Using New Bridges of FreeBSD 15

FreeBSD 15 just rewired its “internet switch” — and the nerds are fighting about it

TLDR: FreeBSD 15 added a much cleaner way to handle virtual network switching, aiming to make setups simpler and faster. The community is split between people cheering the progress, power users calling old methods slow and clumsy, and confused admins begging for better guides before their brains fully melt.

FreeBSD 15 just rolled out a shiny new way to do “bridging” — basically turning your computer into a fancy virtual network switch — and the community instantly split into camps: the wow, finally crowd vs. the wait, what just broke? crew. The new setup is cleaner and more like real network hardware, letting people use a single virtual switch instead of a spaghetti bowl of settings, and one commenter practically stood up to clap, saying they “welcome and applaud any progress on the BSD front.” For longtime FreeBSD fans, this feels like the project finally getting a modern makeover.

But of course, it wouldn’t be a tech thread without drama. One user storms in asking why there’s suddenly a wave of FreeBSD posts, basically implying: who ordered all this BSD content? Another confesses that all the layers of tools, jails (isolated mini-systems), and virtual networks have “knotted” their brain, throwing shade at devs who assume everything is “obvious to anyone smart.” Translation: documentation better level up, or people are out. Then there’s the performance flexer who drops a hot take that these virtual bridges are inefficient and brags they ditched them for a faster setup that finally maxed out a 10-gigabit connection. Meanwhile, someone else is side‑eyeing the author’s long list of disabled network features like, “Why are you turning all this off?” In classic FreeBSD fashion, it’s half engineering talk, half existential crisis, and 100% nerd soap opera.

Key Points

  • FreeBSD 15 introduces a new bridge implementation with native VLAN support, changing how VLANs are handled at the bridge level.
  • Layer 3 addresses on bridge member interfaces are soft‑deprecated and controlled by the net.link.bridge.member_ifaddrs sysctl, which is slated for removal in FreeBSD 16.0-RELEASE.
  • Previously, administrators often created a separate bridge and VLAN interface per VLAN, resulting in complex and verbose configurations.
  • The new model allows a single bridge to handle multiple VLANs, with tagged and/or untagged VLANs configured per member using options like vlanfilter and tagged in ifconfig.
  • VNET jail setups that relied on the unofficial jib script and custom MAC-stability logic are affected, but stable MAC generation is now provided natively for epair(4) via net.link.epair.ether_gen_addr and ether_gen_addr(9).

Hottest takes

"Why sudden surge of FreeBSD-related posts?" — j16sdiz
"I have managed to completely knot my brain over the abstractions" — ggm
"Bhyve bridges are inefficient… Switching to SR-IOV… I saturated the 10 GbE link" — shashasha2
Made with <3 by @siedrix and @shesho from CDMX. Powered by Forge&Hive.