in 14.3

i have no idea

i used both

you mean can't use disklabel/bsdlabel anymore?

scheme is still there

and you can still boot mbt to the gpt?

mbr

so unsure what darwin meant

what around disklabels are deprecated in 15?

ketas, darwin: BSD disklabels aren't being deprecated, only the "disklabel" command, along with the "fdisk" command. they are both obsoleted by gpart(8)

ivy: so fdisk is going away in future fbsd releases?

black1: yes, it will be removed in 16

iirc, it was meant to be removed in 15 but no one got around to doing it, so it was pushed back

okay alright fbsd seems to be taking a radical step in this

radical how? what do you use fdisk(8) for today that you can't do with gpart?

“Radical” feels a bit strong, but I don’t think that’s the question. It’s still the removal of a tool with many decades of history.

(I take no position on whether it’s the right thing to do under the circumstances. But don’t downplay the significance.)

ivy: well these two basically have the same functionality... but it might pose a cost for those used to the fdisk-style syntax for example

anth: sure, but on the other hand very few people use it or even knows that it exists nowadays, so the actual signifiance of its removal is fairly low

the only interaction most people have with fdisk is switching from Linux to FreeBSD and assuming our partitioning tool is called fdisk, which leaves them confused until they find gpart

ivy: I mean most GNU/Linux distros choose to stick with fdisk as a standard utility...

black1: the FreeBSD equivalent of Linux fdisk is gpart

honestly I haven't personally used gpart except in a GUI env

ivy: so gpart was specifically developed for fbsd?

black1: yes, gpart is part of GEOM, which is a FreeBSD feature. maybe DF also has it, i'm not sure

okay alright i see

thx for the information btw

it reminds me of ubuntu getting rid of ifconfig at one point

i wasn't aware any Linux distributions still shipped ifconfig, at least by default

I think debian still does

and KNOPPIX definitely which I gave my grandma for solitaire purposes

ivy: Thanks for the information about fiber a few weeks back. Finally got around to ordering some stuff.

wipt: good luck :-)

fiber?

also

linux updated the fdisk to gpt and kept the same tool name?

i don't have access to any linux systems right now to check, but from what i remember, yes

while here somebody had nice idea to rewrite disks

it's all pentagon's fault anyway

:p

GEOM(4): HISTORY: This software was initially developed for the FreeBSD Project by Poul-Henning Kamp and NAI Labs, the Security Research Division of Network Associates, Inc. under DARPA/SPAWAR contract N66001-01-C-8035 ("CBOSS"), as part of the DARPA CHATS research program.

good

black1: shipped if you install net-tools

is anyone working to improve jails/racct to fix lists.freebsd.org/archives/freebsd-current/2025-October/009183.html ?

why doesn't our malloc(), in debug configuration, report an error when trying to free() a pointer that wasn't returned from malloc?

i feel like this is the most basic behaviour you'd expect from a debug malloc, so perhaps i'm missing something here

ivy: Seems bugworthy.

ivy: how would it know? that's like a valgrind level of out of band data isn't it?

rtprio: presumably it would know because of all the debug information it stores?

like, if it's ~10% slower when debug is enabled, but it can't detect this... what is the purpose of that performance penalty?

i'm annoyed because this was the cause of reviews.freebsd.org/D53545, which i would have expected debug malloc to flag

is bootloader automatically updated when doing make installworld? ( an UEFI boot )

angry_vincent: no, you must copy /boot/loader.efi to the EFI partition

ah ok. they have different time of origin indeed

one in efi partitions is older by 6 monthes

what does ppl running for wireless networkmanager?

or does you guys just appending in wpa_supplicant.conf?

wpa_supplicant for me, but I am not a normal person and shouldn't be imitated

Hey all, I have a laptop which crashes on sleep or wake, not sure which one, but in any case it's not writing to /var/crash. How can I get someone to post to the bug tracker?

tehpeh: why do you need someone else to do that?

