W dniu 22.05.2025 o 23:06, Patrick M. Hausen pisze:
Hi all,

Am 22.05.2025 um 22:15 schrieb void <v...@f-m.fm>:
I think, from what other list members have written also, that many did as I 
did:- looked at the virtualization part of the handbook, found all what
was required there to get started, and thats it. They would probably not
(as I didn't) see the need to look at the advanced networking section if they
were only using a bridge with bhyve or similar.
I have come to realise that there are two sides to this issue, both equally 
valid.

How to configure and use if_bridge(4) correctly was documented from day one
or very shortly thereafter.

But still - for reasons I do not quite understand - more than one 
platform/wrapper
development ignored that documentation.

FreeNAS/TrueNAS surely did and from your posts I read that more jail/VM 
orchestration
tools also "do it wrong".

So I agree - we cannot place the burden on the users with a:

"The documentation was on display in the bottom of a locked filing cabinet 
stuck in a
        disused lavatory with a sign on the door saying 'Beware of the 
Leopard.'"

Which I am prone to do occasionally. Sorry about that.

The technical discussion is simple. An IP address on a bridge member never
was supported and never will be.

The change that is currently discussed simply prohibits a setup that never
was supported in the first place with the good intention to save people from
foot shooting.

I support your suggestion to *somehow* make some more noise about that.

I have been screaming at walls for years about the broken setup of bridges
in TrueNAS on the iX forum to no avail. Tickets in JIRA closed without action 
...
Stuff like that.

Kind regards,
Patrick

I think enough noise has been made. Regardless of its cause and intentions, this kind of noise isn't good for the FreeBSD community. I hadn’t planned to post further in this thread, but I’ve noticed the debate rising again like a phoenix from the ashes. Maybe it’s best to let it burn itself out.

Now is the time to focus on more productive efforts - improving tools like vm-bhyve, helping users with migration paths, and making sure the documentation is consistent and accurately reflects the current state of affairs.

Most importantly, there is a need to show the broader community (or at least not hide) that alternatives are available, such as ng_bridge(4), which allows for quickly setting up a network bridge for fast spinning up a bhyve VM for testing or running a VNET jail for short-term use.

Cheers
Marek


Reply via email to