I have one with that chipset too.

Thanks for all the wifi work.

so, I was just experimenting with the `zfs jail' feature. is the intended method for exposing a jailed dataset to a jail via `jail.conf' doing something like `exec.created += "/sbin/zfs jail ${name} data/set";'? This seems to work without trouble, but I couldn't really find any documentation on any standard way to do this.

oh looks like it is controversial

so there are others which think pkgbase is a steaming piece of shit too? :P

it looks like the apach24 userdir module is not working in 15.0

apache24*

specialbomb: I think all that's needed is "you need to set jailed=on for dataset" cgit.freebsd.org/src/commit/?id=e0dfe185cbca

Looking forward to trying that one out.

oh wsky is here

do you pronounce your nick w-sky or wisky?

it's short for wintersky

oh you already told me that

anyone can confirm my bug?

I still read your nick as "wisky"

it's ok

no im not an alcoholic :P

although beer tastes good

sounds like you need a drink polarian

rtj: already had a wheat beer earlier

Nice, just coffee here.

wsky: cant help with apache though sorry, and sorry for derailing the convo

although I would ask why apache?

nginx is faster and more efficient

No 15 installs here yet.

is there somewhere that shows patch releases? like a list of them?

The announce mailing list.

ok i got it it was just disabled in httpd-userdir.conf

Macer: check the website as well will tell you

freebsd.org

the homepage has a list of security patches and erratas

rtj: yeah, that part is pretty obvious imo, I was more having trouble with "okay, so now how to I automate jails having datasets exposed to them and mounted"

that?

polarian: oh it does? hm.

Macer: and yes that mailing list is the correct one

whenever you get a security advistory (SA) which are also showed on freebsd.org there is an update

however they dont always announce the erratas iirc

specialbomb: Ok had no idea about context. Ty polarian, nice to see you today.

that commit you sent was actually very useful, ty

> The Reg FOSS desk spoke to an OpenBSD developer recently, who thought the change was rash and profoundly counter to established BSD practice. ~ theregister.com/2025/12/05/freebsd_15

I agree!

FreeLinux not FreeBSD!

It's one of my favorite things how it's seprate. I undersand things change. I will greatly miss it though.

polarian: thanks.

rtj: im protesting, I will compile from source insteasd

and if they remove that, I guess I will need to patch it back in :)

I truly havent been around freebsd long enough to give any opinion on pkgbase. I think definitely benefit from not needing to do delta based updates, but I also think there should be special management for base packages instead of treating them like 3rd party. I hope thats what is decided as time goes on

at some point i need to sort out how to update base with pkg too. i guess that's going to be a thing.

although i was more a fan of keeping pkg for packages and base with freebsd-update

I can't really say anything because I've yet to try it. I'm just a creature of habit though.

vortexx: you are a OpenBSD + FreeBSD user, thoughts on pkgbase

also.. i see there is kmods_quarterly_${VERSION_MINOR} ... is there a latest version of that where quarterly can be changed to latest?

similar to pkg+https://pkg.FreeBSD.org/${ABI}/quarterly being changed to latest

next freebsd will be adopting rust

pkgbase is a big mistake, it disappoints me to see some of the major freebsd devs so passionate about the idea of turning freebsd into Linux

speak of metapackages like its innovative

Linux been doing this shit for decades and it always ends up with the same issue, dependency hell!

probably not so much for the kernel and base utilities

So... the parts are distributed as pieces but that doesn't mean the system isn't cohesive or piecemeal like Linux

thats what im thinking, like if the base system is fully maintained by freebsd as it always has been, I really dont see any issue with it in the regards of "dependency issues"

i'm just more a fan of freebsd-update and pkg being separate because then freebsd-update is a bit more static and independent of anything else

Macer: and so then, what im saying is that there should be very clear and separate management of base packages and third party packages

i'd be worried about pkg somehow breaking pkg or breaking how it can update base somehow... something that you'd probably not run into just keeping freebsd-update

i mean i'm sure there will be tons of testing to prevent that .. but freebsd-update was tested and true :)

id say theres always fear in a new tool. do you trust who makes the tool, is the more important factor here in my opinion

i mean i guess i can understand why they want to do it because then you can keep everything in one place instead of maintaining two things to do something one can do

kenrap: exactly...

