I'm currently looking at the code

alrighty. have bastille running on what used to be the old proxmox server :) and managed to make a vnet jail on a vlan

that actually works heh

i wish bastille had an ncurses interface :) like some tui thing

cbsd does

does it?!

github.com/Peter2121/cbsd-tui <- this?

I burned the disc image onto the drive using dd, now it boots :) ; I don't know why it didn't boot when using ventoy.

Ventoy requires some custom something for every OS that it can boot. It may seem like it is some type of generic OS magic-boot tool. But as I understand it ventoy knows about every OS it boots and usually needs to be customized for each one.

Which means that every time an OS puts out a new release I hear everyone complain that ventoy doesn't boot it. Until it is updated to do so.

hmm, i had the assumption that ventoy was a magic-all OS boot tool too, thanks for the info rwp.

I just know that every time there is a new major OS release made by Debian, Ubuntu, *BSD, that afterward I see a lot of people reporting that Ventoy can't boot it. People eventually say update to the latest Ventoy and that this fixes it.

wtf is ventoy

rtprio: ventoy.net/en/index.html

oh

why is my freebsd server killing processes with there is plenty of swap

how much RAM do you have

and what are the processes? i assume they are dying as out of memory?

What is being logged to /var/log/messages about it?

1 gb of ram, 2 gb swap

in the log there's - Dec 20 03:09:13 freebsd kernel: pid 13749 (valkey-server), jid 0, uid 537, was killed: failed to reclaim memory - Dec 20 03:09:19 freebsd kernel: pid 1633 (node), jid 0, uid 1002, was killed: failed to reclaim memory

Initially I would think that whatever process you are running is exceeding what you have. But... How much memory is the program you are running using?

my node process is using around 200 mb

valkey is using 35 mb

is there a pkg cron job or something? i'm seeing something like that in the graph

i don't know what this is but i'm suspecting it - /usr/local/etc/periodic/daily/411.pkg-backup

looks like it's due to vm.pageout_oom_seq

Personally I like the "htop" tool best. It has a graphical memory bar which I find useful. I would "pkg install htop" and then have it run in a terminal while I am running the program and then see what the system is doing. But... Impulse spikes of memory use will be too fast for htop to see.

a pkg cron job runs at that time, and pkg takes too much memory, and then the OS starts killing stuff, regardless of swap due to vm.pageout_oom_seq

rwp: i forget about htop. i used lnav and netdata to see what happened at that time

makr: are you using UFS or ZFS btw?

Remilia: UFS

for starters I'd probably watch top while trying to trigger the crash

then there's dtrace

Hi! IDK if anybody here could be interested, but I had some fun lately with a tarpit for web crawlers, and I run it on my freebsd webserver. Project page: dacav.org/projects/supercazzola ; The result looks like this: dacav.org/L/tarpit-demo

hey

my php-fpm (8.4) on freebsd 14.3 often crashs with SEGV but no core file is produced

does someone has trouble with php 8.4 which crashs ? And how could I have core file in order to understand why it crashs?

ulimit for core is unlimited and kern.corefile is set to a location where the user can write the core file

dacav: as i can see you run lighttpd/1.4.82 on your server. Do you run it with ktls enabled?

Pauli1: not that I know of

dacav: moment, i will show somewaht ...

OK. If that matters, I use lighttpd from pkg

dacav: coe.chickenkiller.com/disposal/finger.png

uhm... butchered picture?

Because I have the same strange thing with my box when lighttpd runs with ktls

That picture is from your site

This one

Right. I never noticed this problem before

So i'm not alone with that. ;)

That's great! :D

Are you sure it is due to ktls?

It seems so because the is corrupt

sysctl kern.ipc.tls.enable

sysctl kern.ipc.tls.stats

I can't reproduce the problem btw

How do I?

ssl.openssl.ssl-conf-cmd += ("Options" => "-KTLS" ) in lighttpd.conf will "solve" that but ....

Oh, you can't reproduce the problem ?

Nope

That's sounds good. So maybe the problem is here only

I open the picture in firefox

I just opened it on firefox (well, librewolf) and even curled it

what if you curl it?

And got the result

Same lynx dacav.org/pics/neo.jpeg

Pauli1: could it be a matter of latency? From your IP you look kinda distant

the server is in London

I'm Germany, near by. ;)

Oh, OK. Then maybe I took someone else's IP

I curl it with another Server and the picture looks fine. Seems the error, mistake, whatever is here on my machine

uhm uhm

Let me try to enable ktls from my client side

nope

(assuming curl makes use of that)

I have no clue what it can be. sendfile, lighttpd on ZFS ...

