For example, each of the 4 servers has 100GB of disks and is FTT=1.
The actual disk space we can use will be 200GB.
In this case, Servers will use 40GB for each. (Free spaces are 10GB for each)
In 1 of servers fault situation vSAN will try replicate 40GB to the other three servers.
Does this cause all VMs to stop working? Or can I operate in one VM without replicating?
@hoon0715 , If a node fails then any FTT=1 data that was on this node is now going to be FTT=0 data (but still accessible and the VMs these data belong to still working) until it gets repaired back to FTT=1.
Note as well that to resync data back to FTT=1 requires not only enough space to do so, but enough space in viable locations to do so e.g. it can't attempt to resync data to any node where the remaining full copy of the data resides (as this wouldn't really be FTT=1 as both replicas would have components on a single node).
Additionally, modern versions of vSAN (6.7/7.0) won't resync any disks to the point that they have space issues - with the default settings it will pause resyncing to any Capacity-tier disk that reaches 95% full due to the resync.
@hoon0715 , Yes, VMs will continue to run with their Objects (e.g. vmdks, namespace, vswp) in a reduced-redundancy-with-no-rebuild state e.g. they can not withstand another failure but will be accessible.