I have a scenario.
I have read that when a host disconnects from Vcenter the Vcenter instructs master HA agent to unprotect virtual machines running on the disconnected host.
1) I tried to test it. I stopped hostd service on host. The host got disconnected from vCenter. However the virtual Machines running on the host remained still protected and when I rebooted the host it restarted successfully on another host as per HA.
2) Scenario Second: I stopped hostd service on host. The host got disconnected from vCenter. I tried to manually connect Host to vcenter(it failed which was obvious). But this time the virtual machine running on this host went unprotected and when I rebooted the host it got powered off on the same host.
Can anyone please help me know that why virtual machine does not gets unprotected in first case (as per concept it should) and why it gets unprotected in second case.
1) I tried to test it. I stopped hostd service on host. The host got disconnected from vCenter. However the virtual Machines running on the host remained still protected and when I rebooted the host it restarted successfully on another host as per HA.
the reason is in in the HA setting, you put host isolation response to "leave powered on" state
2) Scenario Second: I stopped hostd service on host. The host got disconnected from vCenter. I tried to manually connect Host to vcenter(it failed which was obvious). But this time the virtual machine running on this host went unprotected and when I rebooted the host it got powered off on the same host.
here you put the Host isolation response to "poweroff or shutdown VM" state
Hi Ramesh,
In both the cases my isolation response was leave vm powered on.
On Apr 27, 2015 3:18 PM, "dhanarajramesh" <communities-emailer@vmware.com>
ok. the first one is protected due to it has secondary heart beat connection which is data store heart beat. for second case I suspect may be you have configured affinity rule for the VMs to run on particular host. in another view I suspect, there is only two esxi in the cluster and there might be HA slots are not enough to power on vms. check your admission control policy and HA slots calculation.
Hi Ramesh,
I have tested both cases with exactly same configuration. Also, I do not
have affinity rules configured.
The answer I am looking for is if both the cases are tested with same
configuration, then why it is showing a different behavior in both cases.
On Apr 27, 2015 3:39 PM, "dhanarajramesh" <communities-emailer@vmware.com>