I ran into an issue where the AG was no longer present on a specific node of my cluster but I had NOT dropped the AG from another node in the cluster. (http://blogs.msdn.com/b/psssql/archive/2012/06/13/how-it-works-drop-availability-group-behaviors.aspx)
Note: Use the XEvent files captured in the LOG directory to confirm that a DDL command was not issued.
There are protection mechanisms built into the AlwaysOn Availability groups to protect the data. In this specific case I had the registry on Node 1 become damaged. In doing so the cluster registry no longer matched the SQL Server metadata and my AG is removed from the node.
Note: The database(s) still exist on Node 1 but they are no longer joined to the ‘My AG’ availability group. This protects the data on Node 1 from unexpected/mis-configured changes.
It is recommended that you keep the following backups for proper restore of the Availability Group configuration.
- Windows System State – http://technet.microsoft.com/en-us/library/cc771973.aspx
- T-SQL Scripted T-SQL Availability Group
Bob Dorr - Principal SQL Server Escalation Engineer