The power of FreeBSD's Metamode: pastebin.com/70UytHBt

Title: --- kernel.full ---linking kernel.fullctfmerge -L VERSION -g -o kernel.full - Pastebin.com

I couldn't believe this output when I saw it

lol

And I also forgot to share this since I was building both world and kernel together in order, do'h:

>>> World build completed on Wed Jul 26 19:30:49 PDT 2023

>>> World built in 28 seconds, ncpu: 32, make -j128

This is freakishly hilarious

Also, the reason I use that many make jobs is because I have the ram for it and didn't mind extra jobs taking up more memory for a bit of a building speed boost.

As they say: unused ram is wasted ram

kenrap: Wow what cpu is that and how much ram?

ted-ious: Ryzen 3950X and 128GB of ram

I would suggest going with the newer 5950X, since it has a better "cooling" efficiency cpu design

For AM4 socket motherboards

Oh?

They both claim to be 105w cpu's.

I guess the later the model the more efficient they can make it.

ted-ious: The design of the one I use heats up too easily. AMD fixed that problem in that newer model.

Ah.

hey

is there some way using pf, ipfw or other mechanism to exclude a certain source MAC address from bridging?

pf almost has it with route-to, but it doesn't allow filtering by MAC

ipfw doesn't have route-to however

ideally I'd send it to separate bridges depending on the MAC address

the underlying problem is that there are 3 MACs on the same interface, 2 of them have to be bridged as-is but one of them has to be NAT'ted

is operating on an interface well-defined while bridging is enabled?

maybe I could use the interface as-is even if it's bridged, while filtering out the MAC on the bridge

sthalik: let me get this straight.. you have multiple devices on one interface of a multihomed host, and you want the different devices you see to be attached to different homes?

ghoti, that is correct, due to moronic ISP

the host isn't multihomed, this is a TV router

but the TV router requests a public address from ISP's edge router

so I wrote a pseudo-dhcp server using /dev/bpf

ipfw will certainly be able to match traffic coming from a partical ethernet address. But I've never tried to make it pass along ethernet broadcasts.

*particular

it should be enough that ARP works

but I never handled member interfaces on bridges with their own subnets etc

it was always the bridge that had the addresses assigned, now it's both having different subnets

man i hate that i remember you

because that routr is fucked

Hi there, Need to execute script monthly, but at the last day of each month

didn't found a way to do that in crontab arguments

my solution is to execute script daily and test if current date is the last day of the month

here is my test command: [ `date -v+1m -v1d -v-1d "+%Y-%m-%d"` = `date "+%Y-%m-%d"` ] && echo "the last day of current month"

my question is there a better way to do that ?

nerozero: that's not pretty, but pretty good

to the current date, add one month, actually, no we want the first of that month… minus one day.

meena, first of next month - 1 day is the last day of current month

thanks

but will this work in december ?

nerozero: Set your date to December and try

meena, brutal :D

if i do zfs send my snapshot and after destroying my snapshot i want it back can i use zfs import or need to use zfs receive?? as my understanding zfs import is regarding the pool but snapshot is just a special dataset....am i correct?

nerozero: there's probably better ways to figure this out, but this seems like the quickest way to test

tyler82: where did you send the snapshot to?

meena: to an other partition

but same drive

anyone using dhcpcd ( either from pkg or port ) instead of base dhclient?

angry_vincent: still waiting for it to hit base

the same, but there is a pkg

i used to use dhcpcd everywhere ( before FreeBSD became my os )

angry_vincent I think crest does.

I used dhcpcd everywhere as well, but now I stick to base. Are we changing FreeBSD's dhclient to dhcpcd in 14?

i don't think so, only as an alternative

antranigv: not sure if we'll manage to do it in 14. The person working on the task seems … unresponsive: reviews.freebsd.org/D22012

Title: ⚙ D22012 Import dhcpcd(8) into FreeBSD base.

why would it change ? i can see this going the same way as removing some of the other base items. ftr im on keeping it in base just due to the fact its needed for most network connectivity but do not mind if its built as a package to be shipped with releases.

hard subject tho.

