Yeah, for example: Cassandra's pretty happy with a node going down these days and has matured reasonably well. My company is using it for their production systems, and the only problems we've seen were either us doing something wrong, or client-library quirks for certain languages (they favor Java/Python in those parts).
NoSQL has a place, but IMO that place is not primary data storage (in most cases.)
I asked one large, household-name IT company on a recent interview what they'd do if their DC containing all of their Cassandra nodes went down. Their response was "it doesn't go down" (and if it did, they'd lose data)...
...
...
amateurs.
