9 Exchange 2013 mailbox servers all running on Windows 2008 Server R2. All part of the same DAG (so just one cluster).
Inadvertently, powerer off 5 of these mailbox servers - bad idea. According to my reckoning, due to odd number of members, the quorum should be relying on a majority count of members. Since it no lonfer has a majority then the databases should be dismounted on the remaining servers.
This didn't happen. The exchange databases stayed up - albeit the CAS servers went to 100%.
We then powered back up the 5 servers but the cluster did not recover on its own. On each of the 4 servers that did not power down the cluster status still showed down. A restart of the windows cluster service on each of the 4 servers sorted this out.
Question is :-
1. Any idea why exchange did not dismount DBs when quorum was lost?
2. Why did cluster not recover when the servers were back?