Windows Server 2012 in-place upgrade to 2012 R2

Associate
Joined
1 Dec 2005
Posts
803
Has anyone done this? Any gotchas to be aware of?

I know a rebuild is always going to be better... I've got 8 servers I'd like to upgrade with minimal fuss. They're all VMs so I can snapshot and rollback quite easily if necessary.
 
I havent done it, but i would suggest you copy a snap shot to maybe your local machine and try it out that way its not on the network .
 
Yeah two are DCs also doing DNS and DHCP. I'll leave them for now then or it'll be a pain changing network settings on everything.

I'll try the upgrade on a disconnected clone to start with then. Cheers guys :)
 
The general guidance I've always adhered to is to never, ever, snapshot DC's. We took backups using snapshots before we were aware of this and never had any issues other than the occasional totally locked up DC.
Fortunately, we never had to restore from one of these, or rollback to a snapshot because I believe that's when the trouble really starts.

I think things may have changed with MS virtual DC's, but these may only be available on Hyper-V. You haven't specified what hypervisor you're using, and I tend to assume ESXi until proven otherwise :-) .
 
In place upgrade will take hours. Dont snapshot, the delta file will be hudge and will make rolling it in a pain. Unless youve got some awesome storage to mask the io overhead. You are better off with a offline clone / backup.

Have you prepared the domain for 2012 dc's? Im not sure you can in place upgrade a dc.
 
Advise massive caution.

We have some Windows Server 2012 deployments which we tried to upgrade in place out of quickness. We never did a DC though, thought it was too risky.

Don't do a server which is acting as a certificate authority, RAS, acting as a secondary failover cluster (it breaks things!), a DFS branch server in a remote office (it disabled deduplication, remote differential compression and we couldn't re-enable it after upgrade), anything running System Center 2012 server side.

Recommend a full reinstall. Even following the official Microsoft guides were pretty useless, although it wasn't personally me who did the migrations. I did the migraines troubleshooting after the fact.
 
Doing in-place upgrades is getting yourself into technical debt: you may have done things quicker for the time being, but over the life of that server all the long-term problems the upgrade causes will wipe out any time savings from not doing it properly in the first place.
 
Advise massive caution.

We have some Windows Server 2012 deployments which we tried to upgrade in place out of quickness. We never did a DC though, thought it was too risky.

Don't do a server which is acting as a certificate authority, RAS, acting as a secondary failover cluster (it breaks things!), a DFS branch server in a remote office (it disabled deduplication, remote differential compression and we couldn't re-enable it after upgrade), anything running System Center 2012 server side.

Recommend a full reinstall. Even following the official Microsoft guides were pretty useless, although it wasn't personally me who did the migrations. I did the migraines troubleshooting after the fact.

Thanks Jake, exactly the sort of experience I was looking for. So forgetting my two DCs and the two servers running DFS (admittedly quite a simple setup of 1-to-1 replication so they might upgrade OK), that just leaves two web servers, a database server and a TFS server. I think I'll put this upgrade idea on ice and build fresh servers instead since I'll have to do that for half of them anyway.

Cheers everyone :)
 
Has anyone done an in-place upgrade for a VM host? Either in or out of a cluster?

We consider that too risky. Although I do have a crash n burn Hyper V 2012 server we used to have for some clients which I'm tempted to do an upgrade on see if it breaks anything before reusing it. Few weeks off yet though. Certainly wouldn't do that in production though.
 
Well chaps, I bit the bullet and went through with the upgrade to my DCs tonight. Made sure they were both up to date, took Veeam backups and cracked on with it. Everything worked! Well, almost everything - I had to re-auth the DHCP services. But that was it! AD all ok, DNS all ok, CA role on the primary server all ok... it was so smooth it was genuinely spooky.

Tempted to try the DFS servers next, I'm feelin' lucky :D
 
Back
Top Bottom