took the benefits of BSD and ran them through a shreadder, and then doused them in petrol and fucking burnt them

like wtf is the point of pkgbase

"oh look I can remove this core library" woah

you know what you can also do? remove it from the source and compile your own custom kernel

you know, hacking the OS?

the point is to make one ring to control them all :)

Macer: eh... bad comparison. considering how LOTR goes in that regard :D

going back to talking about linux. i mean i don't really have many issues in something like debian which just uses apt to do essentially the same thing. or at least i've never experienced a wtf moment doing it

but i did like how freebsd kept it separate so i know... this is for this.. and that is for that

specialbomb: you mean mount doom went boom? :p

freebsd is mordor, the ring is pkgbase dropping into mount doom

:)

Macer: I might be young, but I have played with both OS enough to know that the Linux way is inferior

the control you have from the separation of base from ports is amazing

my question would have been when is fbsd moving to systemd :)

the fact your base system is one complete component is a great feature

Macer: that would be fork worthy

tbh pkgbase is fork worthy

I get that delta updates suck, but this is not the solution

polarian: im curious since you dont think this is a viable solution, what WOULD be a viable solution to this?

what would be ideal, in your perspective?

specialbomb: leaving shit how it is :)

maybe rewriting freebsd-update to be more efficient

freebsd-rustdate exists which is meant to take sub-minute to update a system to a new release (from what I hear at eurobsdcon) instead of 10 mins

like if you must pkg it, cant it be one big package

one complete redistributable

one complete system

hello I need help to get my server back online

freebsd server back online

sixpiece what happened?

cat tripped over your cables?

it was the squirels

ah right

Alvin?

Alvin was a chipmunk.

the floor is littered with woodchips and apple cores and hulls of acorns

theres a chattering sound... because they were squirrels! Real squirrels!

Macer you're right, same family though

good to know we are using "it works on linux, so we should do the same" as justification

BSD predates Linux

Linux has only been taken seriously in the last decade or two

hmmm... I dont necessarily agree with that. It's less that "it works on linux", its more that "it works, and it just happens to be on linux distros". its not a bad design decision when handled properly

i don't mind having 2 different utilities at all, i would never perform pkgbase + pkg actions together at the same time either so don't see much benefit in having a single command

oh, I do agree with that. I was disagreeing with polarian :p

everything roger said there was very sensible

s2r basically I was trying to upgrade to freebsd 15.0 and I was extremely tired also doing the installation on my laptop as well and I made a change to one of the files absent mindedly and then it might have been ok but I was scared I did some damage so I used a backup and a snapshot and reversed the change then ever since the internet hasn't been working correctly and lots of issues getting it

connected

I still want to perform the update and that's the situation but I'm like right now not even up and running at all , it was simple stuff that made it pingable and internet out but I still can't connect by ssh I have to login to my cloud vps provider and go in via their novnc connection

sixpiece so you went to a previous snapshot? Did you create a boot environment prior to pkgbase upgrade?

I had a lot of firewall rules because I did sendmail on the server to send out emails and there's a open dkim that required some rules and I sort of commented a few out very quickly hoping to get the firewall running so I have some level of protection maybe I should stop it completely I am not sure for now what to do

specialbomb: disagree with me all you like

I want my base system

it does an automatic 2 times per day backups and I just went on a backup that was 10 hours old

no metapackage

no ports updates with base updates

just my base

as a single entity

figuring that it was largely the weekend and not much happened in terms of development

polarian: no disrespect at all, I simply just disagree.

polarian can't you set all that on the repos to active: no?

it's automatically done by my cloud service provider I just requested the backup then it gave a snapshot then I used the snapshot sorry I couldn't answer your question further than that

specialbomb: ik :)

sixpiece and from that backup you have missing files?

<3

s2r: ?

I'm not sure from the backup I am not able to ssh into the connection for some reason ipfw initially was on that blocked all traffic now I turned it off I was using pf and have some rules I turned that on

polarian if you don;t want to upgrade everything set it to no on the /usr/local/etc/pkg/repo/FreeBSD-base.conf

I could try turning it off and then seeing if that works and then doing the upgrade

sixpiece where is the server located, have you got access to the console? What services are running?

