are there any plans to make syslogd speak tls like OpenBSD's does?

anyone managed to get github.com/wireapp/wire-desktop working in freebsd? via deb or appimage or voodoo incantations?

Node.js >= 10, Npm, Git, and npm install --global yarn foobar looks like

Prerequisites

github.com/wireapp/wire-desktop/wiki/Troubleshooting 32bit app perhaps.. i686

wiki.freebsd.org/Linuxulator perhaps

on fbsd

i see electron30-30.1.1_1 in ports

Build cross-platform desktop apps with JavaScript, HTML, and CSS

yeah its 32bit looking at troubleshooting page apt install g++-multilibs

32-bit binary on a 64-bit system

hi, I have identified a bug introduced between 12.0 and 12.1 release. I'd now like to do a bisection search using VMs. What would be the best way to do this? master branch or releng? and which changeset to start at?

dead jim! Production: 13.2, 13.3, 14.0, 14.1

why even waste the effort

some old error is dead version..

is the error still in current release?

YUP IT IIIIIIIIS

Qual, rennj: try freebsd-version -kru there are three versions that can be diffrent, the installed kernel, the running kernel and the userland, if some errata doesn't affect the kernel, no need to rebuild it to change some version id, so kernel and userland can be diffrent, and if installed kernel and running kernel are diffrent you should reboot

yeah that's for Qual , nimaje

so it probably got introduced on current and then cherry-picked to 12-stable which later got branched to 12.1-releng, I would try 12-stable branch between when 12.0 and 12.1 got branched off as it likely contains fewer commits, ultramage

ohh yea stable would have continuity across those points, that was what I was looking for (I was going to go with 'main' based on date)

... I don't know how to tell at what point a branch started, using github's ui

oh I think I found it - the '220 commits ahead of X' page sorts from oldest and seems to start at the branching point

wiki.freebsd.org/LinuxApps heh

back in the late 90's freebsd,linux,solaris was more compatible, so far off the rails now.

deimosBSD, just run a bhyve vm

"were more compatible"?

hah less

it's lame

it was a better time and software drift wasn't as bad

i use to compile same stuff across multiple os's that isn't possible today

past 24 years embrace the suck..

peak 2000

IMHO

back than some stuff were diffrent between systems and now too; if you want to program something for diffrent systems, you have to stick to what is the same between them or add conditionals for those diffrences

back then netscape/firefox i had on all 3 platforms..what you got now?

beos had firefox

so fbsd,solaris,linux,beos same firefox

where we at now?

i dont know what you babbling about same between them

you mean posix perhaps

hp-ux firefox yeah doubt

ftp.nluug.nl/os/HPUX/itrc/downloads.html how lame hp-ux is at this point

old ass compiler

i macromedia flash/adobe flash on all platforms..even beos up to 4.0 or so

java write once, run anywhere...seems better at this point

seems like beos is widely diffrent from some posix/x11 system, so firefox probably had a lot of code exclusively so that it could run on beos

firefox-2.0.0.18

everywhere where you have some jvm

firefox for hp-ux, solaris is like firefox-57

dead jim!

esr-57 w00t is worthless

Where there's a will, there's a way. Everything is possible if people are motivated enough. If the devs don't care about platforms like hp-ux or solaris these days, they will not continue to maintain support

software is less compatible today then in the 80's and 90's

run a virtual machine, you got the cpu/ram/disk power

Well, that's because a changing hardware and software landscape requires changing APIs as well, you're not a good engineer if you consider "tradition" to be the primary deciding factor as to how you'll design your system

nothing to do with tradition, pointing out a fact

compiling software across multiple isa's and os's was better

fuck tradition

Well, in order to be able to compile them, you need someone to sit down, implement support, and test it, it's not a magical switch, and it wasn't in the past either, there's a reason autoconf is such a complicated beast

also, in the 80s, when it comes to the home micros, they each had completely incompatible operating systems and everything; from different flavors to basic to completely different GUI OSes

