desperately need help with CPU masks in vpxd.cfg

[swspjcd]

We currently are running a cluster of 4 ESX 3.5 hosts of which 4 all have E5410 CPUs in them. We are adding a fifth host that has E5530 CPUs in it. I've read several articles on CPU masking and tried several different combinations but can't seem to find the exact one to allow me to vmotion to/from the E5410 hosts and the E5530 host. Enabling EVC is just not an option right now since I believe I have to power down all of the virtuals before it can be enabled. This article ( http://blog.scottlowe.org/2006/11/23/vmotion-compatibility/ ) is extremely helpful and mentions a utility that should provide the info I need ( http://www.run-virtual.com/?page_id=155 ) but unfortunately it looks like www-run-virtual.com is down and I couldn't find the "Vmotion Info" utility anywhere else. Can some please tell me that mask(s) I need to add to vpxd.cfg to allow vmotion between an E5410 host and an E5530 host?

We have VI 2.5.0 build 227637 and 5 hosts (4 are E5410 and 1 is E5530) that are all at ESX 3.5.0 build 283373.

Thanks for your help.

[Troy_Clavell]

Why not just create an EVC Cluster

http://kb.vmware.com/kb/1005764

...and if vCenter is a VM

http://kb.vmware.com/kb/1013111

[swspjcd]

Vcenter is not a VM. From my understanding, if I enable EVC in the existing HA cluster or even create a new Cluster with EVC Enabled, I will have to have all virtual machines powered off to be able to move any host into the EVC enabled cluster. Getting permission to take an outage by powering down 140+ virtual machines is not an easy thing to do around here.

My plan was to enable EVC in the cluster next time we take an outage but that may be months from now.

[Troy_Clavell]

following the instructions in the KB article you should be able to get all your Hosts and guests into the new EVC Cluster with no downtime. Just omit the section the says to power down your vCenter VM and remove/add to inventory.

below is a good example as well

http://malaysiavm.com/blog/how-to-enable-evc-on-esx-35/

[swspjcd]

I will try this. The problem I was having is the first few virtuals I picked to move out of the non-EVC enabled cluster to the EVC enabled cluster, kept giving me CPUID errors so I assumed all virtuals would have the same problem. After I tried a few more, I found several that did not give me the error and were successfully moved into the EVC enabled cluster. Hopefully I can balance enough and still have enough processing power in both clusters as I move hosts without overloading either side.

[swspjcd]

Unfortunately as some others have posted in their comments under I am still getting CPUID errors on some virtuals but not all. It looks like I am going to have to move as many as I can and then bite the bullet and schedule downtime on the others to get them all moved into the new cluster which is unfortunately, exactly what I was trying to avoid....unless someone has the mask for the VPXD.CFG file?!