I am running apache, mysql, different things like icecast and stuff like that even an irc server , php and a few things that were failed experiments like named server, opendkim is a service, sendmail, etc..

sixpiece I ran into some issues while doing pkgbase upgrade mostly related to fingerprints on the .conf file. I couldn't follow step by step the upgrade process from the handbook.

I turned off the firewall and I still can't get into the ssh

sixpiece Is mysql connected to thte internet or listens localhost only?

yes I used granite the ai tool as well as the manual and that helped a lot

normally I just use the manual but started with these ai tools

I can try

mysql failed maybe it's not on

sixpiece I don't remember exactly but couldn't you spawn a shell from mysql if you have the appropiate rights?

it's on but I cannot access it oh right I was sshing into mysql

sixpiece from which version were you upgrading?

I was trying to be conversative but I guess that's my thing I need to do to be balanced centrist

14.3

for some reason maybe it's the snapshot that brought me back in time

maybe I screwed up

sixpiece Have you created a boot environment prior to the upgrade?

s2r: like I will pkgbasify my system lol

Here is a question. For those that really hate pkgbase, what's your "Plan B" for when FreeBSD meets its inevitable demise of pkgbase? :)

which kind of sucks cause I lost months of work by this stupid mistake I was tired

makeworld, makekernel, installkernel (reboot), installworld, etcupdate (reboot)

I am not sure what you mean boot environment?

polarian pkgbase lets you make your system reproducible like the requirements.txt in python

and if thats removed

idk

sixpiece: do you use zfs on root?

I have one from 24 hours ago but I think it was still lost at that time

you should be aware of boot environments if so

sixpiece something like, bectl create 15.0-RELEASE

no it's a cloud instance that they make for me with kvm

in any case I can always fuck off to openbsd for good :P

sixpiece if it's a cloud instance, can't you access the console?

but then wheres the fun in that, I wouldnt be able to moan about pkgbase then

I might get lucky and hopefully the backup is from when I did the work 24 hours ago ...

it's from over a year ago

sixpiece ah ok. so no zfs.

I'll be really lucky but I think it's possible

sixpiece So right now you are trying to revert everything to a known state prior to the failed upgrade, right?

wow just got very lucky I think

it wasn't really a failed upgrade as much as I thought it was but I did make a minor mistake or two deleting configuration I shouldn't have when I was trying to make a merge conflict work

wow it's back I am going to do the upgrade now if I have any issues I will let you know thank you very much sorry it wasn't supposed to be this easy lol

good thing I checked it before that backup got overridden

zip: just submit a PR

kenrap: i have no plan b as despite the hate towards pkgbase you can still manually compile the system as well as ports

kenrap: that isnt going any where

so i guess my plan b is compile form src

cpet: true, src building is the safest option.

sixpiece: you can create be's

kenrap: 15.0 is a .0 most if not .0's are buggy

i have yet to see a .0 that doesnt have some issues

but i dont do any weird things so my brakage is different than yours

It really seems like this .0 has been having a lot more issues than other prior .0 releases though. I think kevans made a good point about the lack of testing in 15.0.

you have not used 5.0

5.x was horrible heh

Got another question, what happens if src building because pkgbasified? :)

*becomes

the whole point in finally releasing pkgbase is to get more use out of it

due to the nature of current i dont think that would be an issue

in theory openbsd recomends you sysupgrade to a snapshot rather build from src anyways

sixpiece are you upgrading right now?

kenrap: with ccache it takes me 28 minutes to compile kernel and world, not everyone has a system like mine so

a lot of people wont bother to compile

s2r yes I am upgrading right now

yeah... its 2025 and i have a family, i'm not compiling my operating system lol

Yeah, pkgbase really makes -current snapshot upgrades accessible.

kenrap: i think the real issue is when a new release comes in people rush to upgrade

kenrap: some goes for seueity issues like russia really want little timmys gaming system and collection of furbies in bmp format

kenrap: the humour i get is when people use something that litteirly says tech release

which meaning its still not production ready

then people like yourself bitch about it :D

I don't bitch because I use -current :)

the only reason i use current if if a specific mission ciritical piece of HW is supported in it

other then that i dont see a point seeing how this desktop is my daily driver

kenrap: the ML's are interesting too such as a freebsd commiter complaiing about current being broken for 9 hours

heh

I just learn to make BEs and snapshots to be able to revert to an older build and stay there until the waves settle down for a newer build. :)

I don't follow the MLs much either.

