[sc:hardware-category ]I recently upgraded the memory in my servers and now it’s time to upgrade the data volumes as well.
When I built the servers, I invested in 2tb hard drives for the data volumes and used DFSR to replicate the data between the servers. This ensured I had multiple copies of the data and protected against hardware failure in a single server.
I had a close call a few years ago when my single server’s power supply decided to die. Luckily it was a nice quite death and it didn’t take the motherboard or drives out with it, but it certainly was at the front of my mind when I built the new twin servers to replace it.
I was down to about 150GB of space left on the 2TB drives and so it was time to replace them with something larger (I could have just added another drive, but I’m already at capacity for the SATA ports on the MB so it was just easier to replace it). The new drive I chose was the Segate Green 6tb drive.
My DFS servers are VM’s and since I am dedicating the entire drive to them I directly connected the 2tb drives to the VM instead of having the overhead of creating a VMDK on them. I wasn’t sure what would happen when I just swapped out the drives so I decided to play it safe.
First I shutdown the first DFS VM, then I removed the old drive and disabled networking on it.
I then powered down the VM host, swapped the drives and restarted the host.
Once it was powered back up, I took the new drive offline, added it to the VM and restarted the VM without any network connection.
Once up, I connected to the VM, created a new partition on the drive, gave it the same drive letter and then created the basic folder structure the old drive had. Windows still knew about the shares and they became available as soon as I had created the folders for them.
I then brought the server back on to the network and waited to see if replication would start.
That turns out to have been a little too optimistic 🙂
DFSR relies on information contained in the “System Volume Information” (SVI)directory on the disk, which of course was missing on the new drive. DFSR interprets that to mean that it hasn’t communicated with the server in a long time and refuses to replicate with it.
My first instinct was to remove the server from the replication group and then add it back in, making sure to force an AD poll with dfsrdiag.exe. But that didn’t seem to work, I found some reference articles talking about deleting some of the information in the SVI directory but that seemed risky.
I decided to simply delete the replication groups and then re-create them, again forcing an AD poll.
After a few minutes, replication restarted and files began to appear on the new drive. Of course replicating 2tb of data takes a while and DFSR isn’t the fastest, but it does work and after a few days everything was back in place.
Of course, I then had to do the same thing on the second server but that went smoothly and both servers now have their new 6tb data volumes.