wow dma is easy, echo ALIASES /root/etc/aliases > /etc/dma/dma.conf; echo '*:' user987⊙hdt > /root/etc/aliases

problem solved

Progress

indeed (c) teal'c

also maybe i need to make mail server tutorial

except i don't like the writings

ketas i don't get how your jails are filtered by pf running on host's bridge. i can use pf to block all traffic on host bridge, but jails happily send/receive unrestricted. they're vnet, but sounds like yours are too wtf

anyone have any ideas why?

i'm on 14.4 fwiw but can't imagine that matters

what

?

i mean jails use normal networking interfaces

so can be filtered

ya and?

ketas: sysctl -d net.link.bridge | grep pfil

kerneldove: meant to highlight you on this

k sec

these sysctls control pf on the bridge

i'musing defaults

ketas you change yours?

bridging did change a bit in 15.0

unsure why this even comes into play

what does the "il" mean in pfil?

traffic comes in

in that point i can filter

ya well i've been p clear so not sure why you're so unsure and confused

kerneldove: pfil is packet filter, which isn't strictly pf. It includes ipfw.

ahhh

so why does bridge matter?

tyvm

bdrewery happen to have a rundown of how bridge changed in 15?

Does a bridge still reset all interfaces as you add/remove VMs?

Oof, seeing lpr going away is a bit of a shock.

what's lpr?

kerneldove: printing

how will ppl print? i never print from fbsd so dunno bout it

kerneldove: CUPS, lprng, etc. lpr is simply the traditional line printer daemon.

dang, end of an era

It would be shame if printing just went away.

I think I'll try 15.1 on a laptop, for kicks.

15.2 out soon no?

.1 just dropped

argh, it used to be easy to get to the releng pages

oh .1

kerneldove: freebsd.org/releng/#schedule

wow 14.6?

kerneldove: Yeah, you tend not to get stranded quickly with FreeBSD.

:-/ Upgrade instructions for 15.1 missing. I wanna be part of the cool kids

cool kids run CURRENT.

I split the difference and just run STABLE

more of a consistent rolling release and it's not quite as exciting as CURRENT

Same here. I run STABLE in production. And CURRENT only for a specific type of development.

bweeoop

for a server I'd probably just stick with RELEASE, but I like being a bit closer to the bleeding edge on my daily driver systems like this laptop

i only upgrade majors when .2 lands

I was kinda forced over to 15 in order to get newer DRM drivers. I wasn't rushing to leave 14-STABLE

ended up upgrading my laptop's wifi card to an AX200 around the time I was upgrading though and that might've pushed me to 15 if the GPU stuff hadn't. I remember seeing that 14 was still behind on iwlwifi patches, and it's already not great on 15/16

I suppose AX200 on iwlwifi on 15/16 is more stable than my old AC 9260 on iwm on 14 though

iwlwifi has been good for me on 15.1-RELEASE. been running the RCs. there is a bug on suspend and resume that is fixed in 15/stable

but, i have a good enough workaround for 15.1-RELEASE for now. I just wrote an rc.d service to bring wlan0 down/up on suspend/resume

that issue is finally fixed? it's still an issue on this STABLE build from a couple weeks ago

really great to see fbsd taking laptops seriously now

important for us to be able to run fbsd on every computer not just server

appears to have been fixed on may 27

I just spent a ton of time getting 15.1 working well on my thinkpad x1, i wrote up everything here: sacredheartsc.com/blog/freebsd-15-on-a-laptop

Im using kde + wayland + iwlwifi

nice!

Did you enable X11 compat in wayland?

Not exactly the smoothest experience to get working, but still, the improvements are crazy from just a few years ago

robert12: i'm not sure. I didnt do anything to explicitly enable it...and i dont see any x11 clients in my session. pure wayland afaik

Cool. It's suppose to be a lot more secure if you do that.