What's the benefit? ipv6 support?

CmdLnKid: dhclient's code is from ca 1997. The FreeBSD project hardened it with Capsicum, but that made it impossible to keep up with upstream. So we never got stuff like DHCP6 in base. If you need IPv6, you'd have to install it from ports.

ah. havent kept up with that

thx

now, the upstream has dried out, ISC DHCP is out of maintenance.

oh no kiddin wow

dhcpcd is under active maintenance, it has IPv6 support and capsicum hardening out of the box.

DHCPv6 is hardly ever needed, for what it's worth.

yes, but when it is, it's not there.

debdrup, yeah, but are we catering to everyone that may not have that knowledge or just the indepth

meena: it does have a bit of problems with privilege dropping until the most recent update, but now that that's addressed I don't know that there's any impediment for it to be imported into the base system.

CmdLnKid: we, by definition, cannot cater to everyone.

it's mostly cloud providers that do DHCPv6, instead of SLAAC.

debdrup, i know i know

Also, again, the plan _is_ to bring it into base - but it can't be brought into base until it does what it needs to do.

im up for either but just trying to understand the motivation as ive not kept up

It's not just enough for it to do DHCPv6, it also needs to drop privileges properly, like dhclient does.

looking for the pitfalls as well

minimizing the broad impact of new user troubles seems worthy when they may not be educated enough to care about the network side of things

At the cost of increasing the attack surface of every FreeBSD installation?

im installed, im connected

oh no i agree but what is a happy medium between secure and functional

while spurring new user interactivity

Sure, but I'd still argue that proper privilege dropping is a minimum viable product kind of featureset for something that will be used by almost everyone.

agreed

at least isc is front and formost among reporting and patching tho

shit i get "were going to patch" before i here any form of were updating here you go

very responsive to security related incidents

how we consume that may very well be a good key to the direction of this

angry_vincent, antranigv: i run dhcpcd on FreeBSD because it supports DHCPv6 prefix delegation

my only concern is the user thats going to replace me 20 years down the road and how well in depth they are at handling network operations

crest: is it just installing pkg/port and enable of rc script?

with short (<10 lines) hooks to feed the delegated prefixes into rtadvd as they're leased

angry_vincent: you can used it like that

and as it has it, ppl care less about configuring it rather than being able to interact with it

or you can use the dhclient rc.d script and tell it use a different dhclient path

it depends on what you want

because dhcpcd can work on multiple interfaces instead of one instance per interface

i do not want to use dhclient at all. only dhcpcd

i don't want to use either

i remember dhcpcd listen on all interfaces

static or dynamic

angry_vincent: it depends on the arguments you pass to dhcpcd

and what's really nice about dhcpcd is that it has full capsicum sandboxing support (not even a patch it's all upstream)

install dtpstree and run dtpstree -Uat if you want to see the multi process architecture in action

├─dhcpcd: [manager] [ip4] [ip6]

│ ├─dhcpcd: [privileged proxy]

│ │ └─dhcpcd: [BPF ARP] vtnet0 10.0.1.177

│ ├─dhcpcd: [network proxy]

│ └─dhcpcd: [control proxy]

so i want ipv4 only and listen on em0

just configure dhcpcd that way if you're already familiar with it

but the base system dhclient should also work out of the box for that

i familiar with dhcpcd on non-BSD. i suspect it is not different

it isn't unless you write your own hook scripts (e.g. calling ifconfig/netstat/route instead of the linux equivalents)

cool! hachyderm.io/@Freaky/110780930586491382

Title: Thomas Hurst: "Here's an rc script for #FreeBSD to apply (and re…" - Hachyderm.io

I installed FreeBSD 13.2 on my laptop via memstick.img. I opted for ZFS-on-root w/ encryption. However, I use a USB keyboard since my laptop keyboard is sort of messed up and as a result I can't enter the GELI passphrase

:( this really needs fixing

I planned to reinstall and either set an shorter password I can type in with that keyboard or remove it all together, but now when I try to boot from the usb drive it asks for the GELI passphrase? why?

