is there some trick to getting tigervnc with enlightenment over ssh?

realdeimos: what problem do you encounter ?

there shouldn't be much to it. set up ~/.vnc/ with config and xstartup files. call vncserver $DISPLAY , and check the log that pops up in ~/.vnc

I eventually figured out the steps to get it working over ssh

thanks for the pointers, would've helped

can i somehow mark a service to start up after another one (or just late like with filesystems)?

still have the issue where my jails aren't reachable unless i do pf reload and no fix in sight… :F

rwp: pkg repos change atomically from build to build, so the only way for packages of existing ports to disapear, is for them to fail to build, that could be some temporary failure in which case the next build could have the package again or some real problem that has to be fixed first (well, it could be some dependency failing and the port in question beeing skipped too)

rwp & rtprio: I did "pkg upgrade -f", "freebsd-update fetch", "pkg-static upgrade -f pkg", "freebsd-update install", but I still get "ld-elf.so.1: Shared object "libcrypto.so.111" not found, required by "php-fpm"" and the only libcrypto.so version in /lib is still "libcrypto.so.30". Any ideas?

andreas303: you have to reinstall all your packages (freebsd-update should have told you that the second time you ran it), libcrypto.so.111 is the old version (openssl 1.1.1) which was removed on the update as 14.x uses openssl 3.0

mhh, i have a mirror that's not being recognized by gmirror. doesn't show up in gmirror list or /dev/mirror, but the component devices are available in /dev/gpt – how do i rectify this? do i just do gmirror label again or will that destroy existing data?

nvm, different setup. this is a zpool. m)

nimaje: Hmm, but I have actually already reinstalled all packages (with "pkg upgrade -f") and rebooted. I will try again.

hm, does the packagename of php-fpm contain the php version? maybe an unfortunate timing of you updating freebsd and phps default version changing

ah, no php-fpm is part of the main php package and all php versions should get build if I understand that correctly

nimaje: Ah, yes, it seems so, because when I do "pkg search php-fpm", nothing shows up.

nimaje: Now I have done a "pkg upgrade -f", but I still get "ld-elf.so.1: Shared object "libcrypto.so.111" not found, required by "php-fpm"". :(

nimaje: When I do "pkg info | grep php" I get a list of php80 packages and "php80-8.0.26".

that one doesn't exist anymore, so that explains why pkg upgrade -f didn't help, switch to a non-eol version of php

trying to setup a vpn-client using l2tp - any pointers on where to look for more information

hi there

I've been following linode.com/docs/guides/install-freebsd-on-linode, but I'm not been able to properly start the installer after booting and using Glish. I'm at this step. I see the hello screen of the installer and I press enter to boot into multi-user mode, but it seems the system freezes. Has anyone been able to install FreeBSD on Linode lately? I tried FreeBSD 14.0 RELEASE first. (Now I'm trying version 12 because

the guide seems to use version 12.0 RELEASE.)

Here's what I get with version 12: pasteboard.co/CkimcYdGEveW.png

Title: Pasteboard - Uploaded Image

I'm gonna try version 13, but I'm losing hope.

IRC is dead. :) Since nobody is ever really online anymore, we should all switch to the USENET.

why not bulletin boards?

dial-in only

That's in fact a great idea. But I don't have a MODEM anymore.

i dont have a phone line

I actually have a phone line and don't even use it.

I'm not aware of any BBSs in my region either.

But the USENET is going great.

Big Black...??

Specially great after Google Groups left it alone.

Big Black?

Don't know what that is.

albini rules

Albini rules? Don't know what that is.

were you cool in the 90s?

I guess not.

you'd have had to be uber cool in the 80s to catch them

I suppose.