I had to build the emacs port with native wayland support

I guess I need to double check because I'm on stable/15-n283794-7fa638d6f7cf and it seemed like the suspend/resume bug was still occurring

mason, lpr is going away? :-( Though on a desktop it is almost impossible to not have cups installed since everything depends upon cups and pulls it in. And cups shadows lpr. lpr is more solid and reliable. But then cups shadows it.

yeah, definitely hit up that bug report if you are still seeing that issue on latest 15/stable builds

though to be fair I was in the habit of always just running netif restart wlan0 every time I'd wake from suspend

hard to remember if I checked whether it was connecting all on its own after resume

tm512: if you look at the bottom of the post i linked, there is an rc.d service for suspend and resume that does exactly that :-)

when I was on CURRENT, I had the stop/start stuff in my rc.suspend/rc.resume, but when I went back to 15, I never got around to re-adding it

I'll check resume right now

(might end up timing out, I dunno)

oh, I guess it is fixed?

sweet!

the last couple weeks I might've not been giving the driver enough time to reassociate with the AP before assuming the driver was borked and needed netif restart

weird, so the other day, I could not bootstrap gammastep's configure script stuff, but now it's working. I'm not sure what changed

I guess sidestepping ports/pkg and building manually is the way to go for stuff like this where I don't like FreeBSD's default build flags

i want to append a line to a file only if the file doesn't already contain the line. is `grep -qF -- "$LINE" "$FILE" || echo "$LINE" | sudo tee -a "$FILE" > /dev/null` 100% right?

well, it's 100% worse than real config management

?!

not what i asked

and what do you use, ansible?

rtprio

puppet

ok and what does it do under the hood to make sure a file has a line in it only once?

file_

uh, it has a fuck ton of ruby

k and that's somehow better than running a command line?

your hack looks like it could work, some of the time

kerneldove, why not just `echo <newline> $(cat file) | sort | uniq | tee file` ? seems more robust

Ozymandias42 how's it more robust?

honestly haven't really thought that far but this doesn't branch. It's idempotent

ya that is pretty nice

well it modifies change time unnecessarily

hmm. true

puppet is ruby but ppl write it like perl :))

kerneldove: if you stay with shell, you should switch to an if there

would kinda prefer a command that doesn't need shell if possible?

does ``grep -qF -- "foo" /path/to/file || echo "foo" | sudo tee -a /path/to/file > /dev/null`` require shell?

yes for the pipe, redirect and the or (poor mans if)

dang, ok mind showing me how you'd rewrite it to be better?

if grep -qF -- "foo" /path/to/file; then echo "foo" | sudo tee -a /path/to/file > /dev/null; fi

mind if i ask why that's better?

because that better communicates the intend and isn't problematic to extend to more complex use cases

is it portable/standard sh?

yes of course

ok that's awesome tyvm

"poor mans if" lol. even in scripts I often do [[ cond ]] && codeA || codeB just more readable sometimes.

especially if stretched over multiple lines

also even that shorthand is trivial to extent. just: codeA => { codeA ; ... codeN ; }

and you always add a comment about how you made sure that codeA produces a succesful exit code, so that codeB doesn't get run if codeA run?

correct me if I'm wrong but that shouldn't happen, right? codeB being run is dependent on the exit code of cond not codeA

..oh yeah. I see. I am wrong. I always treated that construct as ternary operator but it isn't. it just (badly) emulates the behaviour of one

thanks! I've been using this for years but never consideres those kinds of side-effects

hence "poor mans if"

nimaje what do you think of > /dev/null vs >/dev/null like i see ppl use sometimes

the later just skips some optional whitespace, I use both depending on my mood

ok tyvm, really helps to have an expert look over stuff i'm new to

nimaje just tested your code and it doesn't actually work. it keeps appending the text

oh, yeah, I forgot the ! to invert the condition of the if

where at? i haven't done that syntax before

if ! grep …

yep there we go!

tyvm

back from 15.1 ;D

so, should I convert go pkgbase? :D

ya pkgbase is the future

Hey so 15.1 was supposed to bring the KDE option in the installer (it was postponed to 15.1)

Is that still coming?

is vuxml going to be deprecated? thought i read something about that a bit ago

fun trying a 15.0 -> 15.1 upgrade via pkgbase... lets see if it comes back from reboot

wow, big moment

5+ years in the making

Hmm I'll wait for 16 with pkgbase

forgot to reconnect to chat but that worked great

what did

are you sbr?

yes, pkgbase update

that's really great news

so you're on 15.1 now?

did you use the pkgbaseify?

yea, FreeBSD fiend 15.1-RELEASE FreeBSD 15.1-RELEASE releng/15.1-n283562-96841ea08dcf GENERIC amd64 amd64

and yes, I originally convereted with pkgbaseify

nice

run into any issues?