ha

some*thing*

How can I get a core dump or similar?

tehpeh: you can try setting debug.acpi.suspend_bounce=1

and suspending

it suspends everythign then restarts it

you might get a panic you can see with that

you can also try using devctl to suspend individual devices

this page might have more things to try: wiki.freebsd.org/SuspendResume

[tj]: awesome thanks, I'll try that

[tj]: so just ifconfig wlan0 scan and then genrated a new entry in wpa_supplicant and reloading with service netif restart ?

feels like you get stupid/lazy after some years with macos... :P

wpa_passphrase network passs >> /etc/wpa_supplicant.conf

wpa_cli reconfigure

you don't need to restart networking to move around wifi networks

any ui that uses wpa_supplicant should work, but I frequently have a fresh system

morning

I'm so close to getting bluetooth paired to my headphones. I'm following this tutorial: jrgsystems.com/posts/2022-08-20-how…luetooth-headphones-on-freebsd-13-1 . I've successfully gotten to the virtual_oss command, but it doesn't like the -T option. How do I modify the virtual_oss command to make it work?

are you guys using devd with a custom script to suspend/resume when closing lid ( and if external monitor is connected dont suspend) etc? looks like I dont have permission to run zzz as a normal user.. is the workaround there to use sudo with zzz command?

question, time to time, when ssh session interrupts, the terminal windows stops accepting any commands, just ptinting keys like `127;129u[127;129u[127;129u[127;`

is there a way to fix it other then killing shell process ?

nwe, - I never manage to fix it ...

I have a fresh installation of 14.3 on a nano pi R4S (using the rockpro64 image). I does not install pkg because it says it can't verify the certificate but was able to run `freebsd-update fetch && freebsd-update install` to update the OS. I've tried useng `fetch` to any URL and says again it can't verify the cert.

nerozero: try `reset`?

cyric, no verbal command accepted

"reset[13;129u"

that is the "reset and enter

not executed ...

kinda like "visual" in vim

weird

~. also does't seems to

this is the error:

Bootstrapping pkg from pkg+https://pkg.FreeBSD.org/FreeBSD:14:aarch64/quarterly, please wait...

Certificate verification failed for /CN=pkg.freebsd.org

0850E7F5D15E0000:error:0A000086:SSL routines:tls_post_process_server_certificate:certificate verify failed:/usr/src/crypto/openssl/ssl/statem/statem_clnt.c:1890:

Certificate verification failed for /CN=pkg.freebsd.org

and this when I test whatever url using fetch

# fetch google.com

Certificate verification failed for /CN=www.google.com

081066E46A8E0000:error:0A000086:SSL routines:tls_post_process_server_certificate:certificate verify failed:/usr/src/crypto/openssl/ssl/statem/statem_clnt.c:1890:

fetch: google.com: Authentication error

any workaround ?

running 14.3-RELEASE-p5

oh, seems the system date is way off. Let me try to fix that first

that was the issue, sorry for the noise

hey does anyone have a good intro to jails? I tried docs.freebsd.org/en/books/handbook/jails but I could use a little more guidance

spuos: the books from Michael W Lucas are well regarded and there is one about mastering FreeBSD jails.

I using this devd/lid.conf dpaste.org/gz2Ku when I running the script from terminal it working, but it looks like devd doesnt trigger and running the script.. who can I debug this?

psycorama: I do appreciate the "proper" answer, and I often do want one, In fact an in-depth resource helped me understand PF rules. That said, is there a layman's quick guide? I'm trying to set up a quick and dirty game server on my multiuser system after my VPS died, and my friends are getting a little annoyed.

use one of the frontend. that's probably the fastest way. i'm still using iocage but migrationg to bastille

bastillebsd.org/getting-started is probably a good start, then

would you reccomend bastille? I didn't know frontends exist

oh, I guess you would then. Thanks!

i remember a table, quickly comparing the different sollutions. but i can't find it, currently.