*of basic

Compatibility is arguably much better these days, because we have well defined platforms, and we also have well-defined middle-ware on top of those platforms. And if something is missing for your specific OS, well, time to get your hands dirty and implement the support

the ports tree got its name because it was work porting software to other OSes, those file/ directories probably had more patches in the past than they contain now

they should be reviewed and/or be well documented. there's a patch for php that disables inclusion of libpthread. It was added pre-2000 (original commit not available) when threading was 'new' and broke everything. It is still there because nobody knows what the patches are for anymore.

unrelated, but any idea how to fix 'Type clipboard text' function in hyper-v UI when running a frebsd VM? It does nothing. I assume it's supposed to work even without integration drivers. Could it be some kind of terinal driver issue?

apparently linux users also have had this issue since forever. could not find anything explaining the cause or how to fix it, shame since being able to paste chunks of text would be useful while setting up

/quit

setup seems to be stable enough. i had to do some script magic to set up socat and getty for vt100,and disable the extra colorful zsh script

but i can chat from my c128 now

there are some issues though. i dont think irssi or screen or someone likes to be in such a shortterminal

for some reason irssi's status line is in the wrong place. but hey. yay freebsd

hjf: I made my IRC window with weechat very narrow and it surprisingly worked well (with screen)

but yay for c128

yes its weird

hjf: How do you connect? Serial?

the status line appears at the bottom instead of the previous row

maybe some terminal definition issue

vkarlsen: with a cartridge that connects to ethernet and emulates a modem . atdthost.example.com:1234 opens the socket

hjf: Cool!

theres a cheap cart that does this over wifi too

what is your terminal emulator?

DESTerm 3.02

set to vt102 and i did export TERM=vt102 in the shelltoo

and what is the matchin TERM ?

got it

stty -a reports correct # of rows/columns, if any?

let me check... screen Ctrl-A-f does reflow the screen properly fwiw

but maybe there is off-by-one somewhere or some "hard status line" or whatever

I think I should try your setup in the emulator before I order Mega65 :)

saper: hm says 38400 baud0rows 0cols

this thing has 80 cols and 23 rows

it can have more rows but the monitor ihave doesnt work with the non standard mode c128 outputs

uh-oh, vice didn't buiid for -CURRENT amd64

hjf: so the terminal does not report back the # of rows and columns, not good

but how does it reflow with screen?

no idea :)

I am not a terminfo wizard unfortunately

maybe some custom TERM entry with decterm|co#80:li#23:tc=vt102 would do the trick

Hi, FreeBSD uses OpenPAM. I'm trying to switch GNU Guix from Linux-PAM to OpenPAM. Is the latter missing modules, such as pam_env.so or pam_limits.so?

hjf: does it react to printf "\e[18t" ? this sequence can generate <esc> "[" "8" h ";" w "t" back

saper: didnt do anything

buttt

doing "stty rows 23 cols 80" fixed it wooo

wish i could set this without a dot script

hjf: yeah, try a custom /etc/termcap entry like the one above (not tested at all)

then set TERM to "decterm" and it might work.

hjf: sadly, not much seems to happen after DLOAD"DESBOOT",U8 and RUN under x128, black screen :(

saper: run"*" ?

tried adding the termcap and rebuilding the db but it says it's not found

hjf: needed to start in 80 column mode, stupid me

I was going to do a bisection search between 12.0 and 12.1, but... the codebase is f'ed up. completely f'ed up. won't build. not self-hosting yet. multiple blatant programming and makefile mistakes. goddamn.

in hindsight it's now obvious... if people couldn't get a working build then they didn't test their changes, and code review can't catch runtime errors, so of course this environment would result in subtle breakage

I am reading FreeBSD mastery storage + ZFS

So the recommended path to install FreeBSD is not to use the server's RAID card (I bought a Dell R320 for testing, it comes with a H710 raid controller) but to set up the disks individually as RAID0 and then implement the raid using ZFS?

