I am looking in using the High Availability option for Clusters in VMWare for our current VMWare Server setup but I have a question. I want to make sure it only does certain Virtual Machines. Do I do that by dragging the VMs into the cluster separate from the physical hosts? Then setting the tolerated failovers to the number of VMs I want it to failover?
HA will still kick in. You need VC to setup HA and manage it, but the hosts handle the failover and heartbeating themselves. The vc server would be restarted on live hosts, and then when it's up, the inventory will be updated.
-KjB
check the settings or your HA setup. Virtual Machine Options may be what you are looking for.
As Troy pointed out it is under the per VM settings - when setting restart priority you have the option to disable which will prevent the VM from restarting during an HA event -
If you find this or any other answer useful please consider awarding points by marking the answer correct or helpful
The tolerated failovers number is how many hosts in your cluster you are allowing the cluster to allow failure on. hose failed hosts will have their vm's restart on live hosts. Anymore than that number, and no more failovers. It is not for the virtual machines themselves.
-KjB
So as far as tolerated failovers, if it is set to 1 for example, it will allow one host to fail and then if you have 2 hosts in that cluster they will be vmotioned over to the other hosts. If the 2nd fails then it will not do anything with that one right?
Also as far as adding hosts in it. You just have to drag the physical server in the cluster right? Also It will not affect them if they are currently running correct?
Also if you need to take a server down, how do you stop it from moving everything to the other server? Put into maintenance mode?
One other question. To move the Virtual Machines back is that a manual move or does it remember which physical server that they were originally on?
Thanks in advance.
So as far as tolerated failovers, if it is set to 1 for example, it will allow one host to fail and then if you have 2 hosts in that cluster they will be vmotioned over to the other hosts. If the 2nd fails then it will not do anything with that one right?
No - If it is set to one 1 - when the first host fails the vms that were running on that host will be restarted on the remaining nodes of the HA Cluster - when the second hosts fail the VMs that were running on that host will be down and will not be restarted - if the setting were set to allow two host failures then the when the second host failed the vms would restart on the remaining nodes - also note vmotion has nothing to do with HA - vmotion requires running ESX Servers on both sides in HA one of the hosts is down - the vms restart -
Also as far as adding hosts in it. You just have to drag the physical server in the cluster right? Also It will not affect them if they are currently running correct?
Yes - just drag the ESX server into the cluster and it will not impact anything if they are running -
Also if you need to take a server down, how do you stop it from moving everything to the other server? Put into maintenance mode?
To remove a host from a cluster you will need to place it maintenance mode
One other question. To move the Virtual Machines back is that a manual move or does it remember which physical server that they were originally on?
If you have DRS it will move the VMs as the loads increase and the VMs are not receiving their resources - if you do not have DRS then yes you will have to manually redistribute the Vms with vmotion -
If you find this or any other answer useful please consider awarding points by marking the answer correct or helpful
So it does not vmotion them over. It just restarts them on the other host? How does that work does it basically remove them from one server inventory and add them to the other?
Just wondering.
Pretty much - yup.
--Matt
As Matt said Yep -
If you find this or any other answer useful please consider awarding points by marking the answer correct or helpful
From a stand-alone ESX server perspective, sort of. Since the host that was failed, it wasn't removed from that hosts' inventory, per se, but it was picked up by an active host and restarted. Since the servers are all under vc control, the registration/deregistration happens automatically. The restarted machines also perform a quick restart, so they skip a few things (bios load, etc) that a typical poweron/restart event would undertake, allowing the vm to boot up faster than they normally would. This minimizes the downtime you would experience in this situatioin.
But, if you wanted to do this without VC, then you would add the vm into the inventory of a server that was still up and running, provided you were using shared storage.
-KjB
What happens if the machine running VC Server is on one of the hosts that goes down?
HA will still kick in. You need VC to setup HA and manage it, but the hosts handle the failover and heartbeating themselves. The vc server would be restarted on live hosts, and then when it's up, the inventory will be updated.
-KjB
VC will go down but HA will still work restarting VC on another host
David Weinstein,
A new message was posted in the thread "Vmware HA Clustering":
http://communities.vmware.com/message/1160732#1160732
Author : jauner
Profile : http://communities.vmware.com/people/jauner
Message: