We are currently running a 3 node DAG, with 2 servers in our main datacenter.  The active databases are split between these two servers.  Our DR site has a third DAG member that has no active copies.  We have a third site that I have our witness server setup and at the recovery site I have an alternate witness server configured.  DAC is enabled on the DAG.  This weekend we plan on testing failover of our entire datacenter.  I believe I will need to follow the Datacenter switchover procedures to fail us over?  Is this the process:  https://technet.microsoft.com/en-us/library/dd351049(v=exchg.150).aspx

John

It makes no sense to have an alternate FSW in a third site when you have three DAG copies.  The right place for the alternate FSW would be the DR site, but it wouldn't come into play because there's only one node there.

Your link is correct for what you're asking.

Note that failovers and switchovers are different things.  A switchover is planned, and generally less risky.  A failover just happens.  In your architecture, I would think carefully about whether you want to allow a failover.  Consider configuring this on all your DAG members:

Get-MailboxServer | Set-MailboxServer -DatabaseCopyAutoActivationPolicy IntrasiteOnly

I think my wording may have been confusing so let me see if this helps:

Datacenter, 2 servers from the DAG.

DR site, 1 DAG server and alternate witness server.

Third site which is connected holds the witness server.  

I have run the command but only in the DR site, you are saying I should run it against all three.  Also, how can we test failover if we plan a switchover?  I have successfully run all DB's from the DR site but but the other servers were online.   

Thanks. 

I understood everything you repeated the first time and my remarks still apply.

The command is not site-specific, so if you ran it, it applied to all servers.  Please understand that the command says, "Don't fail over databases between sites."  You can manually switch them over, however.  Just switch over a database and see if it works.  In Exchange 2013, it should if you have enough bandwidth between the sites to allow the data to flow.

Full testing requires an outage.  My recommendation is to write a detailed procedure, then use the outage to test it.  Consider hiring a consultant who's experienced in these matters to assist you with developing the procedure.  I recommend you test and refine the procedure at least twice.

Thanks again, we have successfully run off the DR site server, just not during an outage.  Also I re-ran the command since when I ran get-mailbox server it said only my DR was Intrasite only.  After running it again all three servers show Intrasite only.  



Be aware that if you use the .\StartDagServerMaintenance.ps1 and .\StopDagServerMaintenance.ps1 scripts to put a server in and out of maintenance mode, the latter script leaves the server in Unrestricted mode, so you'll want to add running the Set-MailboxServer cmdlet afterward. 

You're welcome, happy to have helped.  Please feel free to mark posts as answers and/or helpful as appropriate.