PyR3X: does your USB keyboard work with your BIOS/UEFI?

meena: I believe so? I came from Debian and used it there when FDE was enabled

but why if booting off of the USB memstick.img does it still ask for the GELI passphrase?

crest: i should remove ifconfig_em0="DHCP" from /etc/rc.conf, stop dhclient, enable dhcpcd in /etc/rc.conf and start it?

meena: kind of upsetting that I released a fix for a problem I don't directly have before my VPS provider, who most certainly do

Freaky: 🤦🏻‍♀️

does anyone know why I can't boot memstick.img from a usb without a GELI passphrase? I can't even reinstall FreeBSD

i don't think so

PyR3X: hmm, just a hunch, you might want to try using `set hw.usb.usbhid.enable=1` in the boot loader's command line before booting into your laptop to see if you can get your USB keyboard working.

you certainly can destroy previous table and reinstall, why wouldn't you able?

I'm currently in NetBSD on another drive do I just 'dd if=/dev/zero of=/dev/rwd1c bs=1m count=1'? or how big is the table?

dd: /dev/rwd1c: Device busy

lol

oh duh needs to be rwd1d for the table

angry_vincent: unless you want two dhcp clients fighting on the same interface^^

crest: or just use `ps -d`

i prefer the dtpstree output format

and ps -d is subject to the usual ps selection rules which processes are included

ps -uaxd would be closer

crest: i now running dhcpcd from pkg instead dhclient. seems straightforward

as it should be

that moment when you try to install www/linux-opera and it fills up /dev/ada0s1a 992M 980M -68M 107% / :(

against the change. one is a server and the other ia just simply a client of any server. while both can do the job its flagrant to say by default that a client should be able to offer services as in network environments one rogue dhcp server can cause you a lot of pain

d'oh, /var/cache/pkg is 50GB

CmdLnKid: dhcpcd is *not* a DHCP server.

jimmiejaz: safe what you need if you need anything then pkg clean

ami mixing this up with dhcpd ?

shit

CmdLnKid: i don't know how your brain works 🤦🏻‍♀️

yeah, just reading the man page for pkg-clean now. I'm so used to use ports, but compiling anything on this craptop is ...not fun.

wrong emoji…

//ignore me

🤷🏻‍♀️ ← correct emoji

My font displays those. . . but they look like ass.

dhcpd = ISC DHCP Daemon

dhcpcd - DHCP Client Daemon

CrtxReavr: i use blobmoji for emoji

That's how you know you care a little too much about emojis.

(and Atkinson Hyperlegible as default sans serif font)

Does someone use iceWM?

Did

Do you know how can I put Thunar File manager directly on System rather than like this imgur.com/nUWrLMi.png

In short, I do not know. If the menu is auto generated, then I certainly do not. Else, have the menu be built from a file, in which case edit the file

it seems that you can add your menus on top of the predifined ones

I mainly had customized the colors, decorations, & perphaps key bindings. Never did much of menu reconfiguring

Number of packages to be fetched: 32737

The process will require 113 GiB more space.

_that_ is not the command arguments i was after

rtprio: pkg install --everything ?

pkg-fetch. trying to pre-download all packages that are installed

-a is obviously not it

pkg query '%n' | xargs pkg fetch # untested

that's exactly what i came up with

Thanks for testing

stupid question: should I use swap if I have 2T of RAM?

antranigv: swap is used for crashdumps, so, maybe?

meena good point, so 2TB of swap to fit the whole ram in? :))

no idea how to size that

The rule of thumb I remember is: twice the RAM until 4GB RAM, then the RAM. But I have no idea whether that's still valid with that much RAM.

I admit to not knowing about FreeBSD but on the, ahem, other systems, swap is needed for hibernation. Enough disk to store the RAM image.

I have a bit of an issue with my laptop with iwlwifi. When I am running the installer, I can connect to my wifi network through it, install stuff, and then finish the installation. But as soon as I reboot into the installed system, I cannot connect anymore.

