Server Fault is a question and answer site for system and network administrators. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I understand that RAID should not be used for backups but what about using it to create backups on physical level? I have a setup of regular rsync backups from all the network to one server. Once a week, all the backups are copied to an external HDD (straightforward with dd) and the HDD is carefully carried home, to prevent loss of data in case something happens with the server room.

Now, would it be a good idea to have a RAID1 array of two or more disks and instead of copying the data - just switch the disk with another one and rebuild the array? Will it be a problem to restore the data from the extracted disk afterwards?

share|improve this question
What advantage do you hope to get with this? – Sven Jul 30 '14 at 9:48
I'd be worried about the possibility the RAID might end up reading outdated data from the disk which has been out of the RAID for a while and then came back. – kasperd Jul 30 '14 at 9:53
@SvW Well, I won't need to stop the services to clone the disk, as it is done now, but have everything backed up 'on the go'. – Pavel Potatis Jul 30 '14 at 10:47
What makes you think you won't need to stop the services to clone the disc with this new plan? – MadHatter Jul 30 '14 at 10:51
I don't see why you are cloning the disks instead of just copying files. – Sven Jul 30 '14 at 10:51
up vote 5 down vote accepted

I have worked at a site where this was an official procedure for cloning one very important Solaris machine. The caveats are:

  1. The filesystem will be dirty if this is done while the machine is in use. You will need to quiesce the file systems if the removed RAID disc isn't to be a very blurry snapshot, and under UNIX, this generally means shutting the machine down briefly (which is what we did) or accepting that you'll have to fsck the removed disc as soon as you boot off it, and that this may involve losing some data.

  2. The removed disc needs to be placed into a chassis that is identical in all respects except where differences are completely unavoidable (eg, MAC addresses of NICs). Even then, some cleanup work will be needed to allow the removed HDD to boot, and be resynced to a pair in the new chassis (we had a formal procedure for this).

  3. The production machine needs to be given a clean disc on which to resync its RAID as soon as it's brought back up.

It was some years ago, and I forget exactly why this particular machine was cloned in this way, although at the time I thought the reasons sound. I still feel this isn't a great technique for cloning, although it works for that, and it's a terrible technique for backups - it's very much more expensive than just doing proper backups.

share|improve this answer
Losing data due to running fsck shouldn't happen on any properly designed file system. – kasperd Jul 30 '14 at 12:42
You may be right, but UNIX UFS probably doesn't count as one of those. – MadHatter Jul 30 '14 at 12:58
Reason #1 alone is enough to call it a terrible technique. Thanks! – Pavel Potatis Aug 1 '14 at 9:35

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.