Anyone have a script that groks the latest support...
# harvester
b
Anyone have a script that groks the latest supported CPU model from all the nodes? So you can set a cluster wide setting and not worry about what VMs can migrate where? If not I might write one.
m
sounds like a good idea. fwiw, kubevirt publishes lots of info about cpu model, features etc. as node labels. is that what you are referring to?
b
yeah, typically the supported models show up with something like...
<http://cpu-model-migration.node.kubevirt.io/Skylake-Server|cpu-model-migration.node.kubevirt.io/Skylake-Server> : true
and
<http://cpu-model-migration.node.kubevirt.io/Haswell|cpu-model-migration.node.kubevirt.io/Haswell> : true
etc.
So if you have a Cascade Procs:
<http://cpu-model-migration.node.kubevirt.io/Cascadelake-Server|cpu-model-migration.node.kubevirt.io/Cascadelake-Server> : true
It'd start with that then seing that not all the nodes have it, move onto
Skylake-Server
etc until it found the latest supported processor
Then you could set the kubevirt setting for the cluster and ensure your vms could migrate to all the hosts.
It just gets a little messy to try to find things on your own when it looks like this:
If anyone wants it, here's a script for getting common CPU models from a harvester cluster and (hopefully) a best recommendation for compatibility. Requires
kubectl
access to your harvester cluster.
👏 1
👍 2