@rmi These fucking idiots should be punished by having to work on a 1.66GHz single-core (with SMT to reveal race-conditions), 1GB RAM, a 160GB 4200rpm HDD and a 56k modem for a year.

Today Google bricked my Chromebook by force-installing a hidden extension that trains a machine vision model on the contents of my screen without my consent, making the whole machine too hot to touch, and I am once again begging someone, anyone, to remove Chrome from Google by any means necessary.

https://research.google/blog/screenai-a-visual-language-model-for-ui-and-visually-situated-language-understanding/

the single greatest Signal message ever sent

@dexter @gumnos @dvl @antranigv Most likely he lost a required route be removing the next hop route when deconfiguring interfaces. You have to restart the routing service to restore the (default) route.

@dvl You probably missed the /etc/rc.d/routing script to bring back the default route. `service netif restart; service routing restart` to be run in nohup/tmux/screen/dtach or similar.

Thanks to @crest! He figured out an Issue I had on my FreeBSD host while I tried to `jexec` into a jail, was because of a uninstall security patch! :) The #FreeBSD #Jails calls are great!.

Outlook wants to stop me writing "python script from hell" in work emails 🙃

@cks Sure you could do a a stateful hack to record the nexthop for replies or you could do it a lot cleaner: create a FIB (aka kernel routing table) and set the FIB per interface.

@henrik @ChrisO_wiki That the life expectancy for fresh Russian troops on the front can be measured in hours instead of days or weeks. A significant number of them will be dead or crippled before the normal incubation time of the waterborne infections.

@doragasu And that's before you get into alternative modes like Thunderbolt, DisplayPort, HDMI etc. *sigh*

@GossiTheDog Hard to believe, but for once Microsoft support staff was helpful 😜​

@dvl: I finally documented why and how to run your own local FreeBSD package cache with Varnish: https://blog.rlwinm.de/why-and-how-to-run-your-own-freebsd-package-cache-3wbg

@dexter @peter @pitrh Depends on how much crap you're running at home. In my case (summer) room temperature would become the limiting factor before the power bill.

Allan Jude channelling Rachel Maddow

@Tubsta @dexter @kp @tuxpowered @antranigv I did document that you have to disable LRO/TSO for if_bridge(4) to work correctly (in almost all realistic configurations).

@dexter Afaik FreeBSD's PF fork doesn't support NAT64 at all and FreeBSD also lacks support divert(4) for IPv6. The only way to do it is with a pseudo-interface e.g. tayga on a tun devices and two static routes.

@Tubsta @dexter @kp @tuxpowered @antranigv The network setup required is a bit more complex, but I shared my working configurations as examples because their needs just happened to be exactly what I tried for a routed IPv6-only bhyve lab setup.

They require SLAAC + DHCPv6 PD on the upstream interface to obtain an the router's address which is a special case (v6 forward + SLAAC) that's fiddly to configure.

The goal is to automatically renumber the downstream interfaces when the ISP prefix changes (probably at least daily) and provide global unicast addresses via SLAAC on downstream interfaces.

I did this using dhcpcd with a small custom hook script to reload the FreeBSD base system rtadvd via its control socket when the delegated prefixes change.

I also used a ULA address (e.g. fd00::53/128) on a dedicated loopback interface to have a stable IPv6 unicast address for the DNS resolver to share that isn't link local and doesn't have to be renumbered. Since the FreeBSD router is on path for all its directly connected downstream interface no extra routes are required.

@ska @antranigv @dexter @bsdcan Without a frontend a reliable maintainable s6-rc setup has too steep of a learning curve to fit it into a short one-off demo. Which doesn't change the fact that runit is only an (unmaintained) process supervisor.