Hi Mitch. Thanks for your message and the technical report (it was instructive). My thinking is in the same line as yours: if I fail 1 drive, because there is more than 20% free capacity in the disk pool, the disk pool should be able to reconstruct the missing data and become optimal. Indeed, I actually have two disk pools, each having 60 drives, each having 1 failed drive. The problem is that one pool (named GPFS in the disk-pool.png picture attached to this message) was able to keep an "Optimal" state with one failed drive even if the free capacity in the pool is very small. But the other pool, named CLOUD, with 1 failed drive has a "Failed" state even if the free capacity in the pool is more than 20% the size of the pool. In the attached pictures (disk-drives-.png) you can see all the drives and the ones belonging to the CLOUD disk pool are marked with green.
I'm sending you in a following message some pictures with the errors from the Recovery Guru related to the CLOUD pool, if you somehow have a little time to look over them.