Hi All ,

In my environment we have a setup like below.

Production site -  Active site - All the mailbox users are connected to this site 

2 mailbox servers site 

DR site - Passive site 

1 Mailbox server 

Scenario : All the three mailbox servers are in DAG .Let assume i have an cluster resources is online on the mailbox server in the DR site because of that mailbox server on the DR site is having the PAM role .So on such case i facing the WAN outage between the production site and DR site .

Question : 

While i face the above scenario i will be having majority of nodes in production site.At the same time i would like to know which server in the production site will hold the PAM role after WAN outage ? Or else cluster resources will go offline on the node at DR site and also will it make the mounted databases on the production servers to dismounted state ?

In the event of a WAN outage, I believe that the cluster resource will fail over to a server in the production site, and that server will be the PAM.  You can control the placement of the PAM with the command:

CLUSTER GROUP "Cluster Group" /MOVETO:SERVER_NAME

Hi ,

Thanks ED for your reply .

If i am wrong please correct me .

If there is no connectivity between the sites ,How could the mailbox server on the DR site will communicate and move the cluster resource to the mailbox servers in the production site ?

Hi ED ,

Thanks a lot for your reply.

After WAN outage , In exchange what would be the backend process that will happen to select anyone of the mailbox server in the production site to hold the PAM role.

Please share me any article which is related to these kind of scenarios and that would help me to have a in depth understanding about the PAM role movement during the WAN outage.

I'd have to do the same kind of Internet search you can do yourself.

I suggest you research Windows Clustering because that's what this is based on.

Hi ED ,

Thanks for your reply and support on this thread .

Let me have a research.In case if i found anything useful on internet , i will share it on this thread.

Hi,

Basic on your environment and research, the primary site will maintain a quorum as it has 2 votes(2 Nodes).
If the server that owns the cluster quorum resource fails, the PAM role automatically moves to a surviving server that takes ownership of the cluster quorum resource.
Then the databases which were active in the DR site will be failed over to Primary site based on the preferences, which will be taken care by PAM (primary active manager) active on the Primary site.
More details about Cross Site DAG / DAC Mode - Scenarios, for your reference:
http://social.technet.microsoft.com/wiki/contents/articles/22698.cross-site-dag-dac-mode-scenarios.aspx

Thanks

Hi Allen Wang,

Thanks a lot for your reply .

1.No of nodes in production site - 2

2.No of nodes in DR site - 1

3.All the databases are mounted in production site mailbox server .

4.DR site mailbox server is having all the passive copies.

5.All the cluster resources (i.e PAM role) is on DR site mailbox server .

In my scenario all the databases are mounted on the production site .But the PAM role is on the mailbox server which is available on the DR site .On such case we have faced the WAN outage between the production site and DR site .

We all know that production site is having maximum no of nodes .But the here the question how could the cluster resources (i.e PAM role ) will get moved from DR site mailbox server to anyone of the mailbox server in the production site after WAN outage  ?

Because why i am asking is after WAN outage DR site mailbox server would not have any possibilities to communicate with the any of the nodes in the production site ? So On such case how could the cluster resources (i.e PAM role ) will move from DR site to production site  even though the production site has majority no of nodes ?

Please correct me if i am wrong .

It doesn't "move" so much as stop and restart. If a node does not have quorum, it stops hosting the PAM. If there is no PAM among nodes that can communicate with each other, they elect a new one. No communication with the other site is necessary.

The same goes for database moves, btw. The active copy of a database that fails via your scenario doesn't really "move" as much as dismount, and then the PAM mounts a different copy.