uskerine: as JBOD - just the bunch of disks

do _not_ use an underlying raid0

I have an ancient 3ware controller and it could be switched from RAID to JBOD

saper: so how did it go with the commodore terminal

uskerine: I've heard there are problems with Dell RAID controllers and JBOD, and here is one Internet search hit: forums.servethehome.com/index.php?t…10p-mini-lsi-2208.22019/post-204927

i had to 3rd-party-firmware patch an LSI a long time ago to make JBOD work

hjf: decterm does not seem to detect Swiftlink provided by vice. I think I have also hit a bug when vice tries to talk to real devices such as /dev/nmdm0A provided by FreeBSD

did you set it to DE00 NMI?thats how i had to set up my cartridge

sourceforge.net/p/vice-emu/bugs/1876 pretty sure there must a bug somewhere since -16385 which looks like some hex address instead of the vice's internal fd number (natural number)

hjf: cool wanted to ask your settings :) I try with userport at @dd00 at 2400bps but that fails with the /dev/nmdm0A problem, will check DE00

does the mega65 emulate a c128?

ooo you can run misterfpga cores in it

hjf: sadly decterm insists on UserPort @dd00 despite me saying "x128 -userport 0" (no device)

cool. i have a proper mister for gaming connected to a crt

hjf: yeah it's basically a very expensive FPGA device

excuse me, I'm not familiar, what is wrong with using hardware raid with freebsd?

800 eur omg

just get a real c128 and a wifi modem for it.

or a rs232 adapter

for 80col output you can use a GBS-8200 cga to vga adapter

ultramage: there is nothing "wrong", but that's not what ZFS likes to use. Sure you can install on it.

real, enterprise hardware raid in a supported configuration is nice

just hot swap the disk,the controller takes care of everything

but just use ufs if you're gonna use that

if using "Enterprise" RAID it's good to have something that has control utility on FreeBSD, to verify the RAID status, the disk status, is it rebuilding the RAID right now and what not

ultramage: use the hardware raid, or use jbod+zfs; but don't put raids over raids. this is the road to data loss unless you know exactly what you are doing

ah okay, zfs

I read about the need to do a custom firware patch on the H710

hjf: no idea how to convince DESTerm "there is nothing at the user port really", do you have RTC?

which I am not happy about it. But it is also my understanding that if you do a RAID0 will the entire disk, it shall work. What is the difference with JBOD?

uskerine: JBOD the whole thing acts just like a controller. you will end up with n disks in fbsd. if you do a raid0 you will end up with one disk in fbsd and any failing disk will be end of your installation

saper: try desterm 2. i read it doesnt try to be as smart and it will use what you tell it to. desterm 3 tries to be too smart

ridcully: why "end of your installation"?

I am asking the same, why the end of your installation

the end of evangelion

I have not read in full the "FreeBSD mastery" booklets, but from what I understood you still can use that raid0 disk in ZFS, partition into GPTs and use those GPTs as providers to implement a RAID in ZFS

I've been using zfs on my Dell T320, happily ignoring the PERF controller for nearly tenyears.

raid0 is borked, once a single disk fails?

mariuss and how is the PERF controller ignored? which PERF controller do you have?

i mean, i never have used it. maybe with more disks the controller can salvage it?

It's in the BIOS/EFI setup

ridcully: no,raid zero (striping) splits the data in equal size chunks

I am listening mariuss, so you can disable the PERF controller and the disks will be still recognised? I can also physically remove the PERF H710, it is a module you can actually remove from the server

Yes.

like data 12345678 will be stored as 1357 in disk 1 and 2468 in disk 2

there is no recovery if one disk dies

I didn't physically remove the controller.

that is what i am assuming

in ZFS what I understood is that the RAID is implemented in software, so theoretically you could use one disk / RAID0 (so technically you are using the hardware RAIDs as providers). Can you partition that into GPTs and then use the GPTs to implement the RAID distributed among several disks with ZFS?

Also, is there any case for servers where it is recommended to use the hardware RAID controller and not using UFS? I have read about the cases for ZFS but I am wondering if it is worth the effort

and using UFS*

uskerine: if you are doing any "hardware"-raid, the OS will see that as single disks. you can partition etc as you like

(or as many disks as you configure in the controller)

ridcully you can or you can not?

because if you can then I see no problem in having one RAID0 per disk taking the entire disk. So you have 6 disks your hardware raid implements 6 raids0, one per disk.