Seems unlikely that someone with 2T of RAM would be using hibernation though. :-)

It says that the scan failed with a -5 error

There is also the corner case that exists right around memory resource exhaustion. Funny things happen there. Sometimes having just a little bit of swap will allow the kernel an escape path to swap out some pages and thereby have just enough memory to avoid running out.

zilti, An interesting problem. I myself don't know. But I would boot the installer and connect with it and then note down the network settings that are active at that time. Then repeat with the booted system. Hopefully there will be an obvious difference that would be a clue?

In the installer one can get a shell and then look at the currently active configuration.

rwp, that is an idea indeed, but I already am telling the installer to keep the configuration for the installed system - which makes me wonder, what could be different that does not get taken over?

Also still not knowing but if I did I would want to know by what method of configuration you are running wifi in the booted system? If you are using a /etc/rc.conf statically assigned IP, DHCP, or if using pkg wifimgr or other?

I'm using DHCP, and wpa_supplicant only so far

I am very new on using FreeBSD on desktop machines, I only used it on servers until now

All of my main systems are wired in but on a laptop which I have not played with for a while in 13.1R I am pretty sure I have tickled a bug interaction between wpa_supplicant and dhclient and if I can find some time I want to explore and poke at it.

Hmm I'll do some research on that, then. I am trying it out again right now and will see what I can find out

I want to avoid casting FUD here because I have not closed on it but it seemed that on my system wpa_supplicant running caused dhclient to wedge itself and get stuck. But I could manually ifconfig and set something appropriate avoiding dhclient and get it to work. It was so odd I feel it must be something I was doing locally that misconfigured things.

I've had stability problems with iwlwifi, and eventually switched to iwm. (Same hardware, different driver.)

13.1.

Also I hear many people are currently using the wifibox solution and very happy with it too. jrgsystems.com/posts/2022-04-20-802.11ac-on-freebsd-with-wifibox xyinn.org/md/freebsd/wifibox github.com/pgj/freebsd-wifibox

Title: JRG Systems - 802.11ac on FreeBSD with wifibox

FreeBSD is so good on servers (which are all wired in) that of course we want it everywhere but I fear that some things on the laptop are currently a little rough still such as WiFi. Many in this channel are running it that way just the same.

V_PauAmma_V: how do I switch to iwm?

Hold on. Let me look what I did.

Which reminds me: have you checked which driver binds to the device using "pciconf -l"? There should be a line starting with either "iwm0" or "iwlwifi0" in there.

Yes, there is a iwlwifi0@pci0:1:0:0

Which FreeBSD version are you running?

13.2-RELEASE

Hm, according to wiki.freebsd.org/WiFi/Iwlwifi my card is supported (Intel Wi-Fi 6 AX200)

Title: WiFi/Iwlwifi - FreeBSD Wiki

Does that still have the iwm driver? I didn't have to do anything for my interface to bind to iwm in 13.1, but maybe that changed in 13.2. Do you have a /boot/kernel/if_iwm.ko ?

I am able to run kldload if_iwm

Hmm. I may be in over my head here. (Which also points to possible problems for me when I upgrade to 13.2 in a week or so.)

I guess I somehow have to set if_iwlwifi on a blacklist to make FreeBSD use iwm

That's where I'm unsure of myself. In 13.1, I had to blocklist iwm to use iwlwifi. In 13.2 (or on your laptop anyway) the converse seems to be true.

Ahh found it, `module_blacklist` in /boot/loader.conf

But setting it to module_blacklist="if_iwlwifi" has no effect

Do you need the ".ko" extension?

That does not help either, unfortunately

Does it get loaded as a dependency?

vkarlsen: Where could I see that?

i was installed bhyve now and kubuntu launches with no issue. ☺️ but i want graphical interface as i want trying how is it working the graphical installation on bhyve...🤔 Any recomm?

zilti: Are you able to kldunload it?

vkarlsen: Yes, but it immediately loads again, seemingly because it re-detects the card

zilti: devmatch_blocklist= in /etc/rc.conf

devmatch is the program that loads modules for hardware after boot time

might be _blacklist depending on your freebsd version

RhodiumToad: that seems to do the trick! Although now it just does not use my card at all anymore. I guess iwm maybe does not support it...

