Hi,
I was trying to nail a gotcha mentioned here: Using VSAN in two or three-node clusters? You must read this
which AFAIK says that it is hard to recover from problems in a 3 node vSAN cluster.
Using HOL 1908, I created a 3 node cluster, created a VM (RAID1, 2+1 components) and nuked a mirror component host vSAN vmk.
Then. the object went south, but in a strange way, I was hoping reduced availability but some windows showed inaccessible (which it was not).
Also, I had obviously a partition, but then it showed 2 components absent nistead of 1, which is what I had expected.
(Partition is 2:1 so the view is seeing the "small" side, so to say)
Also, I was expecting it to move to degraded after the hour, but it kept at "absent" state until I closed the lab, 90 minutes after the fault.
At this point, I would love any light on this. I'm attaching some screenshots showing what I tried to describe.