yes, you can

then you partition each disk (raid) and then consume that in ZFS

uskerine: with raid0 if one of the disks fails, it is al ldead

dead

if your controller can do raid0 with single disks each, then this is their way of providing a JBOD

the disk is dead, but you ahve another raid implemented in ZFS that uses two disks

ridcully that is hte workaround I have read about the PERF H710, which seems to be a common "issue" with the Dells (which lack JBOD/pass through without using a firmware downloaded from who knows where)

And does everyone use ZFS? Nobody uses the actual RAID HW controller and UFS ?

i use the raid controller on my r710

rtprio and do you use UFS?

yes

as long as you don't use more than one disk per raid0, that is a workaround. but be _very_ careful to not mess this up. the lowest layer with raid0 is the road to data loss

Can you share the reasons for such setup and not ZFS?

because i didnt want to reflash the firmware for jbod when i set it up

I am interested because everybody in the internet favours ZFS but I read mostly technical features

and mfiutil has nice controls from the host

what I read is that the problem of HW raid is that if one disk fails you need exactly the same disk to recover

yeah, i bought a couple of spares

and have a hotspare

while with ZFS you can just use a disk that hosts a same size GPT

I see

it's only a 3tb volume; if one drive fails and have trouble sourcing a replacement maybe i'll try zfs on it

(it is my bhyve host, so zfs volumes with it would actually be nice)

I plan to do jails on this server, then my understanding is that I shall have (for HW Raid and UFS): two disks for the host, two disks for the jails and two disks for the data (I will have enough data to put it separately)

with ZFS tehcnically you could do several GPT partitions per disk and then do the RAIDs with the GPTs, so you do not need that many disks. You could even have just two and share them.

I am trying to understand the rationale of arranging this

hjf: nice rabbit hole you got me into - I'll try to find my real c64/1541 combo I should have stored somewhere far away

saper: but not with desterm

saper: there are also many commodore bbs out there

one has even a bridge to chatgpt

and hackaday haha

hjf got CMD Turbo 232 detected by 3.0 @DE00, -acia1 command line options was missing

a yes because its not a user port modem

> cu -l /dev/nmdm1B -s 2400

can't open log file /var/log/aculog.

Connected

AT X4 S11=50 S7=50

it is talking to me :):)

sweet

ATZ ?

hjf: now I have wired nmdm1B into /etc/ttys and I get FreeBSD login prompt from my laptop :)

remember to set it to VT100 haha

you also need to go to desterm menu (set up mouse in x128 and right click opens the menu)

then go to files, load emulation, and select vt100

I have vt102 in /etc/ttys but FreeBSD getty does not need proper $TERM

Hi. I have this external disk running FreeBSD default install with zfs-root-encryption. I would like to mount the zfs pool of such disk to acess its contents. I started using geli attach (+ typing password) but zroot pools are identical names and dont allow me to import.

you're probably plain ANSI atm

found only "VT102 DecId" setting

(look at the status bar for desterm at the top, it should say ANSI-23W)

I naver did this before, so I have no clue, other than googling it.

it's under Files. the emulation is a file you load

FreeBSD getty changed to 2400 and does not like 38400 anymore :)

cybercrypto: same GUID as well or just same name?

cybercrypto: what does "zpool import" say with no other parameters

funny how long it takes to display screen splash license screen at 2400 bps

saper: it shows the zroot pool of the disk I want to access. (nda0p4)

saper: I am trying to understand about this syntax: use the form 'zpool import <pool | id> <newpool>' to give it a new name

due to the fact it is complaining against the names = cannot import 'zroot': a pool with that name already exists

zpool import zroot newzroot

saper: If I force a zpool import -f (using new name) Am I going to break any data? = Last accessed by myhost and domain (hostid=0) at Sun Jun 21 14:34:42 2023

no, you won't break anything

do not force import, rtprio gave a correct command

the rtprio command does not succeed, and raised the warning I described above.

i think you will have to force in this case

rtprio: correct, it only raises a warning telling me to force import

rtprio: let me check

because as far as the disk you're importing thinks, it's sill a part of `myhost`

rtprio: sounds fair.

rtprio: the import worked and I can access data now. really great.

rtprio and saper, many thanks for the inputs. appreciated.

