Bare Metal Backup

I have a Dell 2900 with dual 64bit quad cores and a intergrated factory raid controler.

It currently has SUSE 10.1 running in the main Raid set. I am new to SUSE and just learning UNIX.

A friend of mine told me about the possible problems with Ghost and how partition information is reassembled during recovery as well as the file system my SUSE uses. I am using ReiserFS. “I heard he is in jail for killing his pregnant wife and that support may be a future problem”.

I need a way to bare metal this machine in the event it crashes.

The main role of this machine is a VMserver. Dell shipped the machine with SUSE and I have started learning it to save the costs associated with a Microsoft host OS… Just trying to purchase a Microsoft host OS that supports 16G of ram and dual quad core 64bit processors hurts! Especily when it is just for this role!

I am at the satge where I am faced with so much information to absorb it is all scattered on the floor. I have started moving some of the pieces around and looked at the edges but am in no way close to snapping the puzzle together.

Any advice you could provide would be appreatiated.


When you say bare metal do you intend to cold swap to an identical machine?

Could you pull out the RAID drives in an emergency and slot them into an identical spare? You would have to prepare one or two files where the device ID matters, like the udev rule for activating NICs (you can put rules from both machines with both MAC addresses in there)

Or do you intend to do a full restore from a image?

If you have an identical spare, why not use it in load balancing? That would reduce the time pressure.

How time critical is recovery for you anyway?

I would probably not go with reiserfs if only because it is going nowhere and at some point distros will not fix any more bugs discovered in it, if not already.

BTW, the SUSE 10.1 you got from Dell probably means SLES 10.1, not openSUSE, so don’t get confused there. openSUSE 10.1 has been out of support for a while now (1 year?).

The short answer, is that it’s fairly simple to restore the OS booting either with Live CD/DVD, USB or network. Create partition table on new disks, build filesystems, restore files, then reinstalling the GRUB boot loader.

You may be looking for a commercial backup solution and features, to automate parts of this. Fraid I’m one of those CLI and script hacking ppl, so there’s no way I can recommend anything in that direction.

I’m presumsuming a VMserver means you have many virtual machines. In that case, it is important to get some form of snapshots, which can recover and restart the running live system. It may be convenient for you to copy these files onto a disk which can be saved by your main stay backup system.

There’s no need to worry about the fate of Han’s Reiser, his company had other programmers, the source code still runs, and is publically available. When you next upgrade, you probably should move to ext3 or ext4 depending what’s recommended at the time.

So do you have a preferred way of saving the data, to restore it, in some network accessible way? Will there be other kit to support backing up & restoring data?

Do you want a commercial product or trying to learn how to use standard tools?

Most likely nothing bad will happen if you choose reiserfs, but it’s a bit optimistic to believe that just because the source is open, that there will be always be people around to fix it. Firstly there have to be people who understand the source, then they have to be motivated. e.g. paid to work on it, or have a keen interest in it, or etc. Knowledge decay happens to open source too, just slower than closed source.

I recently found a log message from reiserfs that no amount of searching could find any explanation about. Fortunately the filesystem responded to a --rebuild-tree. To me that signals that the number of people who really know the filesystem is diminishing. So I’m migrating the machine off reiserfs next rebuild.

Which is why I suggest moving off it, when planning the next upgrade. It’s also a bit pessimistic to presume you will have trouble, it’s software and presumbably there’s container file systems and such. reiserfs is working in 10.1, so getting concerned is over reacting.

The fact is, Hans Reiser wasn’t supporting Reiser3, but had moved on to his next research project Reiser4.

I second the recommendation that you migrate away from reiserfs. It
was great when it was the only journaling file system available, but
ext3 works as well and is much more widely tested. In at least the
last 6 or 7 kernel releases, ext3 has not had any bugs by the time a
given kernel reached the “stable” designation. In comparison, 2.6.27
had two killer bugs in reiserfs. That gives you some idea of the
number of kernel developers that actually employ reiserfs - very few.

At one point, all my Linux files systems were reiserfs - now only 1
remains. That is on an openSUSE 10.3 system - when it gets upgraded to
11.1, that partition will become ext3.


You have a stable enterprise deployment of SLE 10.1.
It appears not to be fully backed up, that is the Main Issue.

The OP has not said, there’s been any problem with the filesystem on the system. openSUSE 11.1 and kernel 2.6.27 issues are not very relevant, and doing any OS upgrade/migration requires backup strategy in place first.