This issue has been fixed in ESXi 7.0 Update 3f
If updating to this version is not an option, you can apply the workaround described below.
Workaround:
To identify VMs affected by this problem, you can query the .nvrm file of your virtual machines for the string 'D.r.i.v.e.r.O.r.d.e.r' using the following command:
# find /vmfs/volumes -name '*nvram' -exec grep -l 'D.r.i.v.e.r.O.r.d.e.r' \{\} \;
Note: for running VMs this will only work for those that are registered on the specific ESXi host. You will therefore need to run this command on each of the ESXi hosts in the environment.
Once you have identified affected VMs, as a workaround:
- power off the VM
- delete the .nvrm file
- power the VM on
Note: the .vswp file contains multiple setting for the EFI configuration, which will get lost when it is deleted. If you plan on applying this workaround, please take care to have a current backup of the VM.