This evening, I was cycling through my VMware vSAN hosts adding a 4-port NIC to each of them, and I ran into an interesting Veeam issue.
10PM came, and it was time for my backups to run in my lab. I got an interesting error that I had not seen. Most of the backup jobs I was running were having issues and throwing an error,
“There are currently 2 usable disks for the operation. This operation requires 1 more usable disks.”
The error makes a lot of sense. As per Veeam KB2225, if a vSAN cluster has fewer than 4 nodes, and one of the nodes is down for maintenance, or some other reason, snapshots cannot be created that are required for backup jobs. This is a very interesting thought. If you have a 2 or 3 node cluster, have a host down, or in maintenance mode, and use Veeam to back it up (or any other backup product that leverages VMware snapshots), your backups will fail.
There is an exception to this rule. I realized that even though my VM-snapshot backups were failing, my StoreVirtual storage-based snapshots were not. Since you do not have the same restrictions in creating vSAN storage objects on another storage system, those backups should not be impacted.
This is yet another reason to make sure you design your vSAN clusters as per VMware recommendations. Although 2 and 3 node clusters are supported, it is always recommended to have a minimum of 4 nodes for a better level of availability.