Sorry if this has already been posted, but I've been reviewing http://technet.microsoft.com/en-us/library/dd351049.aspx so that I can author a process for our environment. The article is high-level and doesn't go into the details that I'm looking for. I'm trying to confirm the proper steps needed for my process.
Environment:
- Datacenter A and Datacenter B
- 3-Node DAG: Active mailboxes on both nodes in Datacenter A, Datacenter B's node is completely passive and activation is blocked
- DAG in DAC mode, no alternate wintness server defined.
If Datacenter A is completely down, meaning no Exchange services and no AD services, it appears that I would still use the Stop-DatabaseAvailabilityGroupcmdlet to mark the servers as down. I should be able to use the -ActiveDirectorySite parameter to stop all the DAGs in Datacenter A, but the-ConfigurationOnly parameter isn't needed since AD is inaccessible.
The next step would be to disable UM servers in the primary datacenter
Next comes activating the mailbox servers in Datacenter B. And if I'm not mistake, I'd start by stoping the cluster services on each mailbox server in the secondary datacenter.
Then I can use theRestore-DatabaseAvailabilityGroup cmdlet with the -ActiveDirectorySiteparameter to activate the servers in Datacenter B. Since there's no alternate wintness servers defined, do I need to use the-AlternateWintnessServer parameter?
Once the mailbox servers are active, I move onto activating database. I'd have to start the cluster services on the mailbox servers in Datacenter B. Then I would unblock activation. And finally, use Move-ActiveMailboxDatabase.
If this process sound right to anyone, please feel free to comment. I think I have it right, but because the KB article does not go into specifics, there are a couple things I wasn't sure of.
Also, just out of curiosity, what would the impact be if Content Indexes were in a failed state? I assume this would have to be dealt with before mailboxes can be activated in Datacenter B?
What would the impact be if the Copy Queue Lenght for a given database in a DAG is lagging behind due to a slow network between the datacenters? If I have a copy queue length of 500 for example and I am forced to failover, does this mean that there will be some data loss?
I'd appreciate any validation or additional input to this post.
Thanks in advance