I think I'll take a snapshot, then run pkgbasify

not yet ;-)

I'll wait a few hours and poke thing before I do my prod machine

kerneldove: I did pkgbaseify previously on a 14.x machine and was fine, and most recently from a fresh 15.0 install. But I assume milage will vary depending on complexity of setup

ya i'm just excited about pkgbase being finally here

like i said, 5+ years

heh, woops, forgot I ran a script to restart all my jails in the background

time to destory old boot envs

ah, not surpsing but I had to pkg update -f, for pkg upgrade to pickup 15.1 pkgs, now updating pkgs

surprising*

am now on pkgbase :D

nice!

any hiccups

?

nothing

I created a full snapshot of the disk before (vserver), then ran the pkgbasify script

awesome

last piece, updated jails. doas bastille bootstrap --pkgbase 15.1-RELEASE;oas bastille upgrade -a dev-db 15.1-RELEASE

then one of my fav bastille commands, doas bastille pkg ALL upgrade -y

~20 jails now running 15.1

ok, I wrote this up, will try and do this more often. tuhat.net/o/8by3/p/freebsd-15-1

And yea, tuhat.net runs on freebsd, obvi

nice writeup sbr

rwp: Yeah, it was in the release notes.

14 hours late to the stargate reference, but I feel compelled to give ketas the thumbs up

I am on the way to BSDCAN! And so have not had a chance to read the release notes yet. Bummer that it is being removed in the middle of a release cycle in a point release though. However as I said it is almost impossible to actually use due to cups being forced in ahead of it instead.

Does pkgbase also have a replacement for freebsd-update IDS?

pkg?

pkg-check rather.

scoobybejesus: indeed

Ah thanks

i had a pkgbase question, with the repo "pkg+https://pkg.FreeBSD.org/${ABI}/base_release_${VERSION_MINOR}", how do i bump minor once to move from _0 to _1?

pkg -o VERSION_MINOR=1 upgrade doesn't seem to to do it

rtprio: You need to edit a file? wiki.freebsd.org/action/show/pkgbas…tion=show&redirect=PkgBase#Upgrades

I get the feeling that wiki is old info?

i feel like once it's on 15.1, then it goes back to the exact same file; feels awkward

my pkgbase upgrade from 15.0 to 15.1 went super smoothly with the freebsd.org/releases/15.1R/upgrading docs

really appreciate everyone's hard work on this

pkg -oABI=FreeBSD:15:$(uname -p) -oOSVERSION=1501000

there it is

i guess with _latest i could avoid this

So, someone replied to the announce list post. Is freebsd-update not able to jump to 15.1??

...for systems not installed with basepkg of course...?

Inside a jail I can't use pkg, everytime I try pkg update the process says "Waiting for another process to update repository FreeBSD-ports" but there are no other processes updating.

s2r: There's probably a correct answer, but I'll chroot into a jail sometimes from outside to do update-related things that the jail won't allow in its own context.

mason, Ok, I will try updating it from the outer realms.

s2r: In that particular case, I wonder if something left an errant lock in place.

mason, I think that person probably has a pkgbase updated system and then tried to use freebsd-update on it. It's a one-way trip. If you use pkgbase then you must continue upgrades using pkgbase. I think.

mason: i updated from 14.3 to 14.4 to 15.1 with freebsd-update just fine. i was somewhat confused, that announce@ is open to mere mortals though

Thank you both.

And ridcully - yeah, I was surprised by that too. That said, I bet it's closed now. :P

ohhh passthrough features in bhyve(8)

mason, I tried removing every .lock and sqlite cache.

s2r: Could be a kernel advisory lock without a physical presence. I'm not sure how pkg does its locking.