Also it seems my issue has been fixed already, but not in any released version bugs.freebsd.org/bugzilla/show_bug.cgi?id=261075

Title: 261075 – iwlwifi: 8265 / 9650 problems

Well in any case thanks for all the effort, and I will continue looking into this later.

tyler82: i would use xrdp or something similar

tyler82: connecting to the vnc port on your bhyve host will run like shit

rtprio: i am trying now vncviewer 😔

rubbish 😞

connection refused error 61

i was destroying tap0 interface but should work...🤔

as i was also destroying vm switch

so i was add the switch again...and that should generate tap0 again...🤔

Or not??

wait, what?

you are connecting to the host, right, not the bhyve guest?

and why are you destroying interfaces?

rtprio: becauseci did try three different guide to install bhyve with graphical interface. and i did messed up

tyler82: are you using vm-bhyve?

it was working an hour ago but was very slow

that's what i told you, it's not going to work well. it's not designed for it.

rtprio: yes. i am using thst with vncviewer.

xrdp would probably be most performant, followed by running vncserver on the vm itself

but vm init is so slow...🤔

`vm init` like the command that sets bhyve up initially?

after reboot i cant start vm init. as giving error not configured or not valid directory.

rtprio: yes. actually is "turn on" bhyve

my vms boot in a few seconds

what do you have in rc.conf for vm_bhyve?

there is two way to add the kernelmodule vmm. one is to add rc.conf or boot loader.conf

but not both!!

you shouldn't need to as it's loaded on the first bhyve command, but yes, you can in loader.conf

what's `grep vm_ /etc/rc.conf` ?

i have an assumption why so slow...because vm_dir declared on an other partition. ext4 . so probably need to wait until fuse module will make it visible for bhyve...🤔

or no cluev🙄

yikes. ok, interesting decision, but yes, i suppose that could be partly the reason.

i would make sure both of those are tucked in to loader.conf. the ext4 /vm disk, is it in fstab and mounted on boot?

rtprio: because the isos and images are taking lot of space and my zfs partition is not too big. so space saving. 😁

rtprio: no. not in the fstab....

good idea 😆 i will add it

is this a laptop or desktop system? or something that sits in the closet

laptop

ok, so you need to mount -t ext4 the thing, then `vm startall` and you're saying that takes too much time

i have also used google remote desktop which is weird but seems to work ok

rtprio: ah. 🤭 not even ext4. it is NTFS partition. 😆

so i have to edit fastab.. ntfs-3g or something....at least on linux. i cant remember

yes, the fstab entry for ntfs is a bit weird

much better 😃

i mean much quicker. vm init reaction. 😁 thanks rtprio 👌

but tap0 interface still no clue how to fix. shall i create it manually?

let's back up

did you configure networking on this vm?

and do you have a switch created in vm-bhyve

if i type ifconfig the vm-public switch is there. which is fine. but once the guest boot up automatically should generate tap0 and assign to the vm-public switch

is the guest properly assigned to that switch ?

yep

and it doesn't create tap0 ?

or tapX

no

so you have `network0_switch="public"

` in your guest.conf file right?

and did you start the vm service, automatically?

i think because i was deleting it. it was not easy to delete...needed to force with destroy -f flag...🤔

deleting what exaclty?

rtprio: the tap0 interface. ifconfig destroy -f tap0 or similar the command. i cant remember.

so restart the vm. `vm stop myguest` `vm start myguest` is it recreated?

but i have snapshot so i might be zfs rollback... 🙄

you can't rollback destroying an interface

rtprio: i did delete the guest and recreated mamy times already.

rtprio: wait! what? why not?? 😵‍💫

i have no idea how you came to a conclusion of "i should delete this interface" when observing "i can't connect to the interface." that's literally the opposite course of action

you could vm create a new one and mv the hard disk over; should save you a reinstlal