The "Basic Deployment", as described in the documentation, uses a 2+1 member replica set as the application database. As all writes to this database seem to use w=2, loss of one of the data bearing nodes means that Ops Manager is no longer operational. Users cannot log in, and no monitoring data can be ingested.
Ops Manager needs to provide a better error message (actually, right now it doesn't report any error in the UI at all) when this situation occurs and users cannot even log in.

And if two members of a full three-member replica set were lost, there would still have to be a conclusive error message so users can understand why Ops Manager is no longer functional.

Thanks for considering this.