I currently own a project to get us off our old Storevault S550 SAN and onto our new NetApp FAS2240 SAN. The old Storevault is having trouble authenticating with our new domain controllers as we simultaneously upgrade our AD stack from 2003 to 2008.
Our Exchange 2003 server, a physical cluster built up by an engineer who’s no longer with the company, is running on Windows 2003. All of its clustered storage is backed by the Storevault.
Things get really unpleasant when we start talking about NetApp Snapdrive — this Exchange cluster has Snapdrive 5 installed. Snapdrive 5 can talk to the Storevault, and works on Win 2K3, but cannot talk to a NetApp SAN running OnTAP 8.2 and above. Our 2240 is running 8.2.1P1. The latest version of Snapdrive which I’m using elsewhere in our environment, 7.1, is incompatible with Windows 2003.
On top of all that, we’re dealing with old physical servers in a cluster, so there’s a real likelihood of breaking the cluster if we screw this up, since we can’t just take a VMWare snapshot and roll back if we really get up a creek.
So, we’re pretty well stuck — where do we go from here?
I tested intermediary versions of Snapdrive — every version of Snapdrive v6, and earlier versions of 7, and either they install OK and can’t talk to either the Storevault or the NetApp, or they don’t install at all. Version 6.5 appears to permit communication to the Storevault, however lots of “this API is no longer supported” errors when connecting to the 2240. I’ve seen this error before with NetApp, generally it indicates there’s a version problem somewhere in the chain.
I had found a blog post from this guy who seemed to get all this working:
OK then. NetApp Support, and even my local sales rep, told me this is a special release that actually requires special authorization to even support, and they weren’t willing to grant me support for it; they made some effort to find a workaround, but I decided we’d just have a go at things manually.
So, I brought Snapdrive up to the latest version I could, 6.5, and went on my way.
The downside of this is that LUNs now need to be connected manually via mapping through OnCommand, and scanning for them via server’s Disk Management. But, I was able to use this workaround to get the next phase of the project moving along. So, referencing the ServerFixes blog above, there’s a workaround out there that seems to have worked for some, but in my case it did not — if you find yourself stuck in this situation, it’s worth exploring.
Next we’ll talk about getting data off an S550 Storevault, onto the NetApp, with not more than 30 minutes of downtime, and how we got the 2003 Exchange cluster happy.