Were you uber cool in the 80s?

nay

I was learning to read in the 80s.

i never bothered

hold me closer, tiny dansa

:) beautiful song

not cool toh

tho

:(

it's not cool?

why not?

i dont think any cool was ever "knighted"

*anyone

i suppose paul mcartney had his time in the sun..

but we've never heard anyone say "sir ozzy osborne"

yeah

well said

but i think paul mccarthy and all the beatles are overrated

music is overrated

john lennon is so overrated

the whole thing is overrated

in fact, this conversation is over

:)

i gotta install a freebsd instance at a cloud

thanks for chatting

and for keeping IRC alive

see you in the USENET!

noneoftheabove: i'm not sure the guy who founded Wings could ever be considered 'cool' even at the time

lw: the tv show?

Non-ICEhttps://en.wikipedia.org/wiki/Paul_McCartney_and_Wings

Title: Paul McCartney and Wings - Wikipedia

er, noneoftheabove: ^

ok, i rescind my statement

or rather, reinforce my position that no-one cool has ever been knighted

:-]

elton is likely the coolest

apparently Mick Jagger got one?

and Rod Stewart

f those guys both

how come?

stones are disgusting sub humans

and stewart is kinda gross but i havent heard anything particularly nasty about him

Hi, I have a mini pc I use as a file server. I am interested in getting a second disk to create a zfs mirrored raid. It's for a home use case, would a QLC nvme suffice?

I guess I wouldn't really notice the latency issues if there are any to be honest, the price is good for what it is, I think.

uh-oh - somebody is not happy with my maintaining skills

What is a QLC NVME?

it's lower quality flash memory as far as I understand it

it's not necessarily quality, it's robustness.

it's about how many bits you pack into a given volume of space

ah, is that why they're "slower" ?

yes. and most notably less reliable

QLC uses 16 instead of 8 voltage levels

so the reliability is mostly what I am asking about

less room for error

okay, so not worth it to buy qlc

it is my data after all, even if it is just for home stuff

QLC are inferior in pretty much any metric except for cost per bit

SLC > MLC > TLC > QLC :D

Hi, I need to compile a program that needs gcc14. I have installed it, but as the MakeFile of that program calls gcc it uses gcc 12 instead of 14. How can I set gcc14 as the default gcc?

CC=gcc14

jbo: thanks!

did beastwick quit

sorry, client dropped, I am back

QLC is not worse, it's just the usual price/performance trade off

oh hi

beastwick: QLC is fine, just check the spec sheet, especially the IOPS and TBW (or DWPD) figures

it's worse if you care about reliability (and speed, but whatever on that one)

that's what you want to care about, not the specific technology

jbo: it's not worse. drives don't fail more often because they're QLC. they might offer lower performance, or lower endurance, but that's why the mfr published TBW and IOPS

any reputable mfr (seagate, WD, samsung, micron) will published accurate iops and tbw/dwpd for their disks

at least at the low end, availability of DRAM cache on the disk makes *far* more impact on performance than SLC/TLC/QLC does

this disk has 220TB (TBW) and 5,000 MB/s read 3,600 MB/s write

many low-end drives have no DRAM at all which means sustained write performance can be much lower than even an HDD

Title: Crucial P3 Plus 1TB PCIe M.2 2280 SSD | CT1000P3PSSD8 | Crucial.com

unless I misunderstood beastwick didn't care too much about performance in this discussion

yeah, more concerned about reliability

I know this is all cheaper consumer grade stuff, but still want to know if I should just pay for the $90-$120 disk instead

beastwick: 220TBW for a 1TB disk sounds fine for home use to me. of course it depends on your specific usage. but if you write 100GB per day, the disk will fail after 6 years.

lw yeah that's not happening

it's just mostly a file/backup server with low per day transactions going on

... and 100GB per day is pretty high for home usage

lw, btw, I decided to go for TR 7960x. mainboard & RAM arrived today

and Crucial (Micron) is a well-known brand, with a good reputation, so i would trust their figures

jbo: exciting!

jbo: is this going to be your desktop or a server?

lw, desktop

it's going to be my first AMD experience so I didn't want to risk going server

only 24 cores, surprised you didn't go for more

the 32 cores would cost 1k more :/

ouch

yeah

7960x seems to be very, very good price/value

how much memory are you going to add?