What ever I've to try and try ... ;)

One question, do run FreeBSD 14 or 15

on server freebsd15

14.3 here

Ah

client-side it is 14.3

Thanks

Any time :)

Feel free to ask for details if you need (if I don't answer just /query me)

(I will eventually answer)

Okay :)

for the life of me i can't seem to get rclone to mount smb in a jail :/

i don't remember having issues like this with jails before

Anyone manage to the new MIT kerberos in base working with freebsd15? It does not seem to be compatible with cyrus-sasl-gssapi

Macer: can you use a nullfs mount in the jail?

mount_fusefs: /dev/fuse on /mnt/test: Operation not permitted

rtj: possibly

i know rclone works on the host

i've been messing around with jail.conf and devfs.rules nonstop for the past hour

trying to sort out why rclone just won't work

not really sure what i'm doing wrong here :/

I've been trying to trim the fat on the void box so I can zfs send its fs elsewhere and play with BSD for a while, and boy fucking howdy do Discord, Slack and Signal Desktop use most of the space. Also: xbps never clears up after itself either

ah well... let me shut it down and take a break from it lol

trying to get to a point where i get past the learning curve to start using freebsd for a jail/vm server vs proxmox

but it's things like fighting this... :)

oh wait.. think i missed something

i guess it's allow.mount.fusefs not allow.mount.fuse .. argh! i forgot the "fs" lol

ok. that worked. now i'll take a break. i still have to sort out where to put rclone so i can use it in fstab.. i know it's different in freebsd

If you die in Linuxulator, you die in real life

haha

well right now i'm trying to sort out how i'm supposed to rename rclone to act as a mount helper for fstab

-t rclone -t fuse.rclone -t rclonefs all say invalid fs

the helper does seem to act as a helper when rclone is linked to mount.rclone or rclonefs

wow.. i just remembered.. i put that on my abandoned blog lol... crazy.macer.life/freebsd-jails-rclone-mounting

and it still doesn't work. that was with 14.. i doubt much has changed since then... well a lot did but not this

ah ok. so it seems i found it impossible to mount using rclone IN the jail and had to resort to mounting it into the jail from the host

Hi all, is it possible to put FreeBSD on a chromebook? I keep getting kernel panics with ubuntu, would love some stability.

hello

how can i upgrade from 14.2 to latest version?

pretty sure you can upgrade from 14.2 to 15.0 directly. Conservatively, upgrade to 14.3 latest and then 15.0

Is it possible to put freebsd on a chromebook if I have access to the UEFI to boot into whatever I want?

Good morning everyone! I've asked this in #freebsd-irc, not realising it was the overflow channel. I'm trying to compile world from source, but i'm getting an error while compiling bootstrap. I get an error at _bootstrap-tools-link-[`, namely `Cannot find host tool '['`. What am i missing ?

violette: are you running make with any special options or have you modified any Makefiles?

my first thought is that there's either a malformed option or maybe a malformed environment variable - my second thought is maybe you're running a different `make` than /usr/bin/make?

This is at around line 2830 in /usr/src/Makefile.inc1 if you wanna take a look and see if any variable names jump out at you

Host tool '[' is "/bin/[" and that seems odd that /bin would be missing. And how would anything work if /bin was missing?

I think it more likely that the build environment is somehow just completely broken.

I wonder if $PATH is completely clobbered or something

I would restart from square one. I personally would create a pristine jail and build in there to ensure that the build system is in a known good state.

lol I'd forgotten that `[` is an actual binary and was assuming something wasn't being parsed properly

but yeah what rwp said - start in a fresh environment

This feels like a case where there is a simple error message about not finding the '[' command and we could chase down that detail but having that happen is really an indicator that something waaayyy earlier has gone very wrong. Because [ is a builtin to almost all shells since a few years.

yeah great call

hi all, any ideas about running any of the bsd family on a chromebook?

no idea

Guest95: are you a FreeBSD user ?

no, but i am thinking strongly about it. I skimmed through the michael lucas books

it's good OS, feel free to try

my only concern that i would need to learn more about is docker/kubernetes. I guess you can use instead jails, saltstack, nomad, pot, bastille?

polarian is going to give a talk, nice

Guest95: check out podman too

Guest95: and Sylve

ok

fairly recent development and I believe you _should_ be able to run a lot of Docker containers natively now

that is cool, but what about something like kubernetes?

nprice: have you tried ?

I just run jails tbh

OK, same here

I have a bunch of hacky scripts that make them act like ephemeral Docker containers

but how do you scale everything like on a cluster such as kubernetes?

not much point on a single system

because i would love the stability from BSD but also the scaling like kubernetes

without running multiple systems there's really no scaling that kubernetes will do for you that you can't get out of a single container

right, you would need multiple systems set up, but how would you do that on BSD?

