Welcome to the Forums - I will try my best to answer your questoins -
While we were working on that, other admins were powering up ESX hosts. HA was enabled on these hosts so VM's just start powering up when the hosts came up. The problem though was that several VM's were powered up and were on one ESX host, but were listed as powered off on another ESX host. I'm guessing this happened because all primary nodes in each cluster went down and without the virtual center database, hosts didn't know which one had which VM? That's my first question.
I am not sure what is happening here - but I think what is happening your ESX servers are configured to autostart your VMs and it is this that is starting your vMs - with VC down you can safely remove the VM from the ESX server where they showed power off -
My second is hypothetical, what would have happened if HA was not enabled on those clusters? When the hosts powered on, would they have even seen the VM's anymore, or would we have had to re-register all of them and then power them on? Similarly, in a more normal situation, if HA is not enabled on a cluster and one host fails, what happens to the VMs? Obviously they will shut off, but how would you go about powering them back on? Would you just have to find them in each datastore and re-register them and power them on? I ask these questions because some in my organization are now questioning the efficacy of VMware HA and I want to make sure I fully understand the concequences of having HA enabled/disabled before I argue one way or another.
Depending on how your hosts are ocnfigured - you can have VMs autostart when the ESX hosts start -
My last question is this - given the datacenter fiasco mentioned above, we're now installing a better generator to prevent this from happening again. Two weeks from we're going to have to gracefully shut down everything in the datacenter though to install it. What is the best way to do this with VMware without resulting in another split-brain scenario? Our virtual center database resides on a VM and our virtual center server is physical. I figure we should shut down each VM first, then the hosts, then the VC server. But how should we go about bringing it all up so all the hosts know exactly what to power on? Should I isolate the VM that hosts the VC database to one host, bring that host and VM online and then the VC server and then the rest of the hosts and VMs?
</div>
I would first disable HA - then shutdown the VMs except the one hosting the database - shut down VC and then shut down your ESX servers -
Theone recommendation Iw ould make is virtualize your VC server and make your SQL database physical - you will still be able to have HA working even if the ESX server hosting your VC VM fails - the database is the critical component - all informaiton about your VC environemnt is tored there -
If you find this or any other answer useful please consider awarding points by marking the answer correct or helpful
If you find this or any other answer useful please consider awarding points by marking the answer correct or helpful
