anyone else have AVAGO MegaRAID SAS 9361-8i that causes checksum errors with zpools? the drives are all fine according to smartctl

the controller is in jbod mode

MelMalik: if you want to perform lockdown in jail or are on low end flash storage, you could run everything you want ro, it gets a hassle in some cases, but you said normal eh, hard to say, first things that could be ro are things that are never written to because they lack (suid) permissions

ketas, I want to put the brakes on a hypothetical jailed customer breaking their own system, so I want to mount as many directories as possible RO.

MelMalik: do they manage it? then no real answer except education and system restore capabilities, maybe with fees... if they don't it could be ro and even rootless

i though it was security question first :p

what breaks is not in /tmp and /root and /home... and maybe not in part of or even full /var

but this is like how to still eat while not being able to put fingers into your mouth

ketas, security in the same way a seatbelt is security i suppose. prevent them from flying through the window if things go horribly sideways

regular things don't help?

idk

yeah jails have fancy option to make fs ro even for roots

:p

i mean, it is mostly academic

the idea is

i have jails where service jails have everything ro except /root where all the stuff lives in

a bit hassle

other rw's are /tmp & /var

one jails manages what passwd has in it and what's generally runnable on this

rc.conf is under /root as well :p

unconventional

interesege!

all jails share a / and /usr/local which are separate fses ro

s/ro/and ro/

one has own rw /etc since i gave shell to a friend and he insisted using PASSWORD for ssh

unsure why i created / structure under /root

was already there i guess

:p

had to patch rc to allow extra periodic.conf

users are numerics that are preallocated by some prefix

seems like mess tho

basically did own jail mgmt

base upgrades consist stopping jails, doing zfs rename and then starting them again

and probably host too since it's same

i have some directory what i tar | tar over resulting installworld

where my base config is

sad that unionfs sucks so hard

even after 20+y

that has uses too

right now it just makes my livecd virtially "rw"

if it blows up i have nothing to lose

i wonder if the union concept should be implemented at a lower level, perhaps at the vfs level rather than being a filesystem driver

unsure

hackers mailing list had someone wanting to fix it

but not change concept

!

This year's EuroBSDCon is 1hr drive from me \o/

Never gone, but considering it's *that* close ......

DarkUranium: i have wondered about going there, multiple times

it keeps moving around so it's either more or less appealing

good afternoon!

Is there any quality control in the ports at all? I mean, I installed KDE Plasma 6 on FreeBSD, right clicking the panel crashes KDE Plasma, and pressing CTRL + C in Konsole crashes the entire PC outright.

:(

remiliascarlet: there are binary packages available?

Yes, I installed via PKG.

The only thing I literally compiled from source in the ports collection is Zig, this is so I can use Zig 0.14.0 instead of 0.13.0.

You know. . . I really feel like ports have suffered over the years.

Packages went from convenience to requirement.

So many ports won't even build anymore, outside of poudrier.

requiring packages makes sense since not doing that requires a bunch of code duplication to handle things like pkg-messages. but the poudriere thing i agree about, either we should require poudriere or things should work without it

CrtxReavr: well KDE was never a great thing on freebsd

I completely disagreet.

I remember having a KDE setup that rocked.

and I remember having it on freebsd many years ago and it wasnt the quality as on Linux

Then nVidia released their FreeBSD drivers and it only got better.

I think 5 was very late ported to freebsd kinda, but I did quit kde then eventually anyways

Though. . . I stull normally favored a lighter weight setup.

and no clue about 6, these days I just let the mac do what it can do best.. desktop

blackfox/fluxbox

Lightweight, but still supported gtk/gnome apps.

speaking of drivers, should I not be able to pkg install drm-kmod on i386 14.2?

No packages availabe to install matching 'drm-kmod' have been found in the repositories.

if i want to install it from ports, do I understand the Makefile correctly that I need drm-510-kmod first?

benjamino: drm-kmod is supposed to create a metapackage that installs the correct version automatically, but perhaps it failed to build... i wonder how well tested it is on i386

i mean xorg and lxqt work fine, do I need it at all?

if you have a gpu supported by the drm drivers, it will significantly improve performance, but it's not technically required

i would suggest trying to build it from souce and see what happens, if it fails you can open a bug

ivy: i'll give it a go, thank you!

benjamino: somehow I doubt that if you're on i386 you have a GPU that would benefit from kmod? what hardware are you running?

mtll: its an asus eee pc 1005p, intel graphics

nice

i ran make install clean in the drm-kmod port, its still going...

oh interesting

is that with an atom n450 chip? wikipedia claims it has x86-64 "if enabled", whatever that means

is that one of those weird early machines with a 64-bit CPU but a 32-bit EFI firmware?

some cpus could do more than 4gb of ram on 32bit

it is N450!

morpho: that's PAE though, not x86-64

does this mean i could install a 64-bit freebsd?

benjamino: maybe, not sure

see if there's a bios setting for it?

should be called "Intel 64" or something like that

ill try it as soon as make finishes, thanks, i never thought it could have 64-bit support since its so old

if it's 32-bit EFI (e.g. eeepc 1025C) you need freebsd/freebsd-src #1098 and i don't know if this made it into 14

N450 is way more recent than x86-64

it's just that these atom chips were extremely stripped down

looks like that was only MFC'd last month so probably not in 14.2

CrtxReavr: Well, KDE is pretty stable on OpenBSD 7.7 (snapshot, still waiting for the actual release).

dmidecode says its 64-bit capable

so this might just work hehe

anyway, make failed, it says that drm-510-kmod is unsupported on 14.2 and higher, ill just go and check the bios for the 64 bit option, brb

no option to activate 64-bit in BIOS, ill just try to install the 64-bit OS and see what happens

I assume if you live boot some 64bit thing it'll shit itself well before being able to install something, if there is no 64bit support

