So I've been trying to get vSAN to work in my lab with limited success. I'm running 6.0 U2 and setting up a stretched cluster config. Initially I could get vSAN configured but the health check showed network errors, and data was not available in both fault domains even though all the vSAN vmk IPs were available to every host. I finally noticed the unicast agent IP was set to the vmk for MANAGEMENT of the Witness VM, not the IP for VSAN. I used esxcli to change the unicast agent IP to the correct IP on both storage hosts and now all of my data objects show as healthy and everything is replicated, but I'm getting an error in the vSAN health check for "invalid unicast agent". I am using the VSAN Witness VM with the default configuration, I only changed the IPs on the vmk interfaces for my network. Anyone else managed to get vSAN up without any errors in a stretched cluster? Any ideas why these errors are occurring?
Have a look at that kb, Virtual SAN Health Service - Invalid unicast agent test (2144398) | VMware KB
Also make sure that the routing paths and network in general are configured properly to allow the communication between them.
did you find a Solution?
We have the same Problem, with the difference, that it does not create the unicast agent automatically after VSAN activation. (ROBO config)
To make vsan work, I needed to create the unicast agent manually.
But after that, it also shows the "Invalid unicast agent" error.