Red Hat

Actually this is a Centos 6.x question, but I think it fits here.

I have a client that has a pretty beefy server that will be running all sorts of VMs once I unleash it to the developers. For several reasons, they would like to do a complete clone of the server as is right now, that is with just the Centos OS installed and before anything else (including driver updates etc, as the video card Tesla P100 only runs properly with very specific drivers and doesn't have a whole lot of support just yet).

All that said, the company that custom built the server installed the OS on a 4TB drive. Can I just clone the system as is, onto a smaller, say 1TB external USB drive? This image will never get any bigger, the whole purpose is just to capture the server as is, in it's most minimal/untouched state in case they want to roll back to it at any time.

Will this work (thinking of using CloneZilla), or is this just a massive potential danger waiting to happen?

PS I suppose I should also mention, that there are four additional high performance drives in the server as well, right now they aren't doing anything/empty (and will be empty for the purposes of this clone, but they are mounted).

I would question why they wish to 'clone' the server in the first place. Do they have another identical hardware platform to put it on? If they clone the disk(s) I would say that they need to fully test that clone elsewhere in order to have confidence in it.

I would be inclined to research the correct disaster recovery (DR) from backup procedure for that OS. What format of backup media would be required. If the box is to become a production system then they MUST have a backup device (eg, tape drive) or a remote NFS volume to backup to. That backup (of which multiple copies could be taken e.g. for off-site storage too) could be kept indefinitely. The point you are making is that the system is the data NOT the hardware. Indeed, by definition, the DR would be designed to work on completely different hardware in case the computer room caught fire, the hardware was destroyed, and the same model hardware is no longer available. With a DR backup it would be possible to recover the system even if it took a little time. The data would be available to restore the system.

Usually, it is also important to have hardcopy of the disk layout, partition sizes, filesystem sizes, cylinder numbers, swap space, etc. That combined with a DR backup provides everything for recovery. By comparison, if you clone the system to a second drive and that drive fails (or somebody drops it on the floor) you are stuffed.

I'm not a CentOS expert but those are my thoughts in generic terms.

Something I've had success with is Linux, AIX and Solaris Backup and System Recovery Software which allows you to resize logical volumes and set different network addresses during the restore phase.

There are other clone tools available, such as the one from Cristie Software


There are costs with these, but it really depends what you have got and what you want to do.

Are you looking to take an image of the bare-metal server or one of the guests? There should be tools to clone the guest volumes built in.



Robin