Might not go away until the jail comes down, or if you're very unlucky, until you reboot the host.

fnarr

mason I restarted the jail many times already and every time the same lock error.

s2r: this: forum.opnsense.org/index.php?topic=42054.0 and this: freebsd/pkg 69e28e7a0 might have some info on the lock issue - the first suggests that some bugginess results in a lock sticking around after a dns failure, the second has a command to clear an entry in sqlite that tells pkg something is still in progress

And... both hosts and all my jails are on 15.1. All the programmers and testers rock. Everyone in here rocks. We salute you!

spork_css thanks I will read them.

also just curious if "ls -la /var/db/pkg/repos/*/lock" returns anything...

hmm 15.1 kind of broke my 'build kernel packages without building world' script with flua suddenly not being found and I tracked it down to Makefile.inc1 not having /usr/libexec in PATH when create-packages-kernel calls create-kernel-packages, and now I wonder how I can run-time hot-patch this without editing the file itself

Hi, does anybody uses that native sh shell? Just curious, it's really spartan software. It even doesn't save command history properly. No sure if I should instead install bash and use sudo or doas instead of login as root into that ancient Bourne shell.

I personally use zsh - I figure if I'm already installing ports for other purposes I might as well also grab a shell

pmc: do you use sudo then?

yup

that or doas, depending on my mood when I'm setting up the system

personally I'll only skip those if I'm trying to set up a system without ports, which is usually just a learning exercise to learn more about the base system :)

I was thinking that i will try to stay with native software, but it looks more and more impossible. There ae some great commands like gpart, but this sh shell is really a bit too much.

I am perfectly happy with bin/sh

JurassCZ: sh does support history

JurassCZ: HISTFILE File used for persistent history storage. If unset ~/.sh_history will be used. If set but empty or HISTSIZE is set to 0 the shell will not load and save the history. — sh(1)

it used not to but that was quite a while ago

JurassCZ: it's a perfectly fine shell, but if you do a lot of shell or are used to fish/bash/zsh just use them

pkgs.org says vikunja is in the official repo, but cannot install/search it - am I missing something? (v15)

I think I may have solved my package build issue by adding make bootstrap-tools to the process

Ive solved my problem, its in latest and not quarterly and pkgs.org doesn't show the difference

very recently added, since it was in 14 wasnt originally concerned. a new process learned anyway

is there anyway to determine BEFORE a restart.. of an upgraded kernel that .. a module will not work?

everytime i upgrade i always seem to have to MANUALLY login to a machine to ensure net/realtek-re-kmod198 works

bhyve?

oh

then probably not

voy4g3r2: are you restarting with the new kmod version installed?

the freebsd-update is saying, it MAY not work and to rebuild

but then i am in the chicken and egg issue.. which means i get to drag the machine, from remote location, and plug keyboard/monitor to fix

bhyve

create a VM with the new kernel version, build the kmod port there

the VM has a generic kernel and no modules being loaded

(because the hv NIC driver is there already)

Hello, all. I plumb forgot the core FreeBSD tool (installed by default) that can show the current traffic (in MB/sec) across any or all interfaces. Can you please name it?

ant-x: `systat -ifstat`?

does wpa3 work for anyone

Mahiro, that's it. Thanks. I wonder where I heard of it the first time, because I was not able to find it.

elivoncoder, I have no such executable.

ant-x, It's not an executable, it's a protocol for wireless network encryption

Mahiro: It does support history, however it;s kind of buggy, often it's not stored properly and i really depend on that functionality

JurassCZ: pkg install dash or something

Oh :-) . Beg pardon. I have WiFi-enabled machine with FreeBSD at hand, but I vagualy remember I have only used wpa2.

JurassCZ, sh is a complete Bourne shell. dash is not made for humans!

^ And nor is dash native to FreeBSD.

JurassCZ, don't tell me you are not using FreeBSD's vi, nvi2.

Mahiro: Can't be used as Root. If I will use dash, I will have to use also sudo and never use root

hah. I've figured out how to use nvi, but I prefer vim :D

you're not supposed to use your system as root

wavefunction: this reminds me that we need vim-classic as the default vim :\

JurassCZ: like seriously, do not log in as root, only use su/sudo/doas when required, and do not set NOPASSWD

Mahiro, vim classic, or evi .

