> We have the per-cluster built-in migration now, which is called on every
> start-up. It can easily shuffle files around as needed, and issue API
> calls to update the cluster's BootImages.
I had forgotten about that. Indeed, that's the better solution.
> We have the per-cluster built-in migration now, which is called on every
> start-up. It can easily shuffle files around as needed, and issue API
> calls to update the cluster's BootImages.
I had forgotten about that. Indeed, that's the better solution.