When gVisor is enabled, hyper-threading is disabled (https://cloud.google.com/kubernetes-engine/docs/concepts/san...) and GKE cluster shows only half the CPU as available resources.

Does n2d instance is considered safe from CPU vulnerability and safely enable hyper-threading when used with gVisor?

Disclosure: I work on Google Cloud.

Should be, once the rollouts complete. So once you can see N2D types in the Console for your project, I think it'll just flow naturally to GKE.

Disclosure: I also work on Google Cloud

That's correct! Once the rollout completes, you will be able to use N2D instances for GKE!

And if it doesn't, it was always possible to break GKE state into "offering" node types that were not supported. Source: me, using preemptible instances for a year before they got into GKE ;-)

Modifying instance group templates is your friend.

Yeah, this can be done for nested virtualization instances (images) too.

> Disclosure: I also work on Google Cloud

Wow, half of the comments so far start with this disclosure! Tons of Google people lurking on HN :)

Too be fair, it's just two of them answering questions, and one is both OP and the product manager.