but yes. bastille is the newest of the solutions und seems well mantained

aaah. there it is bastillebsd.org/compare

Hello, does anyone know if we're getting RDNA4 support in FreeBSD 15? I'm sure I read somewhere that there was limited support in drm-6.9, but I tried BETA4 last night and it still crashed as soon as I loaded the module. Am I doing something wrong, or is it still not implemented?

rdna4 support seems to first appear in 6.17

No, it's works well in 6.12, I'm using it right now

6.12 is more than 6.9

I know it works in 6.12, and it probably works in 6.17. What I'm asking is does it work in FreeBSD 15?

those first two things are different than the second thing

(third thing)

the drm-kmod kernel driver is synced to linux kernel revisions

I'm gonna say you are hitting "limited support"

That suggests it would not be expected to crash as soon as the amdgpu module loads.

immediately panicing seems pretty limited to me, it is more support than not detecting anything and refusing to load

you could try master on drm-kmod from git

it was doing the same on 6.6, instapanicing

But I was not expecting that to work, so thought nothing of it

yeah, i wouldn't really expect anything unless you've tested it yourself on linux at the equivalent kernel version

what's going on in freebsd these days

i haven't paid attention in about 13 years

dho: why were you interested in freebsd 13 years ago ?

i used to be quite heavily into the community, was a chanop here quite some time ago.

cool

places i worked stopped supporting freebsd, it became cognitive burden to keep up with the system amongst other things

I see, bad places

no, not really. linux just won the developer market

the slogan that FreeBSD is dying probably won

market share wise, it's certainly smaller than it was 20 years ago

yep, and most likely smaller than 13 years ago

market share is shrinking, but some people still use it, and they most likely will, at least until 2030

dho: regarding the developer market, the thing you might find interesting is that freebsd is now an official OCI target, has a working podman (and has for a while) and can run linux containers in basically the same way you'd do it on linux via podman

...and we have a working libuuid after almost a decade!

not surprised -- was talking with the folks at sun about their brandz stuff doing effectively linux containers on solaris in 2006, did some proofs-of-concept of thesame on freebsd using linuxemu

I had to create, administer, and analiize surveys and survey data a few times in college, so I feel a certain amount of empathy and sense of responsibility when I'm asked. . .

somehow i've managed to work on cloud stuff for ages without actually working on any of the cloud stuff though

i mean, linux jails were a thing for a long time. it's just that you now can use podman to fetch stuf from ghcr.io, docker.io and so on

But. . . some surveys get way too fucking long.

lol hi CrtxReavr, long time

dstolfa: can run linux pods? how long ago did that happen?

rtprio: been a while i believe. i think all you need is pkg install podman-suite and enabling the linux service

you should then be able to use podman to run containers that are based off of linux under freebsd

huh

cool that it's an oci target tho. are there any particularly interesting system-wide development efforts?

dho: it’s a handful of developers pushing the cart, with two larger firms pulling from the front

i assume netflix is still one, who's the other?

please take a look at commit messages, I cannot use any commercial names

lol, wild.

So I’m trying to put a separate network on lagg0 so I can have a “fast” network that bypasses the router so my fbsd nas can just go straight to clients. It doesn’t seem to be working out too well. I figured it would just be making sure samba is bound to both IPs and add a second ip/mask to lagg0. Do I have this wrong?

rclone, for instance seems to be getting hung up after mount and says I/O error.

what's the network topology like?

Macer: you need a kind of policy based routing, it can be done with pf or ipfw

Ah ok. Would this be the same even if I created a second lagg?

are the clients reachable directly or via gateway ?

That is an option. I have 2 10gbit and 2 1gbit ports to work with. I just want the fail over so I can make lagg0 10gbit+1gbit failover and the same for lagg1 if that makes it easier.

failover is one thing, double address scheme and separate network, another

mzar: right now I wound up putting it back the way it was. fbsd nas is on a 192.168.1.x network and clients are on 192.168.20.x vlan

The clients are proxmox Debian containers.

