What is this about “two servers and intentionally damaging” btw? Do you have this resilience handled at a higher levele than I (RAID)? How does that work?
Heh - “complicated” is SO in the eye of the beholder. 
1 Like
I have 2 of the containers running off a ZFS RAID and 2 off a single external HDD for testing purposes.
The default 4 way clone stores the data 2 times locally on disk and “guarantees” integrity and continued servicing of the files as long as N/2 servers are still alive. And guarantees you’ll still be able to upload files while you have quorum (N/2 + 1 instances alive). But this 2 way replication is no good to me if it doesn’t automatically heal missing/corrupted data on each store since, given enough time, the corruption could happen on each host. Best way to confirm it works is to just remove the entire bucket folder locally from 1 or 2 of the hosts and seeing what happens.