At a client's site with 3 ESX 3.0.1 servers and a VC 2.0.1 server, we needed to reboot one of the ESX servers after installing the Dell OpenManage software. The ESX box in question did have 2 (dev/test) VMs running on it at the time I issued the reboot command through VirtualCenter. I watched the 2 VMs stay on that ESX Server even when the ESX server was restarting. Then, once ESX was down on that server, the VMs died off as well. About a minute afterwards, HA said that it detected a host failure on that server. Then the VMs that went down with that ESX Server restarted on a different running server in the cluster.
My question is, shouldn't VC know enough to automatically migrate running VMs from one host to another in this situation?
I believe the proper procedure for this activity would have been to place the server in maintenance mode, manually vmotion the virtual machines to another host, and then perform the reboot.
... but in general, HA has just enough functionality to make you wish for more.
I believe the proper procedure for this activity would have been to place the server in maintenance mode, manually vmotion the virtual machines to another host, and then perform the reboot.
... but in general, HA has just enough functionality to make you wish for more.
Yeah I had forgotten about Maintenance mode, thanks. Incidentally, when I tested on another ESX server and entered Maintenance mode, it did automatically migrate the VMs to other hosts. I then rebooted the ESX server. After the reboot, the server was still in Maintenance Mode, so I did have to exit it from that mode. I also noticed that upon exiting Maintenance Mode, it did re-configure for HA.
manually vmotion the virtual machines to another host
This should automatically happen while the maintenance mode job is running (i.e. maintence mode isn't complete until running VMs are vmotioned off).
After the reboot, the server was still in Maintenance Mode,
This is by design. Lets say you were doing something that required multiple reboots (hardware diag, or patches that required reboot) - you wouldn't want VMs moving back after every reboot!
Dave
I believe that DRS is responsible for auto-migrating VMs when you put a host into maintenance mode (obviously assuming you have DRS configured on your cluster).
Good point. I always enable DRS..so I have never seen it not work
Dave
aint it lovely when things "just work"