Is there a good process for migrating a normal UFS installation over to root on ZFS? I'd be doing it on a separate drive too, but wanted to avoid doing a full installation and then migrating everything over manually

hodapp: you can pretty much just boot from an installer/rescue system and copy it over with tar. you'll need to look up how to create the correct zfs layout, i think this is on the wiki somewhere

make sure you preserve permissions when you copy stuff over or you could have a very bad time of it

'copy it over' = copy the existing root FS over?

yes

I'd assume you'd have to change some bootloader config and stuff too? not too familiar with loader.conf

hm, yeah, would expect those might need some manual patching up

and yeah, preserver permissions and ownership and all those things

consult the cp/tar manpages I guess

for loader.conf you just need to load zfs.ko (if you didn't compile it into kernel), it should pick up the root fs correctly as long as bootfs is set

maybe I should do a very barebones installation to new drive, verify that it at least boots, and get back to intaller/rescue to clone things from existing root

mtll: in case you want to know, 64-bit version is installed and works just fine, haha, i don't know why i didnt try this in the first place, anyway, pkg install drm-kmod works now, great, i can keep this baby for years to come :D

I trying to install linux in bhyve va vm-bhyve show should I get so my virtual machine bootin from debian.iso file?

vm install mydebian debian.iso

yep

rtprio: but should loader="uefi" or?

it uses the template when you created it

if you want it to use uefi, update the template, or use a different template

# pkg clean

pkg: No package database installed. Nothing to do!

mine are all grub, i haven't been bothered to change it

...odd

nwe: perhaps you want to deploy open-stack debian image wiht vm-bhyve ?

it's seamless

rtprio: because after I have fetch debian-12.10.0-amd64-netinst.iso and create vm create myguest and trying to boot it looks it cant boot from iso

Im just entering uefi shell

no need to install anything, just provision with ssh keys

mzar: hush, for now

nwe: which template did you use to create myguest?

rtprio: pastebin.com/4Ha1rG72 found one example on internet :P

... uh

you know that vm create will use templates in /$VMDIR/.templates

the debian one, last i checked, uses grub

rtprio: yes but doesnt I need to create the template by myself and put it in VMDIR/.templates?

iirc when you run `vm init` it copies the templates packaged with it into the templates directory

you did do that, right?

yes but I only have one default.conf

maybe i took them from github.com/churchers/vm-bhyve/tree/master/sample-templates

i set mine up a long time ago and haven't thought about it since

rtprio: I will try with this template :)

another stupid question on this github.com/churchers/vm-bhyve/blob/…lates/freebsd-zvol.conf#L8C12-L8C18

does that mean zpool/dataset?

yeah, a zvol. just a chunk of space on the zpool, without a 'zfs' filesystem

rtprio: what do you suggest use zfs-vol or disk0.img ?

if you're using zfs, zfs-vol would have some advantages, easier/better snapshots

rtprio: another stupid question :) if my rc.conf looks like this vm_dir="zfs:tank/bhyve" so bhyve using zfs then I should just add what the new volume would be named in the template and it will be created?

you should really read the docs, but as i recall you set the things in rc.conf and then run `vm init` that will set it up

and vm create _should_ create the zvol but i haven't used it

will read the docs more :) and thanks for the help rtprio :)

nwe: you can create zvol, and just change disk0 in machine config file before installation

mzar: I will try it out, thats the best way to learn :) read and try things out :)

OK, have a splendid and playful install nwe

Cloudflare Zero Trust Tunnel FreeBSD can you share the document for the construction?

mzar: thanks :)

vxwarlock: wasn't aware freebsd was a supported platform

Thanks for answering rtprio

also why do you want cloudflare on your local network

A great loss

yeah, i disagree

rtprio > vps install freebsd > tunnel cloudflare and domain to tunnel interfaces

yeah, sure, but what problem does it solve

"Where it is not optimal or possible to install software agents, networks can also be connected to Cloudflare using existing network equipment, such as routers and network firewalls. To do this, organizations create IPsec or GRE tunnels that connect to Cloudflare's cloud-native Magic WAN ↗ service."

so at least this method looks feasible

To distribute tunnel services between VPSs | and maybe a little more security

z

Thanks mzar

well you don't run the router in a vps so that's not feasible

just use a wireguard tunnel like a normal person

vxwarlock: np, I just read this on developers.cloudflare.com/reference-architecture/architectures/sase

Snapdragon is ARMv7?

mzar > you read very fast;) I have to examine this document. Thank you for my sharing.

OpenZFS documentation says to use 32k recordsize for postgresql. Anyone with another opinion? (I'm using 32k at the moment because ot that documentation) openzfs.github.io/openzfs-docs/Perf…g/Workload%20Tuning.html#postgresql

why don't you trust the documentation Yaazkal ?

mzar I trust it, that's why I'm using 32k. Wonder what are the othe opinions, many blockes suggesting 8k and others 16k but no one mentioning that documentation hehe

s/blockes/blogs

hm.. I have 128k for postgres, it looks like neglected setup :-/

this presentation says 16k people.freebsd.org/%7Eseanc/postgre…7-postgresql_zfs_best_practices.pdf

so different opinions

but for MariaDB 16K is set

Hello. Are there any script like "localepurge" to be used in a FreeBSD environment?

you mean like "rm /var/db/locate.database" ?

rtprio: no, it's not about "find".

rtprio: it's about localization files like man pages, docs, etc

rtprio: directories like "du -hsc /usr/local/share/locale /usr/share/locale"

fuck my font, oops

no, none that i'm aware of

rodolfo: you can do it per user, software from ports usually respects locale(1), and plese take a look at login.conf(5), also notice ~/.login_conf file

manuals are not traslated, but kept up to date

/quit

lol