I should look at the mailing lists

JetpackJackson: it gets pretty bad when you remind a commiter about jenkins

but then again my first port was fceu

not searching first I manage to port somethingthat was already ported :P

that was a funny day

kenrap: i just boot kernel.old

i have yet to use be

s

but then again i dont fiddle with my system as much as I should

BEs are easy to make and use. I suggest using a date command for using timestamps to name them too.

if you need to have a lot of BE's think the issue is the user not the OS

heh

cpet: lol

I don't make a lot of them, just a couple at most. I just don't think "backup${N}" is very descriptive and doesn't scale with scripting.

i name my backups backup1 and server-backup

how more descriptive do you need them to be

back-how-i-screwed-up-my-config-12-Dec-25-022321

?

I'm not trying to argue with you. I thought it was a nice suggestion. Name your BEs however you want.

backup-kenrap-complains-alot-about-fbsd-datestamp

cpet-now-goes-into-my-ignore-list-kthxbye

i dont need to know if you ignore me or not

you don't have to act like jerk either :)

aww cute thought i was on your ignore list

Where can I check where the command named dialog has gone?

Use "which dialog"

cpet: backup-preconfigchange-{date}

V_PauAmma_V I can't find it. I did a find / -name dialog -print and the only place is found is inside a 14.3 jail. I'm running 15.0

tbf i don't think 15.0 was any less tested than any other release, people just don't take advantage of the time we offer for testing this stuff in general for any release

(still in scrollback)

> if you need to have a lot of BE's think the issue is the user not the OS

everyone has different risk tolerance

cpet: ideally you name these things with some idea of *why* you're creating one in the first place, but the exact sense kind of depends on how you model these things

i mean, you can create a BE and operate on that (let's call it 15.x), or create a BE as a backup (let's call it pre-15.x); in any event, they offer more value than just rollback, whether the majority of people take advantage of that or not

s2r, did you install your 15.0 using the pkg installer?

V_PauAmma_V upgraded via pkgbase

just letting you know thank you freebsd 15 seems like it was successfully upgrading on the last leg with post installation tasks

sixpiece: awesome! :)

sixpiece can you check if you have the dialog program? which dialog

V_PauAmma_V I have checked in another upgraded installation and is missing there too.

Hmm. Not in the tree anymore.

Ah. From freebsd.org/releases/15.0R/relnotes/#userland-programs: "The bsdconfig(8) and bsdinstall(8) utilities now use bsddialog(1) instead of GNU dialog."

So I suspect dialog was shown the door in the process.

V_PauAmma_V so any pkg depending on it should fix its dependency. Thanks!

V_PauAmma_V how can I check if a program is on the tree?

s2r command not found

sorry s2r for long delays

s2r, either of 2 ways: if you installed sources on your host or VM, "find /usr/src -name dialog -print" (or whichever name you want to know about). If you didn't, point your browser to cgit.freebsd.org/src, switch to the branch for your FreeBSD install (so releng/15.0 in your case), click on "tree" in the bar near the top, then look into the appropriate directory (bin, sbin, usr.bin, or

usr.sbin). If you don't find it there, check the contrib directory as well.

tired and speaking to a friend on the phone multitasking

thank you again good night

s2r

kevans: i was giving kenrap a bad time

hello s2r I am interested to learn the thing you were speaking of and I always appreciate the great support of this channel good night thank you

every thing is miraculously working as good as I need it for now and my systems are on freebsd 15

later

Plasma Wayland is a pretty awesome experience

i like wayland. i do not like X.org. finally i can make my system to be free of any X.org traces at all. port's allows this. it was little bit harder few years ago. but qith qt6 it is finally possible as they allows building without X.

there are some quirks though, which can be sorted/patched.

angry_vincent: it seems like you're almost there given the current limitations with the USE_XORG knob.

polarian: I haven't looked at it, I'm on 14.3 for now. I do recall it's replacing freebsd-update which saddens me

hello, i have downloaded the freebsd iso (the big one) and installed it in virtualbox. After the installation, no gcc. I tried to install it via pkg, but sudo is not installed, and su refuses to succeed, and pkg is not there too

is there something to do during installation to have sudo and pkg already installed ?