Mahiro, doas has problems: I have too replcate my environment for root too feel comfortable, e.g.: I use The Traditional Vi (ex-vi.sourceforge.net) as main editor, with an hierarchical system of startup scripts in my home.

-1s/too/to/g

JurassCZ: there's really no reason not to use a shell you like. install one from pkgs and use it for yourself.

Over in #vi, there are some real vi masters!

and it's good hygiene, generally, to use sudo or doas instead of spending a lot of time in a root shell.

What if one want to be feel unique and BSD-ish?

AND

ant-x (I count that as a "shell you like")

ant-x: then you `chsh -s /usr/bin/tcsh`

if you need to be root, there's nothing stopping you from invoking, e.g. tcsh -l or bash --login or fish --login once you have su'd to root.

kona, no, I mean my .profile and .shrc &c.

Is it not bad hygiene to do ``doas sh'' ?

well. for zsh you can put all that zsh stuff in your zshrc

was it 12 when root shell was changed to sh?

and keep the .profile and .shrc to the lowest common denominator

doas will use the environment of the target user, not of the user that invoked doas .

Mahiro, what was it before sh ?

csh of course

doas will but all the shell niceties you have in whatever your user shell are substituted first before doas is even run.

Ah! Not Bourne-compatible?

Mahiro: The kmod is already installed and on 15.0 and when i did freebsd-upgrade -r 15.0-RELEASE it informed me these MAY not work

so i am basically trying to get AHEAD of it before a restart.. because if they won't work.. then i gotta move the machine :()

voy4g3r2: and this is why you need a VM to build it for the new kernel if it is not available from repos

kona, I lose my vi mode, set by ``set -o vi'' in my .shrc .

hrm.. so even though the port is installed now.. it will NOT work.. with an upgrade

ant-x: is that a bourne shell feature?

^ Unless I do that in root's shrc, too some way or other (symlinking &c).

as poudriere does not recommend using jails newer than host

kona, a POSIX shell feature, available in FreeBSD's sh .

(as well as in bash)

voy4g3r2: kernel modules are linked against a given kernel; kernel ABI is not guaranteed to be stable

ok well. idk why you would lose your vi mode then if you are issuing doas commands from your user shell.

but i probably misunderstand what you are trying to say

yup.. so i guess once i do the restart i gotta manually install, thanks

i may hold off for a little

kona: sudo sh will run sh with your current environment including $HOME; doas will $HOME=user-you-are-switching-to

doas sh will not use your existing rc file

you will need a duplicate in that target user's home

i think i can count on one finger the number of times i have needed root to do something in my $HOME

it is not about doing something in $HOME

kona, using vi mode all thorough while on the shell.]

it is about sh loading $HOME/.shrc

Mahiro, yes, I think sudo is differnt from doas in that way.

'it depends'

As for doas, <github.com/stas-badzi/doas-keepenv/blob/main/doas-keepenv> .

at least on my system sudo zsh definitely loads target user's .zshrc

i think i can also count on one finger the number of times i have wanted to have root use my .shrc for anything, at least after i understood why having root have a mostly pristine environment is a Good Idea.

Mahiro, yes, indeed.

but. ok.

so, use sudo.

OR

sudo vs doas holy wars

kona, e.g. I want to edit a file as root. I issue: `e /usr/local/etc/torrc` , where e is my script lanching my self-compiled vi .

i guess, if you must, su to root and then invoke the shell you want, after copying your rc files to root's home dir.

scripts*

Mahiro, why not sudo for weird scripts?

ant: put your self compiled vi in /usr/local/bin and your 'e' script as well.

or maybe a site-local directory

and put that in your paths (or in the system default paths)

better yet, make a local port

build with poudriere!

kona, I will also need to provide my ~/ex and .shrc.

