**** This was solved with subsequent updates to vSphere 5.1****
VMware KB: KB Article
I’ve upgraded a non-production cluster successfully to 5.1 in a non-conventional way. Details on the hardware environment I’m dealing with can be found here.
I found a way to avoid the PSOD’s during the host upgrade, but I had to do the following:
The preliminaries to the list include changing the cluster DRS level to Conservative prior to moving a host to Maintenance Mode.
- Move the host into Maintenance Mode
- Move the host out of the cluster
- Use the HP-ESXi CD to run an Upgrade
- After the upgrade, scan the host using VUM and remediate the one patch needed
- Move the host back to Maintenance Mode
- Move the host back into the cluster
- Apply the host profile
- Change the DRS level in the cluster back to the original setting to even out the resources
The problem starts when you end up having more hosts at 5.1 than 5.0 and you initiate vMotion by entering a 5.0 host into Maintenance Mode. DRS will attempt to even things out and will inevitably try to move some VMs onto 5.1 hosts. Some migrations will fail, and some won’t. If you shut down the VMs to migrate them cold, it will work, but that’s not going to allow a smooth upgrade on the production side since downtime needs to be scheduled and rationed out.
I’m waiting for a fix on this one before attempting an upgrade on the production side.
Categories: IT Pros