HA not HA'ing

[billz0r]

We have a 2 hosts and a NFS Shared storage location.

I'm trying to test HA and FT before implementing for a client.

I got FT working great (stupid 1 vcpu limitation) so worked on to HA. I have HA enabled in my cluster and turned FT off on the VM. To test HA I'm unplugging the network from the host running the test VM and hoping it will restart the VM on the other host but nothing happens.

What am I missing? I have read through the VMWare doco's on enabling HA and there really doesn't seem much to it except tick a box in the cluster.

[idle-jam]

when you able the HA, did you see any error message? if there is none then VMware HA is working fine.  you might want to read the below URL on how they detect failure. i would suspect that you would need to unplug all cables as some of plugged cable might be used by HA as heartbeat.  http://www.yellow-bricks.com/vmware-high-availability-deepdiv/

[sflanders]

More information about the host and VM are required. How many NICs does the server have? How many are you disconnecting? Which ones are you disconnecting? I agree with the previous comment about the heartbeat network still being accessible and the link to Duncan's site is a great reference. What last question would be what is the host isolation response set to? Based on your post it appears you only enabled HA, but still worth checking. See: http://kb.vmware.com/kb/1018325.

Hope this helps!

[billz0r]

HA is only on.

The Host the Test Vm is running on has 2 NIC's in load balance. I'm unplugging both of them so the virtual network and the Kernal can't access anything. I see the Host go offline in vCenter but never start the VM on the 2nd Host. For host isolation I've tried Shutdown and Power off. HA is only enabled, FT and DRS are off. I get no errors in the VIClient other than acknowledging the NIC have been disconnected.

Edit: I've reconfigured the NIC's. 1st NIC for Service console and test VM, 2nd NIC for the Kernel. Unplugged the VM/SC NIC and it worked. I'm worried now that if the Kernel NIC goes down it won't restart the VM's on the oher host(s) in the production evironment.

[billz0r]

Looks like I solved it.

I unplugged the Kernel NIC and it did fail over eventually, it took a little longer than before but it did work.

Looks like the problem I had was putting the VM, Service Console and Kernel all on 1 pair of redundant NIC rather than seperating the functions out.

[billz0r]

Well a new twist, I put all roles (S/C, VM and Kernel) back onto the 2 NIC's in 1 vSwitch but didn't select any of the NIC teaming options and it works.