So I am planning to test the HA to make sure it is configured properly. I would like to do this remotely from through my VPN connection, hence I can not really physically unplug the network cable? My question is how I can simulate an HA situation and more importantly, I would be able to get everything back to the place where they were after. Strategy please.
VI3.5, two HP hosts with HP EVA SAN.
Couple things. We have HP systems, so we can use ILO to shutdown an ESX server running VM's. Do this from the ESX server console, not from VC. You can also find out what switch ports these boxes are plugged into, and simply do a shutdown on the ports.
Couple things. We have HP systems, so we can use ILO to shutdown an ESX server running VM's. Do this from the ESX server console, not from VC. You can also find out what switch ports these boxes are plugged into, and simply do a shutdown on the ports.
Smitty23, yeah, good point. We do have basic iLO (shutdown and power on) but without console module. So I guess I can just shutdown the ESX host without shutdown VMs first, would that be enough to simulate an HA situation? Also, just want to understand HA better, what exactly kind of "failure" to lure HA step in? The shut-down-switch-port option is definitely my consideration too.
Through the VI Client you can issue a a reboot command to your esx host -
You can disable the ports on the switch (production network facing) not SAN SIDE that the one of the server is connected to. OR Just place your server in maintenance mode which will softly migrate everything over to the other esx hosts within the cluster.
If you found this information useful, please consider awarding points for "Correct" or "Helpful". Thanks!!!
If you have shutdown and power on, then you can ssh in, and turn off networking , service network stop. That should kill your NIC, and force HA to kick in. Then, you can restart the server, and get it to come back to normal after restart.
-KjB
PLaying with the networking really would not cause an HA failover - it would cause an isolation response - so it will all depend on the isolation response is ocnfigured for the vms - I still say the best was to test ha failover is rebooting the host form the vi client -
Thanks everyone! To find out which suggestion works better and to my own interest, I would like to try different approaches and get back to you all.
so i didn't get much time to test all possibilities. however, i did turn off the host through iLO, all running VMs were restarted on the other host in the cluster. HA worked. thanks again folks.