Oh! Building it as-is was hard enought (for me).

ln -s will do that, if you absolutely need them automatically sync'd

Indeed.

sometimes you just have to respect religion

But linking root's stuff from a user's stuff is ... the wrong direction of dependency.

everyone is on about the poudriere kool-aid

then put it in /usr/local/etc and have users (including root) that need it, link to it there

I have also modified The Traditional vi to not use sbrk(2) for memory management!

kona, right: I keep that possibility in mind.

but remember it's.... "your" exrc. so maybe putting it in a sitewide share isn't quite right, either.

i'm not following what problem you're trying to solve

like, if five other users have an account on the machine, maybe they don't need access to your exrc.

kona, I agree, but that script is not secret, anyway.

rtprio, invoking doas with the source user's environment.

doesn't doas have some sort of keepenv/setenv

DOAS.CONF(5)/keepenv might help -- will test.

^ also considered dangerous.

But I'll ingnore the warning.

why does he need his environment in his root shell? it's the root shell

they want traditional ex/vi (heirloom ex/vi) with their shrc and exrc everywhere they need to edit stuff.

rtprio, e.g. to use my editing setup, with my $EDITOR=e, referring to my wrapper around my self-compiled vi editor.

ok, aside from why does it need to be wrapped

It is no where near impossible. There are many options to try.

just checkout a copy of your dotfiles into your root shell

but why are you lingering in root shell enough to need your dotfiles

rtprio, there may reasons to have a wrapper around vi. One of them -- to use extention-specfific configuratin of the editor.

rtprio, re: checkout my dotfiles -- right, it is yet another solution.

rtprio, I am just editing files no accessible from my normal user.

Not lingering, no.

Not to mention that sudo solves it automatically and effortlessly :-)

i guess this is just not a problem i've ever had using my normal day to day editors and sudo or doas

Sounds plausible.

I usually copy the configuration file, make another backup copy, edit locally, then install(1) over to the old location

Thanks to all for your advice!

because paranoia

Mahiro: it's not paranoia if the filesystem is actually after you

Mahiro, that's too much of monkey business unless the file is very securyt-critical!

Hmmm, and I could automate the three-stage-operation with a shell script :-)

rtprio: look, every time I have a long, long build running and need to hibernate my PC I am like, can I just do that? what if hibernating corrupts my build?

1) doas cp 2) doas chown 3) e #edit locally 4) doas install 5) rm #local copy

so I have a lot of paranoia/anxiety

oh, i gave up shutting down my pc long ago

ant-x: you forgot backing up persimmons and uid/gid

paranxiety

you realize you editor makes a copy, right?

Will add them in the next commit :-)

you would really want those persimmons

If you are parsimonous.

rtprio: electricity costs and also noise

rtprio, Editor makes a copy? In the FS? No. Do you mean ex-preserve ?

I must go to bed now, to dream about my new imagined smooth and steramlined edit-as-root-via-doas workflow.

Good bye to all.

i don't know what ex-preserve is

hi, im trying to upgrade from 15.0 to 15.1 and i need to do the bootloader step. but im not finding the ??bootloader location?? here is my output termbin.com/x708

im trying to do this step: freebsd.org/releases/15.1R/installation/#upgrade-loader-uefi

suggestions ?

it's one of the ones matching \efi\freebsd\loader.efi

riight

that's the longest efi list i've ever seen

lol

yeah i dont know why it has so many crap there

if you reinstall / change OS, that usually needs to be cleaned up by hand

remake that partition

during install

Mahiro: Your VM option, for the upgrade question i had and hte network card, you would recommend making a bhyve of 15.1 and have the module compiled on that and MOVE it to the host? i am a little slow :)

rtprio: ex-preserve man.freebsd.org/cgi/man.cgi?query=ex&sektion=1&manpath=4.4BSD+Lite2

doesn't really explain it much, there's a paper though that does. it's an ancient vi/ex feature

i was born after 1975, i don't use ex

I think I used ex once

I've used ed more than I've used ex

if you used vi in the 80s or 90s, exrecover and expreserve were features of vi as well.

mostly because vi was a feature of ex

Hrm, the 15.1 install media fails to thrive on my T420. Doesn't configure the network, doesn't see disks after it fails to use the network and restarts, which isn't surprising. A bit disappointing.

mason: 14.3 installer does the same.

I retract above. I used 15.0 in my last install.

And the 15.0 installer had the same behaviour as described by mason