1

I have an oldish (physical) Windows Server 2003 R2 box, which is a file/print server. I suspect the RAID card has gone bye-bye, after a reboot Windows is checking the data volume, its been 1.5 hours or so. I'm starting to prep for the worst and I have a few questions.

First up, I have extensive experience running Windows shops up until the 2003 R2 sort of time period. At my current job we're mostly running Ubuntu for applications, with Windows servers fulfilling background roles like AD, file & print, DNS, DHCP, etc. As a result, I have not kept pace with the finer details of Windows Server 2008 and above.

Some config and other info that's relevant:

  • This server had its RAID volume split into 2 partitions, C: = the OS, D: = the data.
  • Only the data (D:) volume is backed up. We use Bacula for this. The last backup was 24 hours ago, so the worst case is that I lose 1 day of changes to data.
  • There's as near as makes no difference 1TB of data.
  • Our MS licensing is all good, such that I can install the very latest server OS if I want.
  • I have a VMware cluster in this site, I'll be building a new VM to be the replacement file server.
  • This file server was involved in some domain DFS-related stuff, namely:
    • It did not host any DFS namespaces, only targets
    • It hosted 2 folder targets under DFS namespace A. Neither of these targets were replicated to other server.
    • It hosted 2 folder targets under DFS namespace B. Both of these are replicated *using DFS-R, not NTFRS) to a server at a remote site, which is still fully operational.

First, I'm planning to use Windows Server 2008 R2 for the replacement file server, as I think 2012 is still a bit too new, especially for someone who hasn't kept up with the finer details of the newer releases. Are there any gotchas I should look out for when adding the new 2008 R2 server to the existing DFS setup?

Second, my VMware datastores support a maximum virtual disk size of 512GB. Since I need at least 1TB, what's the typical way to do this? Create multiple virtual disks and create a Windows striped volume on top of them?

Third, how do I go about restoring data and joining the existing DFS-R setup without generating a massive amount of replication traffic between here and the remote site?

ThatGraemeGuy
  • 15,314
  • 12
  • 51
  • 78

1 Answers1

1

There are no particular gotchas replacing a 2003 DFS replica with a 2008 R2 replica. Obviously, you won't be able to take advantage of the newer DFS features, but you probably weren't expecting that anyway. Just make sure you remove the old server as a target, and add the replacement server in its place.

I'm not sure about your disk size limit on your datatstores; my honest response would be to create new datastores with >2MB blocks and use that/ehose datasotre(s). (Assuming that's the cause of the disk size being limited to 512 GB).

Regarding minimizing the replication traffic, the best solution I've found is to pre-populate the server with the data in quesiton, in this case, restoring the data backup to what will be the DFS target on the server before adding it back in as a DFS target.

HopelessN00b
  • 53,385
  • 32
  • 133
  • 208
  • Thanks. Unfortunately I cannot reshuffle virtual disks and create clean datastores in the timeframe I'm working with, so that's out for the moment. For the DFS(-R), you seem to be confirming what I suspected, i.e. put the data back, share it, and add the share, and DFS-R is smart enough to fix it up. I think it was dear old NTFRS that needed very specific restore steps, using ntbackup and so on, or else it tended to clobber both sides and leave you with a lovely mess. – ThatGraemeGuy Nov 06 '12 at 19:22