Hi. I am in the process of enabling HA for my cluster which has 3 hosts.
1st Host - Total memory: 48GB and Utilized Memory: 35GB
2nd Host - Total memory: 96GB and Utilized Memory: 38GB
3rd Host - Total memory: 96GB and Utilized Memory: 45GB
I want to make sure of a few settings before enabling HA
1) Once I enable HA, should I be worried about the hosts restarting automatically?
2) I plan on using 20% for "Percentage of cluster resources reserved as failover spare capacity" within Admission Control Policy. Would that be considered appropriate and suitable for HA?
3) I am a bit confused about the "Host Isolation Response". I know the default is "shutdown" but what exactly does this option pertain to? I will be manually modifying certain VM's so their restart priority is higher than others but just wanted to get a better understanding for the "Host Isolation Response". I've read a few articles but can't seem to grasp the meaning behind it.
4) Lastly, I understand, by default, VM Monitoring Status is disabled. Is there a need to enable this or should I proceed to enabling HA without it?
Appreciate any help in this matter.
Thanks.
Welcome to the community -
1) No - I assume you mean what happens if when ESX host restarts - if you have DRS enabled the VMs will vMotion if they are not getting enough resources -
2) HA does a good job of monitoring resource but by setting this will ensure there should be enough resources
3) HA communication is done by communication between each ESX host so an isolation respone is when one of the hosts loses communication with the remainder of th nodes in the HA cluster - so the host that is isolated will either power down the VMs and they will restart on the remaining nodes or they leave powered on and they will continue running on the host that is isolated
4) That feature is enabled if you want to monitor individual VMs - so it is not necessaery unless you are interested in montioring individual VMs
Thank you for your response. I take it in order for VMs to automatically vMotion to other hosts (should a host fail), I would need to enable DRS? I was under the assumption that enabling HA would perform that function.
That is more so my ultimate goal. I want to make sure VMs get vMotion'd over to other available hosts should a host fail or lose connectivity for long periods of time.
Thank you your help.
A good reference is:
http://www.yellow-bricks.com/vmware-high-availability-deepdiv/
http://www.yellow-bricks.com/drs-deepdive/
vmotion is only possible between 2 _live_ hosts. If a host fails, HA will do a cold restart of the vms that were powered on on that host. It's not a vmotion so memory state of those vms is lost.
Elisha
Elisha,
Thank you for your response. My main objective is to make sure the VMs (from a failed host) get automatically migrated or vMotion'd into another Live host. Am I incorrect in thinking that HA will not do that? Sorry to sound confusing but just want to make certain that HA will provide the capability to perform the task as discussed above.
Thanks again for your help.
HA will automatically migrate vms from a failed host to another live host in the cluster and power them on, but it doesn't do vmotions. vmotion can only be done between 2 live hosts, not from a failed host to a live host.
Elisha
As the others have indicated with HA there will be a brief outage as the VM is restarted on the new host because vmotion requires two live host to work