I'm not sure there's a great answer to that right now

would be standing up multiple systems with jails and front them with an nginx load balancer or something

you can certainly do it but there's gonna be a lot more manual "plumbing" for you to handle than you would running k8s

yea that is what i was afraid of, that would be the only thing keeping me on linux but then there are downsides to that which freebsd would solve

Guest95: one solutions could be running 3 vm with (bhyve) and install linux and kubenretes om them?

hmm interesting so the main os is BSD but then run linux VMs on them?

seems like it might be slow if you do that

Guest95: itś depends if itś for learning (homelab) they it would be okey, but not for production then i propbly to slow..

right, this would be a prod system

hmm dont know iif you can give a your bhvyve direct-access to disks if that maybe will be better performance.. *brainstorming* :D

i appreciate the brainstorming

but try it out? I mean kubernetes node in AWS /GCP are virtual-machines :)

you might be surprised about how lightweight bhyve is even for full-hardware virtualization

Guest95: so you have to run Linux for your kubenretes, no worries, FreeBSD is not any real competitor

I should setup 3vms installing linux with k8s, and test it out..

usually, you need k8 for your shareholders, if your business can scale, they will be pleased OC

but Netflix doesn't care

but maybe they are hiding it ...?

a giant pile of shell scripts can get you surprisingly far lol

can you explain further mzar what you mean or how they are doing it?

I don't know TBH

no worries

k8 is history, now everyone is focused on these ollamas

lol true

you think it is a bubble that will pop?

irc is history, now everyone is focused on Microsoft Teams

:')

does anyone else move distfiles to a place not on /usr/ports/distfiles? I do that but when I attempt to builkd it chokes on $DISTFILES for some reason... worked around it by changint /use/ports/distfiles to a symlink

yes, ollamas will take over bitcoin miners

There's definitely a level of bubble involved but I think a lot of the technology has proven too useful that it'll simply go away

WWRKDIRPREFIX still works ok though

yea agreed, but i feel like everyone will just switch to the best model available at the time and the best one wins

shbrngdo: WRKDIRPREFIX or WWRKDIRPREFIX ?

shbrngdo: how goes it ? are you still running FreeBSD ?

shbrngdo: WRKDIRPREFIX= /var/ports and DISTDIR= /var/ports/distfiles works excellent for me

hm. i was going to use freeipa as a dc .. but i guess they don't have anything for freebsd

only fedora and rh it seems

How to use rsync to backups in zip files?

X-Y problem

er, I'm wrong

as far as I know rsync cannot be used to make backups to zip files

but one can use tar(1) to backut to zip files

s/backut/backup

zip_: are you doing backups with tar ?

One day I'll switch away from IRCCloud and use a client which lets me turn off handle notifications

lol

I'm doing backups with Time Machine and backblaze but I'd probably use restic once I kick the Mac to the kerb

Oh I also burn things like photos and music to blu ray because immutable backup storage is good actually

LOL

there's always option b. change your nick :)

does anybody know why samba is packaged with different major versions? i guess that's because upgrades to samba can break things a special kind of way?

i've had to cycle it a couple times and it seems like removing / installing newer works but i was just curious why it has samba416 samba419 and samba420 instead of just samba

i see it packaged that way in a lot of places - i assume there are client-specific incompatibilities or whatnot, but it would be nice to have "samba" just point to whatever is the latest

or something along those lines similar to how we do with php and golang and nodejs

I’ve never seen it that way other than in FreeBSD for minor versions. I could see samba 3 and 4 being separated though.

nprice: samba.org people publish security updates for 3 different minor versions so…

kind of like mediawiki I guess

Hi I am trying to setup mdo(1) on a fresh install of 15.0. I have mac_do kldload'd and sysctl security.mac.do.rules="uid=1001>uid=0" but I still get the error when running mdo from my user: mdo: setcred(): Operation not permitted

I can't find a setup guide, and I thought I did everything the manpage said. so I am a bit stuck

how to make a dir drwxrwxrwt?

$ ls -l /

drwxrwxrwt 2 root wheel 512 Jun 6 2025 tmp

Chip1972: by setting the sticky(7) bit

how, in mnbame

how, in name of Christ?

ops

how? in name of Christ.

with chmod(1) the mode takes 4 octeds the manpage has the details: man.freebsd.org/cgi/man.cgi?query=c…+and+Ports&arch=default&format=html

God bless you... could you print it?

when you tell me the hostname of your cups server

and your printer is not on fire

Amen

(19:25:07) nologin: Chip1972, man chmod; $ chmod 1777

Saved by an angel...

Chip1972: we'd prefer if you didn't insult people for being satanists.

my name is satanist, but this misinterpretation is intended and for me to laugh

how to change nick?

Chip1972: /nick newnamehere

 

mysterious

so i'm trying to use samba to provision a new domain (controller)

and i'm getting some sort of python error

hm.

deimosBSD you think is better to upgrade 14.3 first then 15.0?