On Thu, Aug 7, 2008 at 6:07 PM, John Andersen jsamyth@gmail.com wrote:

On Thu, Aug 7, 2008 at 2:52 PM, Andrew Joakimsen joakimsen@gmail.com wrote:

...

Can someone recommend an "easy" disaster recovery solution for an openSUSE system running MD RAID 0? Ideally I could tell a computer-illiterate person to insert a DVD, press "yes" at the prompt and everything will be restored to a default state. --

By MD RAID 0, I assume you mean software raid? (lets be precise here).

You have virtually nothing in the way of disaster recovery with Raid 0, loss of either disk means your data is gone forever.

So I don't see any "Insert the DVD" sort of solution here. Even with a full backup you would have to potentially repair the failed disk then do a restore. And if your system was on the raid-0 (as opposed to just your data) you are right back to doing a bare metal install.

With Raid 1 you get mirroring and you can lose one disk with no data loss. You can recover from that fairly easily (in fact I have lost one disk of a mirror and didn't even notice. I've since started using monitoring).

Disaster recovery is seldom as easy as shoving in a dvd. That's why they use the term "disaster".

If he really meant RAID-0, HP supports "One-Button Disaster Recovery" via some of its hardware / software combos. I think you're looking at a couple $K minimum for hardware/software.

http://h18006.www1.hp.com/products/storageworks/drs/ http://h18006.www1.hp.com/products/storage/software/datapexp/index.html

Traditionally you needed to have your backup on tape. Then you booted their custom restore OS and restored the tape.

Greg

On Fri, Aug 8, 2008 at 12:37 AM, David C. Rankin drankinatty@suddenlinkmail.com wrote:

Short answer: NO

Basic solution:

(1) Backup all data from the current machine - daily; then if it dies (2) Insert the openSuSE install disk; you will have to give more answers than just "yes" (3) Restore the safe, secure, backup of your data

You don't really give enough information about your current backup scheme to know what will be left to restore after a failure.

Where is your data backed up currently?

I don't want to do that. I want to create some sort of bootable medium that can restore the system to the current state. Sorry I meant it is currently in MD RAID 1. There is no current backup. There is no need for constant backups. If a year from now the system fails (yes I understand if a hard drive physically fails that needs to be addressed first) a year-old backup will be just fine.

What you guys are telling me is that thereś nothing like "Norton Ghost" that will work with Linux software RAID? N�����r��y隊Z)z{.�ﮞ˛���m�)z{.��+�Z+i�b�*'jW(�f�vǦj)h���Ǿ��i�������

On Friday 08 August 2008 11:18:08 am Jeffrey L. Taylor wrote:

Quoting Andrew Joakimsen joakimsen@gmail.com:

Just one word: Katrina. RAID-1 will do nothing for a disaster scenario like flood of the building, collapse of the roof, etc. Real backups, preferably stored somewhere off site.

Though OT, that's a valid point. I keep backups both at my mother's house (1 mile), my work (20 miles) and my aunts house (three states away).

My work keeps a live hotsite available with virtualized machines about twenty miles away in a building with a diesel generator and 2 weeks supply of fuel underground.

On Friday 08 August 2008 04:53:31 pm Rajko M. wrote:

On Friday 08 August 2008 05:40:54 pm Kai Ponte wrote:

...

Though OT, that's a valid point. I keep backups both at my mother's house (1 mile), my work (20 miles) and my aunts house (three states away).

Data are safe. Now how to backup us.

I dunno, I'm just a manager.

I don't know how to actually *do* anything.

Andrew Joakimsen wrote:

On Fri, Aug 8, 2008 at 12:37 AM, David C. Rankin drankinatty@suddenlinkmail.com wrote:

...

Short answer: NO

Basic solution:

(1) Backup all data from the current machine - daily; then if it dies (2) Insert the openSuSE install disk; you will have to give more answers than just "yes" (3) Restore the safe, secure, backup of your data

You don't really give enough information about your current backup scheme to know what will be left to restore after a failure.

Where is your data backed up currently?

I don't want to do that. I want to create some sort of bootable medium that can restore the system to the current state. Sorry I meant it is currently in MD RAID 1. There is no current backup. There is no need for constant backups. If a year from now the system fails (yes I understand if a hard drive physically fails that needs to be addressed first) a year-old backup will be just fine.

What you guys are telling me is that thereś nothing like "Norton Ghost" that will work with Linux software RAID?

OK,

I'm beginning to understand. You have a computer that is *already* configured and produced no *data* that significantly changes the setup on a daily basis so you just want something that can restore that system in case of a disaster? Is that close?

Since you are raid1, you have the disks mirrored, so for failure of 1 disk, it is just a matter of replacing a harddrive and letting the software rebuild the new drive. So do you want to address the situation of a double drive failure like caused by fire, etc.?

If so, there are a few ghosting packages, but all installs are somewhat hardware dependent. Do you have a backup machine with the same hardware?

A little more data may produce a little more useful answer.

On Sat, Aug 9, 2008 at 1:56 AM, mourik jan c heupink heupink@merit.unu.edu wrote:

...

...

What you guys are telling me is that thereś nothing like "Norton Ghost" that will work with Linux software RAID?

Take a look at clonezilla [http://www.clonezilla.org/]

mj

Wow. Thanks for that clonzilla clue. That is a sweet package. A little rough around the edges, but quite usable.

I've been trying to clone older specialized machines into Vmware machines and Vmware's own tools for this will not run on old platforms, and requires you to install software on the old-platform. Some of the machines I want to clone only exist as single hard-drives, the actual computer is long gone.

Clonezilla allowed me to clone these fairly easily, temporarily parking the image to my linux server (via ssh), then booting an empty vmware machine with clonezilla and installing the cloned copy there-in.

The only tricky bit was getting the exact same number of total sectors in the virtual disk as the real one or the image fails to boot. (invalid system disc, insert disk and press any key) You can only really be sure of this by using the command line vmware-vdiskmanager to set te exact number of sectors.

Procedure: (Posted here to feed the spiders.....)

1) Use Clonezilla CD to clone the disk to server via ssh from the real hardware.

2) step into the bios and write down total number of actual sectors, cylinders heads and sectors per cylinder.

3) Create a virtual machine. But don't boot it.

4) exit vmware and open a dos box/Shell in the directory where the virtual machine is stored.

5) use vmware-vdiskmanager to manually create a disk in that directory using the actual values for total sectors as the size parameter:

vmware-vdiskmanager -c -a ide -s 123456789sectors -t 2, where 123456789 = the total number of sectors of your real disk and the -t disk type matches your disk type. 2=preallocate.

Optionally you can boot the VM with a windows disk and escape to a shell and type fdisk /mbr at this point.

6) restart vmware and edit it to use this manually created disk as ide 0, deleting the first disk (and the file that contains it).

7) set the VM's CDrom to the Clonezilla disk or the ISO there of.

8) boot the VM machine. It should boot from the cdrom, if not go into the bios of the virtual machine, and adjust the boot device to be the cdrom first. (F2 Immediately).

9) When booting into Clonezilla, set it to restore from the ssh server , and to not install grub (unless you need it), and to use the CHS figures from the backup. Select option to not overwrite the mbr if you put it on above.

The Cloned VMs booted first time.