The tutorial "Back Up and Restore with Filesystem Snapshots"] provides adequate discussion of the use-case where data and journal have been placed on separate volumes, but does not address the use-case where (just) data has been distributed across multiple volumes (which is likely functionally equivalent to separating data and journal).
This should probably also mention that the technique is not suitable (or supported) for sharded clusters.
Need to answer a few questions before proceeding:
- Is this a normal kind of deployment? Do we want to support a deployment where the database might be fractured between multiple volumes?
- How would a user configure this? Would be helpful for us being able to validate any instructions produced
- Does the existing instruction work for this situation?