Greeting everyone,
I have this environment with 3 databases all using SQL Server 2012. They are all using Windows Server 2012 and they are all in the same Failover Cluster.
We use SQL Server 2012 AlwaysOn for HA and DR. The 3 nodes are participating on the same High Availability Group.
From the 3 nodes, 2 are located in a datacenter A and the other node is located on the datacenter B. Today there's a VPN Site-to-Site closed between site A and site B for the communication. No dedicated link is available for this.
The HA between node 1 and node 2 on the datacenter A works perfectly fine. For the node 3 on the datacenter B there's configured a listeneron AlwaysOn for it on the range of the network of datacenter B since i don't have stretched VLAN. The problem starts when we have a Disaster Recovery situation envolving node 3. Whenever the VPN that links both sites drops down, lots of packet loss or something related or even if one of the sides is offline and the VPN drops the Failover Cluster gets crazy and shuts down! Yes, it shuts down. You are not able to see any of the configuration of the Failover Cluster. Just like if you click on the Failover Cluster and click on Close Connection. But obviously you can connect back to the cluster until the VPN get's back online. Really really strange.
Looking at the logs when this happens it tells me that the Cluster lost connection to the Quorum.
The quorum is configured today as Node Majority since it's the best pratice for odd number of nodes. The info was taken from this link: https://technet.microsoft.com/en-us/library/cc731739.aspx
Also, all 3 nodes are configured on the Availability Group as synchronous commit as we want the less data loss as possbile. Also all 3 nodes are marked as participants of the cluster (preferred owners).
Is something related to the quorum or any other thing misconfigured that i'm missing?
I take any help on this if possible.
If any adittional information is necessary please let me know.
Thanks in advance!