Bhyve Hypervisor Building

Phishfry

Phishfry

I am wondering how people build their hypervisor. I am thinking of the Xen hypervisor as an example. A stripped down Linux install.
Building with NanoBSD taught me about using FreeBSD for a purpose built appliance.
My thoughts are that a Bhyve Hypervisor could use the same approach with stripping out un-needed components.
I thought about going as far as ripping out all the unneeded drivers and just passing the devices in loader.conf.
Keeping just the motherboard ethernet and other bare bones needs.

Do you build a custom hypervisor install or just use the stock install?
 
I am still new to Bhyve and I have re/built my hypervisor over and over. Still trying to find that sweet spot.
I have went from one extreme of installing a full system including Xorg to a bare install.
That was why I was wondering how people approach it. Bare minimum or treat it like a regular system.
The low Bhyve overhead really makes either approach usable.
Right now I am running a second purpose '10G router' on the hypervisor with 2 Chelsios serving up dnsmasq/gateway for two other boxes.
I am not sure if I want the 'router' on the hypervisor or move it to its own VM.
I am trying to get a grips on my 10Gbe speeds on the hypervisor before testing in a VM.
My initial thought were it would be quicker on the Hypervisor but my NVMe tests showed little loss on a VM.
So I need to test networking next.
Other thought include crashes of 'router' might bring down my Bhyve.
But realistically how often does networking bring down the machine.
 
I am trying to learn Bhyve without the assistance of helper applications. I wrote a rc.daemon script to start my VMs on bootup.
 
You must log in or register to reply here.
Back
Top