A backup or raid system with ntfs partition without losing my data?

[jaspers]

Hey guys we have a newb to linux here

Right, here's my senario, I "had" a 3tb external hdd formatted to ntfs which I have lots of my dvds ripped on and was connected to a media center PC. I also have a 2tb WD MyBook World network drive with lots of important files on it that I have collected for years (ie. photos, family videos etc...) which has started having problems. I also have a 1tb WD MyBook World Edition which has failed.

I decided to build my own overkill NAS, I purchased a nice case (Hoojum Cubit) a nice mainboard and another 3tb drive exactly the same as my other.

I took the drives out of their external cases and put them in my new NAS box and I run my ubuntu 11.04 from usb stick. I have already setup samba and mediatomb, installed webmin and got it sharing the files/folders I want. but the hdd with all my files on is still ntfs.

So my question is what's the best way to mirror my hdd with the one I just purchased?

[zombykillah]

Hi there,

As you already see Linux can use NTFS.

Usually when we talk about mirror a hard drive it happens before the file system.
If you want more information about software-raid / Raid 1 is mirror you may look into "mdadm".
What you probably wanna do is to create a mirror with two of your new drives and copy all the data from your old one into it.

So first you should make sure what hdd has the information on it.
I guess it's sda
The first partition will be sda1

Now I would suggest you copy all data 1:1 onto the second drive, just to make sure.
Which may be sdb

With the command:

dd if=/dev/sda of=/dev/sdb

All data from DRIVE A will be written on DRIVE B!

Now you should be able to read your data from bout drives!

With:

fdisk /dev/sda

You may alter your partitions into a raid partition.
Your hdd are bigger than 2TB?! ... damn fdisk won't work ... anyone know the new tools??
Guess it was gdisk or something like this...

Well anyway that done you can create your RAID 1 Device:

mdadm --create /dev/md0 --level=0 --raid-devices=2 /dev/sda1 /dev/sdb1

The ntfs data can now be found on device /dev/md0.


If you have more than 2 drives you may wanna create a RAID5 instead of an RAID1

[Irithori]

parted or gparted can deal with gpt drives.

Imho: using ntfs as a data basis for a linux daemon is discouraged.
It comes with a performance penalty and ntfs is not a posix filesystem.

I would suggest to
- get yet another new 3tb disk
- build a raid1 with the two new empty disks
- format the raid with ext4
- and rsync the files over.

Reasons:
a) the original files are not in danger, becasue they are only read
b) the resulting fs is a linux native one, which is way better in terms of performance and posix conformity under linux than a ntfs
c) after the copy process, you gain a spare disk if one of the raid disks should fail

[Irithori]

Additional hint:
I always create the raid partitons a little bit *smaller* than the maximum size of the disk.

Why?
Because, if a raid disk fails and you buy a replacement from another brand or even only model, then there is a chance, that this replacement disk is just a few sectors too small.
Hence it then cannot be used in the exising raid.

[HROAdmin26]

I would echo several of the same things already mentioned:

- Don't use NTFS in a Linux system (it's not native, it fragments, it runs via FUSE)
- The Linux md (RAID) driver is sufficient for RAID needs...

BUT, RAID is *not* a backup method. You can Google for "raid is not a backup" if you'd like to read more about why. (I mention this because of the thread topic.)

So, if you have (2) 3TB drives, maybe you'd like to get a third and:
- RAID1 (2) of the drives
- Use the 3RD drive as a location to rsync, tar-and-copy, etc. your files to a backup location.

If you are not concerned with "high availability" (which is what RAID gets you), then perhaps you just need 2 drives where data from one is copied (backed up) to the other on a periodic basis. My point: Do not plan/believe that a RAID set is a data backup.

[jaspers]

Hey, thanks for the replies,

All I really want is the easiest way to keep data from one drive equal to another. Is there any solution which could just check for modifications to the data and fs on the main drive and clone it on to the other (kind of like a 1 click backup or something)?

The drive is ntfs because I didn't change the fs due to there been important files which I don't want to lose, I could format the new drive to ext* and copy the data over before creating the mirror method (which takes ages)/

[HROAdmin26]

All I really want is the easiest way to keep data from one drive equal to another.

Rsync is designed for this - schedule it via cron to run as often as you want.

I could format the new drive to ext* and copy the data over

I would still recommend moving to a native Linux filesystem. It will pay off in terms of longevity and performance.

[jaspers]

Ok thanks for the advice guys,

I've taken a quick read about Rsync and it sounds ideal for my needs. I'm also taking the advice of many and gonna use native Linux filesystem partitions for my data.

Thanks again.