›doas‹ instead of ›sudo‹ may be the way

mdo(1) and mac_do(4) is another option and native to FreeBSD.

vtorri: `su -` then use the password for the user `root` (not your password)

ridcully : result : "su: Sorry"

trying the other ways

SomeVisitor: doas is not found

doas needs to be installed. same as sudo

can't you remeber your root password?

mdo : not enough permission

there is no pkg bin in /usr/bin

its in /usr/sbin

and you need root permissions to install anything

i know my su password

but i can't even get to write my password

su fails before

are you member of wheel?

I wonder if it's a pam issue?

tsoome_ what ?

use `id` with your current user, and check if it's in `wheel`

from su(1): In particular, by default only users in the “wheel” group can switch to UID 0 (“root”).

since you run in a vm: switch to another terminal and log in as user root there

ridcully everything in vtorri

You might want to download the memstick or iso image and boot into it, so you can chroot into your install and reset your root password

(Oh wait you already got one)

i'll reinstall again

booting into single mode is usually enough to change root passworx

it's not very long

Oh yeah, do what mzar said.

i'm reinstalling freebsd in VB

at the step : "Distribution sets" or "Tech preview" choice, what should I use ?

kenrap: it is an artificial thing, re USE_XORG. it is automatically appends RUN_DEPENDS for a given entry in USE_XORG ( which is in itself failproof approach - better install more than is needed ). However for purists it is an obstacle. There are complex or lesser understood dependencies. It can be treated specifically depending on what is the goal. I do not like excessive, unnessecary dependencies. But I trained by many years of

Gentoo. FreeBSD while allowing many things to be optional is little bit more conservative in configure options.

angry_vincent: gotcha

32.1T / 69.2T copied at 390M/s, 46.40% done, 1 days 03:42:38 to go so i guess about 2.5 days for me to expand my vdev

i guess later i'll find out if that is vdev dependent

i don't think zfs does any type of reflow if you add a new empty vdev to it

this weekend i'm going to dismantle my old 4TB x 12 pool after moving it to the pool i'm expanding. should be fun with zfs copying

it's nice to see that you are progressing Macer

Macer: you sure are having a hell of a time. makes me feel like I should actually be finding this failing drive in my disk shelf :P

Macer: if you are good with zfs send/receive please give me some working examples for my notes. I run into a lot of times I think it could be useful, but the documentation online is always way over my head. If I just "see" it in action though, I might be able to pick up on it

SponiX: what sort of issue do you have with send/recv? they are somewhat… very simple tools?

one produces a byte stream, the other consumes

like tar c -f - | tar x -f -

Remilia: can you give me a working example of doing it from machine1 to machine2 over a network? I'd like to be able to do it between my virtual machines, and also between my vms and a bare metal box too

SponiX: `zfs send -LRc zroot | mbuffer -q -s 128k -m 32M | ssh target-host 'mbuffer -s 128k -m 32M | zfs recv -Fdu targetpool'`

err

zroot@snapshot*

that's a full replication stream

incremental is adding -i with the previous snapshot to send arguments

so something like `zfs send -LRci zroot@backup-202511 zroot@backup202512`

but the previous snapshot has to exist on the target pool

Yeah, you say that is simple. But that still boggles my mind LOL. I will check back in with you the next time I have a situation that needs this ;)

you do not have to use ssh of course, can just netcat if you trust your network

you also do not need mbuffer, I use to to soften the inter-stream blows

I'm up for using any method that works. netcat might be a bit faster than SSH though

I use nc but I also do it over an IPsec tunnel

there's a bunch of howtos on the internet for this

Well, since these machines are all within my local network (NAT). the additional layers of security aren't needed IMHO

Remilia: I know there are, I've looked at a LOT of them. And haven't found any REAL WORLD examples that I can easily modify to make work in my setup

well the above is what I used many years ago

If I get one good working command, I will just add that to my notes, and pull it from there and modify it any time I need it

right now I have a more complicated one-liner because of mediawiki

well... that's where I pulled that line from

a 15 years old file with all sorts of one-liners I came up with

If I get this spare system built today and still have free time. I might look at doing a snapshot of a users home directory and pushing that from one vm to another