128GB

my current rig has 64GB and that is getting right at times

single-socket MB i guess? (i don't know if TR can dual socket at all)

yeah, single socket

Asus TRX50-SAGE

it doesn't nearly have as many PCI lanes as the w7-34xx but... costs difference is HUGE

so I'll be giving AMD a chance :)

would have liked more RAM but... again, cost :D

huh 4.2GHz base clock on that, that's not bad. i think that's higher than the boost clock on my 5800X3D... i wonder how they do that

larger die means larger heat spreader i guess

lw, yeah, I looked at it long and hart. I really think that TR 7960x is an excellent option for "us"

jbo: yeah, i was mostly looking at EPYC because of the future expandability (with 2-socket MBs) but maybe i should do TR instead

haven't ordered the cooler yet

Title: Watercool Heatkiller IV PRO Threadripper-Kühler Kupfer Nickel, 149,95 €

that's the current plan of action

i'll add my usual rec for Noctua, but i don't know if they do TR coolers

oh you're doing water cooling

yeah, no chance I'm going to throw 350W at a noctua

I have noctua in ALL my PCs. I adore them. but nope.

I plan to OC this rig so 400W...

ah missed the TDP. that explains the high base clock!

:)

the NHD-15 I currently have is already struggling with my overclocked i7-8086K

(I won the silicon lottery on that one)

jbo: in the mean time i've moved my -current builder to a VM on my M1 Mac ... the performance is really pretty impressive considering the hardware, and i'm pretty sure this will save a lot of money on AC power

(it was previously on a Ryzen 2700X)

lw, nice! I don't think that I'll have the luxury of considering AC power after this rig :s

what

if you don't have AC power what are you going to plug it into

I mean in terms of power consumption, sry

ah

well hopefully it won't use much power at idle, AMD seem good at that

don't know what to do for GPU yet

never had a non-quadro workstation and I like that nvidia provides native freebsd drivers

but the current selection is... meh

I have currently a Quadro P5000 which is 7 years old. I am considering just putting that into the new TR rig

jbo: assuming you're not a hardcore gamer, i recommend AMD GPUs of older 1-2 generations, so e.g. RX 6xxxx

well supported by amdgpu and good price/performance

lw, AMD doesn't provide native freebsd drivers, right? it's linux KPI crappery?

correct, but it runs Factorio with no problems

lol

FFF!

my desktop has an RX 6800 XT, i've been using it with Wayland (sway) since November and had no problems

I'm an X person (because bspwm)

i understand leaning towards nvidia on freebsd though given the native drivers... but do you remember the EGLStreams nonsense?

lw, I "remember" it: forums.freebsd.org/threads/nvidia-current-status.81980

Title: Solved - Nvidia - Current status | The FreeBSD Forums

lw, not sure if that changed to. I just decided not to go for wayland :p

what's the currently preffered IPSec/IKEv2 implementation for FreeBSD? still strongSWAN?

lw: i use strongswan, but i think debdrup had ipsec working without it

dstolfa: with static keys or something else?

i know you don't need an IKE implementation for static keys but i think that's generally considered insecure nowadays

lw: i don't know what debdrup uses, i just vaguely recall it. i only use ipsec via strongswan to connect to the university vpn

hmm, i updated a RPi from 15.0 to 15.0 but two commits later and for some reason it's not booting

i would definitely say strongswan

yeah, you don't wanna settle with weakgoose

sounds like a good reason to stick with -RELEASE if possible

lw, the wonderful world of -CURRENT? :D

jbo, paulf: i don't see how this can be a -current issue since the two commits in between updates had nothing that could possibly affect booting. i suspect it's more of a generic freebsd-on-rpi issue

time to grab the microHDMI cable and see what's going on

ok, found the problem

[8!] witch /mnt/boot# ls kernel

[9!] witch /mnt/boot#

i think this is a pkgbase fuckup

kernel-less as a service!

