We have a Metro stretched cluster environment with 26 hosts, 13 at each site. There are affinity groups which "should" keep HA and DRS VMs at each site, unless the site is not available.
Now, one of our hosts is not feeling too well. 
We want to test it without affecting production by putting test VMs in it. We want it to participate in HA, so that if the host crashes/bounces the VMs will come up on another host, but we really shouldn't have any other machines depending on it for HA as well..
We do not want the host to participate in any DRS until we know that the host is working properly.
Please assist in the proper way of setting this up without redesigning the entire VM environment. I thought about pulling it out of the cluster, but then there's no HA.
Thank you,
Scott.....