I had to do a Exchange Datacenter Switchover Procedure on a 3 Node Dag. Two on primary site and one on Secondary site.
Primary datacenter needs maintenance and both primary nodes have to go offline. so we had to do a datacenter switchover prior to the primary datacenter maintenance . The procedure is simple if you plan it properly. lets see the aspects we checked. Before the maintenance of the primary datacenter.
- Made Outbound Mail flow to go via the secondary site.
- Made MX Records / Inbound Mail flow to come via the secondary site.
- Public DNS Records changed to the secondary site.
We made sure DAC mode is Enabled with Value DagOnly.It will avoid the split brain syndrome across these nodes.
Get-DatabaseAvailabilityGroup -Identity DAGNAME | Fl Name,DatacenterActivationMode
To Enable DAC – you can set it with
Set-DatabaseAvailabilityGroup -Identity DAGNAME -DatacenterActivationMode DagOnly
Make sure witness server and Alternate witness server is specified
Make sure no activation block is set on any mailbox servers
To Check
Get-MailboxServer | fl name,DatabaseCopy*
All Activation block has been removed
Get-MailboxServer | Set-MailboxServer –DatabaseCopyAutoActivationPolicy "Unrestricted"
Check where the quorum and move to secondary site
Get-ClusterGroup "Cluster Group"
Its showing my primary site server.
So we moved it to the secondary site
Get-ClusterGroup “Cluster Group” | Move-ClusterGroup –Node SECONDARYSERVERNAME –Verbose
Moved all the active databases to Secondary Server
Get-Mailboxdatabase | Move-ActiveMailboxDatabase –ActivateOnserver SECONDARYSERVERNAME
Now
Stop-DatabaseAvailabilityGroup –Identity DAGNAME –ActiveDirectorySite "Default-First-Site-Name"
To check your site name
nltest /dsgetsite
To check on Completion
Get-DatabaseAvailabilityGroup –Status
Stopped Dag nodes will be removed from Operational server list.
Now go to your secondary site server – Run
Stop-Service clussvc
Ran below on Secondary site
Restore-DatabaseAvailabilityGroup DAGNAME –ActiveDirectorySite "SECONDARYSITENAME"
Now Alternate witness become active and it formed a single node cluster without the primary site servers and databases stayed active on secondary site as we moved prior.
Opening Failover cluster to verify the same.
Now Primary servers can be safely turned off
—
Once Maintenance is over in the primary datacenter. Run below
Start-DatabaseAvailabilityGroup DAGNAME –ActiveDirectorySite "Default-First-Site-Name"
Failover Cluster is Back to Normal
If you see witness is not changing back to normal . you can run
Set-DatabaseAvailabilityGroup DAGNAME
Note :
Odd Nodes – no witness server works on node majority
Even Nodes – works with a witness server to act as a node.
Make sure Quorum is moved back to primary site
Get-ClusterGroup "Cluster Group"
Get-ClusterGroup "Cluster Group" | Move-ClusterGroup –Node PRIMARYSERVERNAME –Verbose
Now make sure all copies are healthy or you can update it with
Update-mailboxdatabaseCopy "Databasename\server" –DeleteExistingFiles
Make sure Mail flow and Public DNS changes is done via the primary site.
Datacenter Switch-over process has been successfully completed.