.1.x is untagged

20 is tagged

why not add a 192.168.20./24 address to the nas?

lagg0.20

That’s what I tried first.

Adding an alias

alias is not a vlan

that's not an alias, subinterface

Oh. Ok. I was working on the assumption an alias was the way to go there.

So is actually he creating a virtual interface?

*be

I’ll see if I can hunt that down in the handbook.

if lagg0 is untagge, use lagg0.20, if lagg0 is vlan1, use another lagg

Yeah. lagg0 is untagged vlan1

Macer: it will work, unless you want to use lagg0 in bridge

is bridge(4) involved ?

Sounds like the quick and easy is to make an actual physical interface.

Like just have the untagged data on lagg1

please don't be afraid of tagging

Well. I’m a little afraid of tagging and getting locked out. lol.

do you have ipmi console there ?

O suppose I can ensure there is an untagged port.

you can leave untagged vlan 1, just add tagged vlan 20

Hm. I don’t think so. That is my ancient dual Xeon isilon with the ancient Java based ipmi 2.something that for the life of me I haven’t found a way to access in forever even with IPMIView.

I bet you can configure it to use ipmi sol, it's old feature

Yeah I’ll have to take a harder look. It should just be a physical port tag shouldn’t it?

Yeah I was trying to sort that out too so I could use the serial on lan connection. I just haven’t had time to look at it too hard.

vlans_lagg0="20" ifconfig_lagg0_20="up 192.168.20.254/24"

Oh. Fbsd can just tag it itself?

sure, but you need to configure port on switch

I’m guessing the “20” is the vlan id?

to pass tagged vlan 20

The IPs are 20.x but vlan id is 201. That’s awesome though. I’ll try that when I get home and see how it works. Maybe that was the issue all along. Although. I was able to ping.

I’m honestly not sure what happened after that. rclone smb mounting didn’t work for sure. Not sure if that’s the reason.

I’m just looking for a direct path that doesn’t hit the router since the router is gbit only. I thought that wouldn’t be an issue… until it was.

If this works I may actually make a data only vlan.

new samba uses heavily DNS, perhaps the client is still looking for address from vlan 1

Yeah I thought that too so forced the address with hosts

And also tried setting up rclone for the ip

Since the local dns points to the 1.x address

I thought once I was able to ping the alias and change it to the ip in rclone it would work for sure. It didn’t.

take one step at a time, create tagged subinterface on lagg0, test connectivity with the clients, then teach samba to behave

And direct nfs isn’t an option in unprivileged lxc containers.

Yeah. I’ll have to give it the college try once I get home. Going through the router is hammering it. lol

It will give me the opportunity to take the time to properly label them in the ubiquity interface too.

Macer: good idea; were you trying to ask any LLM for support ? what brought you in here ?

No lol. I’d die if some LLM sent me to irc.

cool

I just figured I was configuring it wrong in FreeBSD and the nas sounds like the better place to start.

It just runs vanilla fbsd

I wouldn’t trust an LLM anyways. It would probably tell me to wipe it and install windows server.

I just asked Gemini "Would you send me to IRC for any answers you were struggling to find?"

hhe

"The direct answer is no, I would not send you to IRC."

"If I were struggling to find an answer, my protocol isn't to send you elsewhere, but to try a different approach _with_ you"

dho: the amount of people seeking help significantly dropped since asking LLMs became an option

double edged sword but i think in general it's a good thing

hmm when I using FN+FX on my laptop, but when I using my external Keyboard the FN isnt reconized :(

the fn key is typically a 'hardware key' and usually changes to its behavior require bios stuff. not surprised an external kb fn key wouldn't work

dho: yup.

dho: okey

(also usually the only thing you can do with it in the bios is tell it to swap with ctrl since vendors decided to place it where ctrl ought to be)

