We created a VM-to-VM Anti Affinity DRS rule on our cluster entity. We also set the "das.respectvmvmantiaffinityrules" parameter from "true" to "false" on the affected cluster entity.
The affected cluster is a 2-node vSphere 7.0.3 U3c (with vCenter 7.0.3 U3c) cluster with one VM on each host and separated by the VM-to-VM Anti Affinity DRS rule. VM-A-1 is placed on ESXi-A-1 and VM-B-1 is placed on ESXi-B-1.
We reset (hard power-off) the ESXi-A-1 via its remote management interface and observed, that the VM-A-1 was not restarted on the remaining ESXi-B-1. We expected that the above-described vSphere cluster HA parameter would change this VM-to-VM Anti-Affinity DRS rule from a must into a should rule during the failover conditions.
We got the following event message: The anti-affinity rule set on your virtual machine prevents the placement of virtual disk.
But the datastore where the VM is residing is shared among the two ESXi hosts. As far as we disable the Anti-Affinity rule, the VM gets successfully restarted due the failover.
Did anyone else face this behavior or has an idea what's wrong here?
Kind regards in advance.
I have received feedback from the VMware support, and this is a known bug.
It should be resolved in vSphere 7.0 Update 3 but obviously it's not resolved. VMware is aware of this and is working on a fix for it.