DougtheWee
Contributor
Contributor

vMotion configuration on HA Cluster

HI all. Wondering if someone can point me in the direction of a walkthrough of the requirements and setup procedure of configuring and enabling vmotion on a 2 node cluster.

I have a server running vCenter Server and two nodes running ESXi 5.1 and "seem" to have successfully created a 2 node cluster with HA and DRS enabled. I've deployed a couple of VMs to the cluster and they have been deployed to seperate hosts automatically, and showing as being "vSphere HA" protected, though interesting if it right click on the VM the "Fault Tolerance" option is greyed out. Is this normal?

ON the hosts summary I can see both hosts have a HA state of connected with one as master the other as slave, vmotion is enabled, but showing as not configured for FT. I'm guessing I have missed a step which is why I'm lookign for a walkthrough to see what I have missed.

Thanks

0 Kudos
9 Replies
karthickvm
VMware Employee
VMware Employee

HI Doug,

Check if the FT traffic is enabled on the VMKERNEL port.

http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=101342...

Karthic.
vRNI TPM
0 Kudos
DougtheWee
Contributor
Contributor

Thanks for that. No I dont have FT logging enabled on any of the NICs. But I think I have a more fundamental question. What I am looking for is a solution where VMs will be migrated automatically to alternate nodes should one fail. This may be a stupid quesion but does FT need to be running to accomodate this, or will the fact that I have built a HA enabled cluster support what i need on it's own?

0 Kudos
aravinds3107
Virtuoso
Virtuoso

Check Turn on Fault Tolerance option is disabled and see if you have done all configuration

If you find this or any other answer useful please consider awarding points by marking the answer correct or helpful |Blog: http://aravindsivaraman.com/ | Twitter : ss_aravind
0 Kudos
aravinds3107
Virtuoso
Virtuoso

What I am looking for is a solution where VMs will be migrated automatically to alternate nodes should one fail

HA would do this for you.. If one of the host fails the VM's would failover to other host in cluster (you will have minimum downtime)

If you find this or any other answer useful please consider awarding points by marking the answer correct or helpful |Blog: http://aravindsivaraman.com/ | Twitter : ss_aravind
0 Kudos
DougtheWee
Contributor
Contributor

Thanks that encourages me that I'm going in the right direction. I'm guessing though that if I go down the full FT route, the failover to a new host will be transparent rather than having some downtime whilst the VMs move to the new node.

0 Kudos
aravinds3107
Virtuoso
Virtuoso

I'm guessing though that if I go down the full FT route, the failover to  a new host will be transparent rather than having some downtime whilst  the VMs move to the new node.

Yes that's right but FT has certain limitation like only 1vCPU, no snapshot, no Storage vMotion, 4 FT VM's per host etc..

Read VMware Fault Tolerance Requirements and Limitations

If you find this or any other answer useful please consider awarding points by marking the answer correct or helpful |Blog: http://aravindsivaraman.com/ | Twitter : ss_aravind
0 Kudos
DougtheWee
Contributor
Contributor

Having read through that article it looks like FT is not going to be viable for what I want due to the limitations of the vCPU. I will be running VMs with multple vCPUs which puts the kybosh on that.

What sort of failover times are we looking at for a VM to move to another host? I'm running a dedicated GE network for vMotion.

0 Kudos
aravinds3107
Virtuoso
Virtuoso

What sort of failover times are we looking at for a VM to move to another host?

HA would get triggered when the other host in the cluster doesnt recevie an hearbeat.

Read http://www.yellow-bricks.com/vmware-high-availability-deepdiv/ which will talk in depth about HA

If you find this or any other answer useful please consider awarding points by marking the answer correct or helpful |Blog: http://aravindsivaraman.com/ | Twitter : ss_aravind
jdptechnc
Expert
Expert

DougtheWee wrote:

Thanks for that. No I dont have FT logging enabled on any of the NICs. But I think I have a more fundamental question. What I am looking for is a solution where VMs will be migrated automatically to alternate nodes should one fail. This may be a stupid quesion but does FT need to be running to accomodate this, or will the fact that I have built a HA enabled cluster support what i need on it's own?

If you are looking for "zero downtime" in the event of a host failure, you would need FT, but as mentioned above, there are some severe limitations for FT vm's, the biggest show stopper for most people being only one vCPU per FT protected VM.

Enabling HA would handle automatically restarting your VM on an available host, which satifies your "automation" requirement.  But not this is not a vMotion.  If hosts in the cluster detect a host failure, the VMs hosted by that failed host are restarted by one of the surviving hosts, so there would be a brief outage.

Please consider marking as "helpful", if you find this post useful. Thanks!... IT Guy since 12/2000... Virtual since 10/2006... VCAP-DCA #2222