(and sometimes the only thing you can do is have it work as ctrl for "popular hotkeys")

itś not a really big issue, more nice to have, because itś working fine in macos etc :P Itś a keychron wireless keyboard I using btw :P

the issue isn't with the keyboard, it's with how the fn key works on the laptop

you won't find a keyboard that has a fn key that will work, basically

I have an even more annoying issue, that devd doesnt trigger my script when closing lid (so it going to sleep mode) it works if I running the script :)

dho: okey

mzar rtprio: that's what i have in rc.conf... i'm a little confused about the defaultrouter part

wouldn't i wind up using a different router for each subnet?

that's me trying to split the physical vlans .. although i'm curious if i have to add the vid to 20.x

(still)

would i still need a vlans_lagg1="201" there for the vid?

there are 2 ways to do it: or at least there were in my rc.conf history

oh, if you need vlans_ depends on how you set the switchport

Macer: you cant have more than one default gateway, without setting up something like multiple fibs--which has its own sets of complexity

i'm curious at this point if i even need the vlan tbh

Macer: what are you trying to do?

since now it's a dedicated port

zi: i'm trying to split 2x 10gbit + 2x 1gbit to (1 x 10gbit + 1 1gbit) x 2 for separate subnets on the same fbsd server

so the fbsd server has a fast lane that doesn't need to go throgh the router (which is 1gbit) for samba, nfs, etc

so i'm essentially making 2 networks on it. one that talks and one that transfers copious amounts of data

the 1gbit ports are just for failover

why separate subnets?

because the clients need to go through the router to reach the other subnet

and i want them to go straight to the nas

the router is 1gbit max .. so vlans talking to each other have a bottleneck

a small switch with 1000/10g ports would probably do everything you need without anything special

(config wise)

no second subnet or other stuff

i have a EdgeSwitch 16 XG

but i don't see how i'd be able to get to go vlan to vlan without the router telling things where to go. i'd need a switch with routing capabilities and i don't think mine does that

my solution was to just put the fbsd server on 2 subnets

i dont suppose you can make a quick diagram of the bits involved?

it has 4 nics 2 10gbit and 2 gbit .. so i was just going to make lagg1 with the other subnet and see if it will traceroute to clients without hitting the router

you don't need additional gateways, as the nas already knows how to get to 192.168.20.0/24

so just ... ignore that

you would just put all of your 10g clients on your switch and connect a 1g port on your switch to your router

Macer: i don't think you need a routing switch for that

everything local to the switch will do 10g

stuff leaving the switch will go 1g through the router

no routing on your side needed

yeah. that's what i'm trying to sort out right now ... before i had nas on 192.168.1.x and clients on 192.168.20.x .. so clients connecting to the nas are traveling through the router to get there

just collapse everything to a single subnet

then i lose the vlans. i want to keep them :)

that's how i originally had it

for what?

i have an iot vlan, guest vlan, proxmox servers (containers) vlan,

to keep things separate

are you doing firewalling or something between them?

not yet

right now i'm just trying to figure out how to have a fast data path between the clients and nas

then, id probably just collapse it all to a single vlan. you can logically divide it up into /26s for organization, but all on the same /24

i'd be kind of working backwards doing that

you only need vlans if you need some separation

and if you need vlans, you'll need routing

yeah. unless the nas and the clients are on the same subnet then it would be handled by the switch directly and not have to bother right?

should turn it into a direct path

if everything is on the same subnet/vlan, then the traffic would be switched locally, yes

yeah. i'd like to keep the vlans while creating a fast data path. so that's the goal right now and my solution is to just make another physical connection.

as in you want to give other devices dual interfaces and dual ips?

either way. you really dont want more connections imho.

if you want vlans, setup vlans and routing

ok that didn't work lol

i went with that. i have lagg1 and an ip but can't ping

ah great

exactly what i was trying to avoid lol

eh, can't ping what

i can't ping something on the 20 network

20?

and you set the port on the switch to be native vlan 20?