We have just come across a problem with ESX 3.5 Update 1...
We have a cluster of 4x IBM x3850 M2 hosts with ESX 3.5.0, build 82663 (ie Update 1) installed.
We have a VirtualCenter 2.0.2 VM running on this cluster, to manage our large Pre-Production (test lab) ESX 3.0.2 server environment.
One ESX 3.5 host in the cluster of 4 mentioned above, was yet to be patched with the 3.5.0 Update 1, and this was the host that the VC 2.0.2 was running on. This VC VM was vmotioned to a host which already had the Update 1 installed.
Then the VC 2.0.2 failed - our VI Clients were disconnected from the VC 2.0.2 VM, and failed to connect again. We could login on the VI Client, the client would connect ok, and show "Loading inventory..." for an hour or more and nothing would appear.
We tried restarting the VC service on the VC server, and restarting the VC server too, but nothing would get the inventory appearing. Running Task Manager on the VC server showed very high CPU utilisation of 70-100% and long periods on 100%, much more than the 20% or so that it would normally do.
To see if it might have been the Update 1 which caused this, we powered off the VC VM and registered it on the older ESX 3.0.2 cluster that it used to live on a couple of months ago, which uses older IBM X3850 M1 hosts. It powered up ok and we were quickly able to login to this VC using our VI client, and it was performing much better, with lower CPU utilisation than on ESX 3.5 Update 1!
We didn't get any other complaints from the 50+ VMs on the ESX 3.5 cluster about performance - only for the VC VM!
Has anyone else had similar problems with ESX 3.5 Update 1?
Just a quick question, I noticed that you are using VC 2.0.2 instead of VC 2.5 for ESX 3.5.
Any reason for this?
Just to clarify, we are NOT using VC 2.0.2 for managing the ESX 3.5 hosts, but for our ESX 3.0.2 hosts! We have another VC 2.5 server to manage the ESX 3.5 hosts.
It just so happens that our VC 2.0.2 is a VM on the newer ESX 3.5 cluster.