👍

hjf: now I can say "I can reproduce your problem" :):):)

what worries be is why nmdm(4) stops accepting writes at some point (Input/output error followed by 0 returned from write())

read side seems to work

hjf: can I save the baud rate somehow on the desterm disk?

saper: you can, apparently, if you use the "dialing directory". but the default is always 2.4k

hjf: /usr/bin/printf 'decterm|vt102 80x23 for DEC Term on C128:co=80:li=23:tc=vt102\n' >> /usr/share/misc/termcap && cap_mkdb /usr/share/misc/termcap

those let me use the "decterm" terminal that has 80x23

this is my /etc/ttys like but you probably do not need it "nmdm1B "/usr/libexec/getty std.38400" decterm on secure" (with tabs in between)

but it works quite nice, thank you for the entertainment :):)

probably you could connect socat to /dev/nmdm1A and then use that ttys(5) like to get FreeBSD login prompt

hjf: 57600 or more bps seems to very slow, the CPU can't handle it?

saper: i've found 38400 is the fastest it can do. not sure why it doesn't go any faster, it supports up to 115200.

if i use 115k it's just as slow as 9.6k

saper: btw: #c64friends

yeah we are very offtopic here, thanks

I just wonder why write(2) fails with Input/Output error after the other side of nmdm(4) has disconnected and reconnected. I can't reproduce this with a simple cu(1) connections. cu has no problem with the other side being temporarily gone.

i mean, there are people far more knowledgeable about commodore than me in that chat hehe

What's the best solution under FreeBSD for the problem Debian solves with update-alternatives(1)?

interestingly the new terminal definition dees not set cols/lines after login, but when the screen starts it takes the term size from there

Specifically, I've just switched the shebang line of a bunch of scripts from `#!/usr/bin/lua` to the unsatisfactory `#!/usr/bin/env lua54`.

As far as I understand neither of lang/lua5{1,2,3,4} provides an executable named lua. Not in /usr/local/bin/ nor elsewhere.

java has javavmwrapper, not sure about lua

|cos|: If you're using the ports framework, there's USES=shebangfix

saper, vkarlsen: Thanks! Your answers covered getting me to understand the options available.

|cos|: HTH

Not sure if this is exactly the right channel for this, but I'm running into an error trying to do a make installworld onto an RPi4, from 13.2 to 14.1, it is telling me "Required library libcrypto.so.30 not found." and I'm not sure how to resolve this one. In my /lib I have libcrypto.so.111.

And just to note, I had to move the /usr/obj/usr/src/arm64.aarch64/tmp directory out of the way because I built my world/kernel on an amd64 machine and it refuses to install the world or kernel on the Pi itself if that directory is there because everything in there is not arm64 executables but amd64 executables.

how did you build the world?

On the amd64 machine, I did `make TARGET_ARCH=aarch64 buildworld buildkernel`, later on the Pi, I have /usr/src and /usr/obj NFS mounted ro from the amd64 machine and after moving that tmp directory, I was able to successfully do `make installkernel` but `make installworld` fails with the above.

the tools at /usr/obj/usr/src/arm64.aarch64/tmp can be used to install on the amd64 using DESTDIR on the storage device of the arm device Ithink

so you need tools built for amr

for arm

Is there some way to force install libcrypto.so.30 and then continue with the installworld from there?

I think you have a deeper problem than just this

This was never a problem before, but I recognize that the OpenSSL library changed between 13 and 14.

can you install on amd64 by attaching arm storage there?

can you post the full log of your installworld somewhere please

like bsd.io

The Pi is an active system, I do not want to shut it down to do this. I want to update like I would any other system. And right now, the full log would just be it starting the "Install check world" step and immediately giving me the error.

There must be the way to install native tools for the Pi, but I don't know the correct make incantation (/usr/src/Makefile lists some possibilities)

I think I got it working. I went into /usr/src/secure/lib/libcrypto and did `make install` in there, and then when I went back to /usr/src, make installworld proceeded just fine.

but I think this should be fixed somehow

It IS fixed, for my case, I was only missing libcrypto.so.30 to be able to finish the world install, it is there now, the world install can proceed. It was only needed because of OpenSSL changing versions from 13 and 14, if that hadn't happened, this wouldn't have been a problem to begin with.

normally FreeBSD can handle similar situations just fine

that's why I think it should be fixed