Hi, during maintenance of a multirole Exchange 2013 2-node DAG in DAC mode we have to shutdown node2 AND the witness (deployed on its own VM, on the same Hyper-V host as node2). 2 DB's (1 active on each node), no replay.

Action plan:

Run maintenance script on node2 and check switchover - shutdown node2

Shutdown FSW

Is node1 with the 2 active DB's available after this ? And is cluster operations automatically resumed after starting first FSW and then node2 ?

Thanks :)

Why you want to shut down the FSW in Maintenance window 

If you shutdown the FSW  the DB's will not be available 

My best approach will be 

1) Put the Node 1 in MM mode. Move all the active DB's to 2nd node.

2) Upgrade the node1 

3) Remove the server from MM mode 

4) Put the Node 2 in MM mode. Move all the active DB's to 1st node.

5) Upgrade the node2

6) Remove the server from MM mode

rebalance the DB's according to your setup

Sathish, the reason for doing this is that we have to reboot the Hyper-V host with both node2 and the FSW.

Afterwards the VM hosting the FSW will be moved to another host. I understand that the 2-node DAG is operational without the FSW - would the best solution then be to shutdown and move the FSW before stopping node2 ?

When you reboot the Hyper-V host, you'll lose quorum as two of the three cluster voters go offline.

Why not look at moving the FSW share to another server so you when you reboot the Hyper-V host, you don't lose quorum? You can use the below command:

Set-DatabaseAvailabilityGroup DAGName -WitnessDirectory D:\witness -WitnessServer server1

Once done, you can go through Sathish's steps below:

1) Put the Node 1 in MM mode. Move all the active DB's to 2nd node.

2) Upgrade the node1 

3) Remove the server from MM mode 

4) Put the Node 2 in MM mode. Move all the active DB's to 1st node.

5) Upgrade the node2

6) Remove the server from MM mode

7) Rebalance the databases across both nodes

Once you're done with the maintenance, it's best to keep the witness on a host where you have no mailbox servers so you don't run into a situation where a host failure causes loss of quorum and takes down all the mailbox databases. This prevents the host being a single point of failure for Exchange.

Let me know if this answers your question.

Thanks.

As a side note, you can force quorum manually if a host goes down with the witness and a mailbox server. To do this, use:

Stop-DatabaseAvailabilityGroup -MailboxServer FailedServer

https://technet.microsoft.com/en-us/library/dd335133(v=exchg.150).aspx 

Thank you both :)

So then we could just shutdown the FSW and move the VM to another host without losing quorum ?

But also, with dynamic quorum in W12, the last node should survive a sequential shutdown of both a node and the FSW in a 2-node DAG ... ?

http://www.msexchange.org/articles-tutorials/exchange-server-2013/high-availability-recovery/exchange-2013-dag-dynamic-quorum-part2.html

https://social.technet.microsoft.com/Forums/office/en-US/ee5bac97-45af-4738-8753-6ef5322bc590/how-does-dynamic-quorum-work-for-a-two-node-dag?forum=exchangesvravailabilityandisasterrecoveryI realize that my original post was inaccurate, giving the impression that maintenance of the Exchange servers was the primary goal. In fact its the Hyper-V host requiring reboots due to config changes.

• Proposed as answer by Allen_WangJFMicrosoft contingent staff, Moderator 1 hour 5 minutes ago

Thank you both :)

So then we could just shutdown the FSW and move the VM to another host without losing quorum ?

But also, with dynamic quorum in W12, the last node should survive a sequential shutdown of both a node and the FSW in a 2-node DAG ... ?

http://www.msexchange.org/articles-tutorials/exchange-server-2013/high-availability-recovery/exchange-2013-dag-dynamic-quorum-part2.html

https://social.technet.microsoft.com/Forums/office/en-US/ee5bac97-45af-4738-8753-6ef5322bc590/how-does-dynamic-quorum-work-for-a-two-node-dag?forum=exchangesvravailabilityandisasterrecoveryI realize that my original post was inaccurate, giving the impression that maintenance of the Exchange servers was the primary goal. In fact its the Hyper-V host requiring reboots due to config changes.

• Proposed as answer by Allen_WangJFMicrosoft contingent staff, Moderator Tuesday, September 08, 2015 6:18 AM

Yes exactly in windows 2012 with dynamic node should survive on sequential shutdown of both the nodes.

As you stated the title of this post made us to focus on the maintenance of exchange , Anyhow you are in the right track now and you can proceed :)