partition, for instance, it jots it down and continues until eventually it sees what appears to be the end. Once the tool has scanned the entire drive, it outputs its results to the screen for you to check and edit. It also optionally can write this reconstructed partition table back to the disk.
gpart has been around for quite some time and is packaged by all of the major distributions, so you should be able to install it with your standard package manager. Don't confuse it with gparted, which is a graphical partitioning tool. Of course, if your main system is the one with the problem, you need to find a rescue disk that has it. Knoppix and a number of other rescue-focused disks all include gpart out of the box.
To use gpart, run it with root privileges and give it the disk device to scan as an argument. Here's gpart's output from a scan of my laptop's drive:
greenfly@minimus:~$ sudo gpart /dev/sda Begin scan... Possible partition(Linux ext2), size(9773mb), offset(0mb) Possible partition(Linux swap), size(980mb), offset(9773mb) Possible partition(SGI XFS filesystem), size(20463mb), offset(10754mb) End scan. Checking partitions... Partition(Linux ext2 filesystem): primary Partition(Linux swap or Solaris/x86): primary Partition(Linux ext2 filesystem): primary Ok. Guessed primary partition table: Primary partition(1) type: 131(0x83)(Linux ext2 filesystem) size: 9773mb #s(20016920) s(63-20016982) chs: (0/1/1)-(1023/254/63)d (0/1/1)-(1245/254/56)r Primary partition(2) type: 130(0x82)(Linux swap or Solaris/x86) size: 980mb #s(2008120) s(20016990-22025109) chs: (1023/254/63)-(1023/254/63)d (1246/0/1)-(1370/254/58)r Primary partition(3) type: 131(0x83)(Linux ext2 filesystem) size: 20463mb #s(41909120) s(22025115-63934234) chs: (1023/254/63)-(1023/254/63)d (1371/0/1)-(3979/184/8)r Primary partition(4) type: 000(0x00)(unused) size: 0mb #s(0) s(0-0) chs: (0/0/0)-(0/0/0)d (0/0/0)-(0/0/0)r
To hammer home the point about how easy it is to back up the MBR, now I have an extra backup of my laptop partition table—in this magazine.
As you can see, it correctly identified the two primary partitions (/ and /home) and the swap partition on my laptop and noted that the fourth primary partition was unused. Now, after reviewing this, if I decided that I wanted gpart to write its data to the drive, I would run:
$ sudo gpart -W /dev/sda /dev/sda
That isn't a typo; the -W argument tells gpart to which disk to write the partition table, but you still need to tell it which drive to scan. gpart potentially could scan one drive and write the partition table to another. Once you specify the -W option, gpart gives you some warnings to accept, but it also prompts you to edit the results from within gpart itself. Personally, I've always found it a bit more difficult to do it that way than it needs to be, so I skip the editor, have it write to the disk, and then use a tool like fdisk or cfdisk to examine the drive afterward and make tweaks if necessary.
gpart is a great tool and has saved me a number of times, but it does have some limitations. For one, although gpart works very well with primary partitions, it is much more difficult for it to locate extended partitions, depending on which tool actually created them. Second, take gpart results with a grain of salt. It does its best to reconstruct drives, but you always should give its results a sanity check. For instance, I've seen where it has identified the end of a partition one or two megabytes short from the actual end. Typically, when we partition drives, we put one partition immediately after another, so these sorts of errors are pretty easy to find.
Now, if you have destroyed only the partition table, you hopefully should be restored at this point. If you managed to destroy the boot code as well, you need to restore it too. These days, most Linux distributions use GRUB, so with your restored partition table, if you are currently booted into the affected system, run:
$ sudo grub-install --recheck /dev/sda
Replace /dev/sda with the path to your primary boot device. If you use an Ubuntu system, you optionally could use the update-grub tool instead. If you are currently booted in to a rescue disk, you first need to mount your root partition at, say, /mnt/sda1, and then use chroot to run grub-install within it:
$ sudo mkdir /mnt/sda1 $ sudo mount /dev/sda1 /mnt/sda1 $ sudo chroot /mnt/sda1 /usr/sbin/grub-install ↪--recheck /dev/sda
If the chrooted grub-install doesn't work, you typically can use your rescue disk's grub-install with the --root-directory option:
$ sudo /usr/sbin/grub-install --recheck ↪--root-directory /mnt/sda1 /dev/sda
Well hopefully, if you didn't have a profound respect for those 512 bytes at the beginning of your hard drive, you do now. The MBR is like many things in life that you don't miss until they are gone, but at least in this case, when it's gone, you might be able to bring it back.
Kyle Rankin is a Senior Systems Administrator in the San Francisco Bay Area and the author of a number of books, including Knoppix Hacks and Ubuntu Hacks for O'Reilly Media. He is currently the president of the North Bay Linux Users' Group.
Kyle Rankin is a director of engineering operations in the San Francisco Bay Area, the author of a number of books including DevOps Troubleshooting and The Official Ubuntu Server Book, and is a columnist for Linux Journal.
Webinar: 8 Signs You’re Beyond Cron
11am CDT, April 29th
Join Linux Journal and Pat Cameron, Director of Automation Technology at HelpSystems, as they discuss the eight primary advantages of moving beyond cron job scheduling. In this webinar, you’ll learn about integrating cron with an enterprise scheduler.Join us!
|Play for Me, Jarvis||Apr 16, 2015|
|Drupageddon: SQL Injection, Database Abstraction and Hundreds of Thousands of Web Sites||Apr 15, 2015|
|Non-Linux FOSS: .NET?||Apr 13, 2015|
|Designing Foils with XFLR5||Apr 08, 2015|
|diff -u: What's New in Kernel Development||Apr 07, 2015|
- Drupageddon: SQL Injection, Database Abstraction and Hundreds of Thousands of Web Sites
- Play for Me, Jarvis
- Non-Linux FOSS: .NET?
- Designing Foils with XFLR5
- Not So Dynamic Updates
- Flexible Access Control with Squid Proxy
- New Products
- Users, Permissions and Multitenant Sites
- diff -u: What's New in Kernel Development