Tech Tips: VMDK File restores using Veeam - Side-by-side access

Veeam is a pretty awesome tool for doing full VM and file-level backups of VMware vSphere VMs. Every once in a while, however, you might try to do a restore that is a bit out of the ordinary. One such scenario is when you want to do a full restore of the VMDK files (descriptor and flat) and then mounting the drive alongside the original disk on the original virtual machine. You might want to do this so you can compare files between the current copy and the restored VMDK. When you attempt to mount the restored VMDK, you will run into the following message upon reconfiguring the VM:

Message from ESX Host: Virtual disks [truncated output] have the same UUID. Virtual disks with same UUID should not be assigned to a virtual machine because duplications can lead to problems when the guest boots. Make sure virtual disks have unique UUIDs. Do you want to continue?

To avoid the duplicate UUID problem, try this: after restoring the VMDK files to your host, ssh to your ESXi server and browse to the /VMFS/Volumes/ of the restore point. Once there, run the following command against the VMDK descriptor file:

vmkfstools -J setuuid .vmdk

Replace with the name of the VMDK as seen in a directory listing of the VM’s folder in the VMFS/NFS volume or in the vSphere Client (C# or Web). This will generate and set a new (random) UUID and allow you to mount the drive to the original VM without problems!

WFW Copied from http://blog.clearpathsg.com/blog/bid/303793/VMDK-File-restores-using-Veeam-Side-by-side-access

Tech Tips: Overcoming VMWare ESXi’s 2TB limit with Dell PERC 6 and H700/H800 using disk groups with many virtual disks

VMWare ESXi (version 4, I wonder about what ESXi 5 really brings) has a 2TB limit on datastore size.
So say you just bought a new Dell server and disk array, with let’s say 24 500gb disks, and wanna run it in RAID-6 for a total of 10230gb (22 * 465gb).
You’ll soon discover VMWare will limit the datastore to a little less than 2Tb. I’ll not go into the mess that is the reason for this, just accept it. Or buy ESXi 5 for US$ 10k.
One (poor) solution, would be to setup a few smaller RAID arrays (eg, four 5-disk 500gb RAID-5 arrays); each array totalling around 2tb, and create a few datastores, each with a little less than 2tb.
That’s bad because obviously you’ll waste a lot of disks; but it works.
The best solution is to make use of a little-advertised feature of RAID cards: disk groups. Since PERC 6 (PERC 6/i, PERC 6/E, H700 and H800) the card has 3 concepts: Physical Disks, Disk Groups, and Virtual Disks.
When you use OMSA, or go into Ctrl-R during boot, the ‘Create Virtual Disk Wizard’ makes it easy to create a Virtual Disk in a specific RAID mode.
So when you create a new 24 physical disk RAID-6 virtual disk, it automatically creates the Disk Group for you (it’s a ‘wizard’, right?).
What’s not so obvious is that you can put more than one Virtual Disk inside a Disk Group, effectively creating more than one LUN on a RAID array.
To do that, during the first VD create step, reduce the VD “size” field to 2000gb; it will create the first VD in the Disk Group.
Then, repeat the VD creation; it will allow you to re-use the same RAID level and Physical Disks, to create new VDs in the same group.
This way you can create a few VDs (all below 2Tb) without ‘wasting’ physical disks. In VMWare ESXi, create Datastores, one for each VD: set the VMWare block size to 8mb so that you can create large VMWare Virtual Disks in them.
For a large Virtual Machine, you can create many VMWare Virtual Disks (each up to 2tb, each in different datastores) for the VM; inside that VM’s operating system, use the OS’s features to JBOD each virtual disk into a large volume.
You can do that with Windows’s Dynamic Disks, or using Linux software raid md (in JBOD mode): just make sure to use JBOD or other non-stripped mode, because the RAID is actually done and working at the physical controller card Disk Group level.

reference: here