Dear All
In a situation where a boot partition fails, and the partition cannot be recreated ie because the drive is faulty, here is a quick recipe to possibly recover your data.
Say your faulty sata drive is connected to sda, then connect another clean drive to sdb.
Boot your server using the sme current CD, and start up in Rescue mode. Mount your system as prompted.
Format the new drive connected to sdb as ext3.
Then mount that drive
mount /dev/sdb /mnt/newdrive
Initially test your drive or drives using smarctl, see the Howto. You can also seperately download the Ultimate Boot CD (UBCD) and use that to run various tests on your system including drive diagnostic testing for a variety of manufacturers drives.
Back to the mounted system. Ascertain with smartctl short and long tests the status of your drive ie it's really faulty or just spat out of an array or lost the partition for some unknown reason.
Assuming you can mount the faulty drive as well as your new drive, you can then copy data from the faulty mounted drive to the newdrive using standard tools eg cp command, mc and tar (to preserve ownership & permissions).
Copy files from their source locations to /mnt/newdrive
When finished umount the newdrive
umount /mnt/sdb
Shutdown
Build a new server from scratch and configure it as required.
Then attach the newdrive with the retrieved data on it, say to sdb (on a single drive machine), or to sdc (on a RAID1 machine).
Mount the drive
mount /dev/sdc /mnt/retrieved
Copy data to working system.
Make sure you set ownership and permissions correctly if using the cp command. Use chown & chmod to do this.
Note it's better to use the mc (midnight commander) move function as original permissions & ownership are retained, or alternatively tar and untar folders & files.
When finished unmount the newdrive
umount /mnt/retrieved
Shutdown and disconnect newdrive
Reboot server and check all is OK.
Refer to the various Howtos for info about RAID, testing drives and formatting drives (see USB howto). Google is also your friend if unsure of any commands or usage syntax. I will add more detailed steps later when time permits.
Having retrieved your data now say a BIG THANKYOU to the powers that be of your own personal religious or non religious choosing. Rush out and buy a big USB drive, attach it to your server and configure the built in backup to run daily incremental backups, and NEVER make the mistake of having inadequate backups again.
Oh and the other Golden Rule, before you touch your server to replace faulty drives (eg in a RAID array) or do any work on it, or for that matter even breathe on it, manually instigate a full backup (by changing backup schedule settings or from the admin console), and wait until that backup has finished before doing maintenance work. That way you have a full backup of the machine as at the time you shut it down, rather than 1 day ago. A lot of data can change in one day.
16 Replies
4246 Views