(including StrongSwan pki bs)

SponiX: you cannot do a snapshot of a home directory unless it is its own dataset

I have one user that doesn't want to move over to the new box because of the additional setup required. If I can clone their home from one vm to the other, that should eliminate any additional setup needed

you do not need zfs for that

Oh?

tar solved that issue back in 1970s

Guess I could just do "rsync"

I would not call rsync 'just do' because it's more complicated than zfs

the easiest way is `tar` seriously

well for me rsync between boxes is what I normally use for any backup routines. So that is "simple" :P

The one that is "simple" is the one you already know LOL

SponiX: `tar cJf homedir.txz /usr/home/username`

though that would include path prefix

`tar cJf homedir.txz -C /usr/home username` will chdir to /usr/home so your archive will only have username as the directory inside

then on the target system `tar xf homedir.txz -C /usr/home` if the username is the same

might need to fix owner/group after that if they differ

SponiX: to reiterate, to *only* replicate a single directory and underlying stuff with zfs send, you *have* to have that directory as a separate zfs dataset (filesystem)

send/recv works on snapshot level, and snapshots work on datasets

Okay, I appreciate the input/guidance. zroot/home/username and zroot/poudriere/ports/username both exist in "zfs list" and I think this makes them a dataset

rsync -avP * username⊙110: and the like works fine for me already though. I just would like to learn the zfs process :P

my day is going perfectly cursedly... i worked around the inability of the ums driver to do its thing using uhidd. I'm successfully (ish) using X11 on a system that needed hw.dri.debug to be at 447 in order for the driver to load successfully (suggesting a race condition), and that when I leave the session, will go back to unusably jailbarred VTs, and I will need to go to my ssh'd login session in

order to shut the machine down.

MelMalik: I'm not tracking much of that. BUT, it doesn't sound like much fun

It's a bit of a lorry crash.

Remilia: FreeBSD nc can do a bunch of FreeBSD specific things among others it can request/require per socket scoped IPsec processing

which should be really interesting for those with ConnectX6 cards and their IPsec transport mode offloading

I've barely even used nc - haven't even scratched the surface of what it is capable of. inxi -Fxxzc0 | nc termbin.com 9999 on Linux is about my nc use LOL

Alma Linux chat right now is talking about a "v2" build of Go only running on "v3" hardware... What a snooze fest LOL

I feel very lucky that my X99 system makes it into the v3 category though. v2 support is getting hard to find

dunno, 1.25 works on my old home router

might be because I build it for v1

Yeah, if I wanted ever last shred of performance out of my boxes I'd rebuild everything with the fancy flags. I'd rather run 1-3% slower though and have stability instead LOL

not sure what you mean

I can only build it for v1 because it is AMD Jaguar

pretty sure v2 would crash with illegal instruction

Was talking about my machines/situations. I could rebuild with compiler optimizations more fitting if I really needed to get a small performance bump

HI

BUT, would rather just run the pre-made packages and have better compatibility/stability

Remilia: Nice... I've done my own when doing bulk builds using poudriere, and just put them in as file:// local pkgbase sources. Have never done the webserver portion and shared them out though, even though I've read the documentation :P

pkgbase huh

I've only used poudriere for ports stuff, not for base system

probably very different use cases

Yeah, this was for ports, I've never rebuilt the base system for pkgbase. In fact, I have no clue how to do that, but might research it for lolz

I've tinkered with pkgbase, and blown a few of my systems up with it already

both dev vms are just built from source for their base right now

SponiX: you mentioned pkgbase though?

'local pkgbase sources'

sorry I don't really get it, in my case I have the usual repositories for pkg (and previously for old style pkg-add etc.)

hmm, yeah, I did mention it. But not 100% sure that is what it actually is, or what I actually ment

pkgbase = base system in packages via pkg

it was putting a source file for pkg in like /usr/local/etc/pkg/custom-ports.conf

Yeah, guess it wasn't actually pkgbase then LOL

Yeah, it was actually just building pkgs and a regular pkg source file. Not ACTUALLY pkgbase I was doing -- My bad

I just looked off of the pkgbase configuration that was on a system already to make the pkg-src.conf for local

and just did edits based off of that and the handbook

SponiX: in my case I’d be doing it locally.

