Hi ASThorpe ,
The questions you have are related to HA and DRS .
HA is cluster feature for fail-over in case of host failure while DRS is a feature which helps with optimal utilization of available resources.
To answer your questions.
If we want to allow for the loss of one host (through hardware failure) would I need to add Server Z to both Windows and Linux host groups? Or would Server Z automatically replace a downed host from either host group?
Host failure is a HA event so the fail over will happen at the cluster level and it will support both the host groups as they are a part of the same cluster.
Would we be better off setting the rules as should rather than must?
It is better to run the DRS setting as should rather than must as this would protect your environment against any host failures. In other words if you have set it as must , even if there are other hosts available to run on the VMs will not power on on those host no matter what .
Finally, am i correct in thinking that the fail-over host will not be 'usable' if we put one of the hosts in a cluster into maintenance mode and additional capacity is required?
That is correct . The same explanation for the same . Placing the host in maintenance mode is a trigger for DRS and not for HA so only your DRS rules trigger in that event.
Hope this was helpful!
Thanks for your speedy reply!
I'm still struggling to understand how the VM's would know they could run on Server Z (failover host).
I agree that host failure is at cluster level, and in the event of any host failing, server Z would kick in.
With the VM host rule set to must I just can't understand how the VM's would 'know' they could now run on server Z (as it's not part of either of the windows or linux host rules).
Hi ASThorpe ,
The rule should be set to should as against must so that in the event of a failure the the VM will power on another host.
This is to say the should is a more flexible DRS rule and which would let the HA failover to kick in.
However if the must DRS rule is activated . The VMs would run on the rest of the VMs remaining 2 hosts in the linux host group.
I hope this helps