Upgrade Steps Outline – DPM 2010 to 2012 SP1

System Center

What follows is the basic outline I followed to upgrade System Data Protection Manager 2010 – DPM – running on a Windows Server 2008 R2 host to DPM 2012 SP1. Obviously each step has small libraries of documentation you could read, warnings you could ignore, and whatnot – but if you’re simply looking to make sure you’ve got the right steps planned out, this is what I did to upgrade our setup:

Please note – I did this same set of steps across both of our systems (primary site and secondary site).

  1. Ensure Windows is up to date.
  2. Ensure DPM 2010 is up to date.
  3. Ensure you are running SQL Server 2008 R2 (I have SP1 — 10.50.2550.0)
  4. You are set to install DPM 2012 at this point. Install was simple – put in the disk, provide some passwords… next, next, next. Continue reading “Upgrade Steps Outline – DPM 2010 to 2012 SP1”

VHD Parent Locator Fixup Failed – DPM 2010 ID 3130

Error:
DPM failed to fixup VHD parent locator post backup for Microsoft Hyper-V \Backup Using Child Partition Snapshot\SERVER(SERVER) on DPMSERVER. Your recovery point is valid, but you will be unable to perform item-level recoveries using this recovery point. (ID 3130)

So I’ve seen this happen a couple times now on my DPM 2010 installation. I wasn’t too overly worried about it because it was an occasional alert and usually the next set of recovery points would be fine. However, after a couple VMs started having the same recovery point error on a near nightly basis it was time to start figuring out what was going on.

I found this thread which contained a link to this KB article and that was the solution for the original poster. Unfortunately that KB didn’t really apply to me because all the hosts/vms in question are 2008 R2. But, what it did do is point me to the root of the problem itself. What I had done a while back was setup checkpoints on the systems in question and then a short while later (when I was done testing a couple things) I deleted all the previous checkpoints for those systems and kept only the current running point in time. The problem was I never took the downtime to merge the differencing disks and naturally I later forgot that I needed to. So, what would end up happening is I would have say 5 minutes of one of these VMs being off as I made configuration changes and Hyper-V would begin the merge process. However SCVMM didn’t show anything like that happening, so when I was done my config changes, I would boot the VM and Hyper-V would cancel the merge process.

So, in short the solution to this error is quite simply shut down your VM, open the Hyper-V console (not SCVMM) and wait while the merge process completes. Depending on the size of the differencing disks it may take a (long) while.

And for reference – the easiest way to determine if you have a disk merge that needs to be completed is to browse to your VM storage and check to see if there are AVHD files in there. You can also open the VM’s XML configuration and look for this line:
disk_merge_pending type=”bool” True