But I think doing it over ssh is the quick and easy.

And the only reason I was having a hard time is because I had two disks completely fail on me.

Otherwise. It went rather well. It expanded. Scrubbed. And I had more space.

Works as designed (tm)

And with regard to the second disk. I think it getting kicked out was my fault. I kept trying to clear the1 bad read over and over so I think that caused zfs to kick it out the pool lol.

I just had a good idea

installing freebsd arch-style

s/arch/gentoo/

I think my idea to have another shot of Gin is better

what does that mean polarian

in theory I should be able to installkernel and installworld to a formatted disk

then configure the installation from the host

and for at least bios boot (because I libreboot) no configuration is needed

loader will be detected automatically and freebsd should boot

you'll also need to run make distribution if it's blank

but yeah

isley: why make distribution?

It is "an idea" but what makes it a good one?

also the whole idea is to avoid distribution because it will be pkgified in 16.0

or at least you used to, maybe i'm out of the loop in this brave new pkgbase era.

need a long term no pkgbase solution

SponiX: it gets rid of the horror of pkgbase

there are a LOT easier ways to not have pkgbase on CURRENT though

pkgbase can co-exist with other methods, you can still build from source as well

polarian: because installworld doesn't install everything.

for one, you can just disable the pkgbase repo, and buildworld/installworld/installkernel like normal - Profit

isley: oh right... fuck

Why? I doubt 'make distribution' is going anywhere

SponiX: if the install is pkgified, I heard installworld/installkernel break the system

o0x1eef: I assumed make distribution will soon be changed to make the pkg tree used by pkg to update the system

polarian: you heard wrong ;)

it will work otherwise though, that's how i used to jump from ufs to zfs root before we had zfs booting and still relied on a tiny ufs slice.

It wouldn't break the system, it would update the system via another method.

'make distribution' is too useful to get rid of imho, I use it all the time when setting up jails

I'm having issues with NIS on my server, does anyone know how to wipe all nis master (accidentally set up as master) stuff so I can start again as a client?

polarian: I kinda grasp what you are trying to do. But as of right now, you are trying to fix a problem that doesn't even exist

scratch that, I just rm'd everything and restarted some stuff

i mean, installworld explicitly gets angry if the system is pkgbase these days

what's wrong with packagebase? I'm too new to FreeBSD to realy know anything about it

not trying to start anything promise

yeah i was just curious about opinions. seems like a well-trodden debate

most of the arguments i've seen against it feel like they're just emotional complaints, not really of a technical basis

that's a great point

i mean for tech stuff in general. the big things like /usr merge or systemd

pkgbase is... incompatible with doing `make installworld`, the package database and the system come out of synchronization

i think

indeed

oh ok

tht

ran into a disk quota...

o0x1eef: yeah but jails will be pkgified too

SponiX: doesn't exist... yet :)

I think it might be premature to assume there won't be a choice. We have both UFS and ZFS. Etc.

o0x1eef: just preparing for the worst case scenario

I also contemplated just dd if=/dev/random

use a BSD not a Linux wannabe

join #emacs

oops

acdw: TREASOn!

...

:)

im kidding

dd if=/dev/random of=/#freebsd bs=1m

shilling GNU in a BD channel

i actualy use notepad.exe

s/BD/BSD/

with wine

what.. why?

because notepade.exe is bae

o0x1eef: you wish :P

who is the moderator for this channel

i really like the new version with Copilot integration

.....

There's a YouTube channel called Dave's Garage or something, and he features an episode about that. He use to work at MS, and wrote the task manager.

now it's my turn to say I'm kidding :)

OG notepad is honestly pretty good. great for notes and such

like, a pad of them

o0x1eef: yeah I heard of them

watched a few of their videos

hes a little too... supportive of windows to my liking

Pretty interesting stories :)

I wouldn't hold that against him

meh, its his opinion, I just disagree with it :)

BSD is my life now, I only use Linux when I must

He seemed to start at microsoft when he was young and he was successful there so it is natural he would be in camp windows, and I think he knows its faults too, at least from the videos I saw

whats NO_ROOT and METALOG

they need to be set inorder to distrbute*

its not in build(7)

so NO_ROOT is to compile without root

and METALOG is for pkg

but neither of which are documented in build(7)