or, I guess in this case: kernel as a service? :D

boot-lite

jbo: nowadays: "you didn't pay for the kernel, what do you expect?"

but i think the actual reason is in my pkgbase build i set KERNCONF to "GENERIC-NODEBUG GENERIC LFVM", then i updated this system, and i guess somehow this ended up with the default kernel being not installed

lw, btw. I figured out why my windows11 bhyve VM suddenly gets stuck during boot consistently - but I don't know how to fix it

probably because it had FreeBSD-kernel-generic installed, which installs in /boot/kernel.GENERAL, and it did not have FreeBSD-kernel-generic-nodebug installed, which would install in /boot/kernel

s/GENERAL/GENERIC/

lw, when I don't pass-thru that PCIe USB controller card, it boost successfully every single time. when I have the passthru enabled, it gets stuck very early on (just after the windows bootloader is invoked). I have no idea why. It booed fine with the passthru for several days. and I really need that USB-Controller in the VM :D

jbo: i don't know why you persist in this weird config like passthrough of USB PCIe adapters, it seems like you're just inviting problems

lw, how else am I going to get a USB device into the bhyve guest?

ok, now this might be a 15.0 issue, the system just hangs halfway through boot le-fay.org/tmp/30d/9fE0NG.jpeg

Hi, can someone help me building an c++ 14 program? I'm getting: error: no template named 'tuple_element_t' in namespace 'std' and I don't know how to fix it.

BTW, is not my program, but github.com/elixir-nx/xla.git

Title: GitHub - elixir-nx/xla: Pre-compiled XLA extension

martinrame: did you try c++ -std=c++14?

lw: I cannot call c++ directly, this uses a MakeFile.

So far I'm building it with: std=c++14 CC=gcc14 CPP=gcc14 ROOT_DIR="$(pwd)" BUILD_ARCHIVE="~/xla_extension.tar.gz" make

But I'm not sure those are the correct env vars.

the std isn't

try CXXOPTS="-std=c++14" (or gnu14 maybe)

paulf: thanks, let's try that

Mm, same error

thougn if you are using BSD make I don't know that at all well so I could be wrong

martinrame: you probably need to ask the software vendor about this

lw: yes, I asked, but as they work on Linux they can't help: elixir-nx/xla #78

Title: Cannot build on FreeBSD 14 x86_64 · Issue #78 · elixir-nx/xla · GitHub

78 – xdvi package is empty bugs.freebsd.org/bugzilla/show_bug.cgi?id=78

martinrame: okay... how are you invoking the build? are you using the provided Makefile? could you please paste a full log of the build attempt?

lw: I'm invoking as this: CXXOPTS="-std=c++14" CC=gcc14 CPP=gcc14 ROOT_DIR="$(pwd)" BUILD_ARCHIVE="~/xla_extension.tar.gz" make

lw: pastebin.com/RS7JSKkw

Title: ERROR: /usr/home/informemedico/.cache/bazel/_bazel_informemedico/1fd8e3102f29134 - Pastebin.com

This is the output

you need to work out how the build system works - it looke like basel, and I see -std=c++0x

oof... I remember the last time I had to Bazel :S

I only ever tried to use bazel once all I got was java exceptions

what do folks use for service management in freebsd? like, auto-restarting things that have failed, stopping auto-restarting if that isn't working

% man service

> auto ?

paulf: sad but true

jaredj, init/service/rc

jaredj: I guess the manual page doesn't do it in 14-R. I think what you're looking for is 'service foo enable'

jaredj: no real standard solution for that, but you could look at daemon(8) which has support for auto-restarting services on failure

iirc daemon can do that auto-restart stuff

crude example: blog.insane.engineer/post/freebsd_daemonize

Title: Daemonizing on FreeBSD

in particular i have a matrix server, conduit, running inside a jail that's using iocage. apparently there's something someone can say to it that will cause it to panic with a message "not implemented yet" and die

