I'm sure our listeners will be interested in this PDF:

Redundancy does not imply fault tolerance: analysis of distributed
storage reactions to single errors and corruptions

https://www.usenix.org/system/files/conference/fast17/fast17-ganesan.pdf

While testing zfs in a non-mirrored config seems .... pointless, the
commentary is interesting: the vast majority of databases tested both
fail to detect corruption, and happily propagate it to replicas. In some
cases, notably if the node is the current leader, then the entire
cluster can become unavailable as those errors are diligently
transferred to other nodes.

Ouch.

A+
Dave