Re: pkgbase, I'd like to think I could enable a configuration setting in pkg whereby the command would automatically fail if I fail to indicate the repo I want to update/upgrade. I can't exactly remember the edge cases folks have brought up when criticizing pkgbase, but it seemed to stem from a POLA failure and accidentally ruining your system by

running a command you've always previous run and having it perform different operations

So.. why did my 8x4TB raidz2 result in 29T available? Shouldn't that be 24TB? dan.langille.org/2025/12/09/creatin…idz2-from-8x-4tb-devices-here-we-go

polarian: you just need to set NO_ROOT

polarian: I spent over an hour getting my old kernel build script to work with 15, my distributekernel line is now `make distributekernel INSTKERNNAME=kernel.$LCONF KERNCONF=$CONF DISTDIR=/var/www/kernels/$LCONF NO_ROOT=1 `

and that works just fine after buildkernel

same definitions work for packagekernel, but I have yet to figure out how to create-packages-kernel (it worked once and I cannot remember what I did to get it to work)

dvl, going out on a limb, I guess it's because of compression?

V_PauAmma_V: The zpool is newly created. What's been compressed?

Yeah, I misread the rows. data04 has 0% CAP. I don't know, then.

Your fault for not supplying a printout on alternating green and white fanfold paper instead of a screenshot. :-)

V_PauAmma_V: You have just given me flashbacks..

Sorry-not-sorry.

Hey Dianora.

hiya

Slum'n in LiberaChat?

indeed.

How have you been?

busy!

Busy can be good.

Trying to keep more of an eye on here as I find time is all.

EFnet's more and more of a ghost town every day.

Well.

It could have been fixed.

And LiberaChat has so many FreeNode-style rules.

People forget IRC is supposed to be fun.

kevans: what's the current way to upgrade a pkgbase system using the src tree on the host system?

I wouldn't think that would change.

welcome aboard

You can expand your nick if you like. You aren't stuffed into nine letters like there on the EF.

sad day today woke up to a panic

I do miss fanfold paper, those printers were fantastic.

CrtxReavr: life is not fun

MelMalik, umm . .I've been on Libera since day one.

And I'll stick with nine characters, thankyouverymuch.

Dianora: you still work on the comm ports ?

Keep it at nine, CreatureFoxRetraversed doesn't have the same ring to it

OTOH, 9 isn't enough to show off my pincers to full effect.

They would have to be shortened, as Basmil told Thorby

Yeah. That's why I have to do without on EFNet.

Remilia: thx

so distributekernel should go to /boot

and then distributeworld should go to /

then set timezone hostname, that lot of stuf

then in theory you can boot the install?

just doing this for pure knowledge at this point

and for jails I assume you just distributeworld into the root of the jail, and then write a jail.conf entry

are my assumptions correct?

what is distributeworld and distributekernel ?

mzar, see build(7).

OK

used while building a release - are you guys building new release ?

seems like drm66 is not the latest and drm69 is, see how that runs with my intel gfx

How do I match a - with tr(1)?

tr \- _

(Or whichever replacement character)

Quoting it works too

What about as part of a set of matching characters?

tr '-_' '_-'

Or rather tr '_-' '-_'

To avoid it being parsed as an option: tr -- '-set1' '_SET2'

Interesting. That's not in the synopsis, at least on 13.5.

It's not tr-specific

Other manual pages mention that, though.

But yeah.

has anybody got experience with zrep script to sync between systems? I had a sync running hourly to a remote host and now I see it hasn't synced since october. Is there any way to refresh it? I tried to sync it manually and I get many: warning: cannot send 'zroot/jails/ubuntu2@zrep_0003b5': Broken pipe. Error: Problem doing sync for zroot/jails/ubuntu2@zrep_0003b5. Renamed to zroot/jails/ub

untu2@zrep_0003b5_unsent

I used syncoid for replication

scoobybejesus_tl I will try it.

I just started using sanoid/syncoid, its pretty straightforward

specialbomb I started to use zrep because I saw an article on KlaraSystems, There're also articles on sanoid, seems nice. Will try them.