:D :D :D

it's been a good little server other than this - i guess synapse takes a lot of ram, but conduit hasn't - i only have 1GB of RAM on this vultr vm and it's been pretty much working

thanks for pointers! reading up

if you installed the Matrix server from ports, the rc.d script probably uses daemon(8) already so it might just be a matter of adding some arguments

> -r, --restart

um ok, that seems easy

andreas303, /lib/libcrypto.so.111 is in /lib and part of base. Earlier you said you only had /lib/libcrypto.so.30 there. That's telling me that your FreeBSD base is not updated. That would be a problem with the base upgrade not the pkgs upgrade.

andreas303, I would double check that your base version is "freebsd-version -u" reports 13.3-RELEASE (I am still on 13.2-RELEASE-p10 but freebsd.org/releases/13.3R/relnotes is out now). If not then that is the problem.

Title: FreeBSD 13.3-RELEASE Release Notes | The FreeBSD Project

andreas303, You would need to do "freebsd-update -r 13.3-RELEASE upgrade" (or replace 13.3-RELEASE with your next release version, 14.0-RELEASE maybe) to fetch it and then "freebsd-update install" to install it. Then reboot. Then follow up with "freebsd-update install" to finish the base upgrade. Then verify that you have upgraded with freebsd-version.

That will bring in the new /lib/libcrypto.so.* version that you are reporting as currently missing from your system. And then the precompiled pkgs will work.

rwp: andreas303 updated to 14.0, so openssl updated from 1.1.1 to 3.0, that is why the .so.111 is the old version and the .so.30 is the new version

andreas303, Searching further into the scrollback I see you mentioned 14 not 13 so I should have said 14.0-RELEASE there. But I still think the problem is an incomplete base upgrade. Since you reported that you only had /lib/libcrypto.so.30 there.

and the incomplete update is thatphp 8.0 is eol and was removed from the ports tree, so pkg upgrade -f couldn't install a version for the new abi and the old package was still used

lw: first listed kernel in KERNCONF is the default kernel

rwp & nimaje: Oh, OK. Hmm, so what would you recommend me doing to get back on track with my failed upgrade? Should I remove all php80 packages and try to install php81/php82/php83 packages instead? The /etc/os-release shows FreeBSD 14, so that made me believe that the upgrade was complete and finished.

rwp & nimaje: I'm a slight newbie when it comes to solving unexpected FreeBSD problems. I'm still learning. :-]

andreas303: yes reinstall that stuff for a non-eol version of php, as I already told you at 11:35 UTC

nimaje: OK, I'll try. Thx!

Are their any free beginner level OS/kernel design courses (video preferred) that center around FreeBSD? By "beginner" I mean, you already know C, know the how to use/operate freebsd and it's internals, but don't know much about the code/algorithms behind it.

that's not "beginner" anymore :P

Well, begnner in the sense that I know nothing about OS design :)

TIL: freebsd's kernel has a function named 'hammer_time()' which I found kinda funny

pages.cs.wisc.edu/~remzi/OSTEP is good, but it is about unix-like OSs in general, not freebsd

Title: Operating Systems: Three Easy Pieces

is there a smarter way to search mailing lists?

Didn't M. Lucas write a book about FreeBSD design? (Not video, but that's what jumps to my mind.)

The Desigan and Implementation of the FreeBSD Operating System?

That rings a bell.

that one is by Marshall McKusick and George Neville-Neil, sadly I haven't read a FreeBSD book yet

I have installed FreeBSD 14 on my ThinkPad X280. Everything is working very well except HDMI. It works if I boot the machine with the cable plugged. But not if I hotplug it. I have tried both with the modesetting-driver and with the XF86-video-intel driver. This seems to be a problem only in FreeBSD. It works very well in OpenBSD. The card is an Intel UHD 620