ya but this is excessive wear and tear from what i read

what partition scheme should a freebsd bhyve guest get in the bsdinstall zfs configuration dialog? handbook says gpt but bios or uefi? i start the vm with uefi bootroom so i'm assuming GPT (UEFI) right?

why do I keep losing nickserv authentication

doing scripted bsdinstall. why do we have to provide a plain text root password in ROOTPASS_PLAIN? seems like a security risk to have the plaintext pw floating around

Hi, I ran freebsd-update fetch just now and it said that it doesn't want to overwrite /etc/ssh/sshd_config because my version has been changed. Is there a way to diff my version with the would-overwrite version so I can see if the new version has any changes I want to integrate?

adiabatic it would ask you about that

adiabatic usually it says "diff? view? merge?" etc

At only the fetch stage? Or do I have to press on to get prompted to integrate the changes?

adiabatic freebsd-update install

Onward ho, then.

adiabatic good luck!

"No updates are available to install. Run '/usr/sbin/freebsd-update fetch' first."

OK, fair enough…

Just wait around until 14.0 gets released, and then handle it?

8:41 PM <bsdbandit01_> good evening im running freebsd 13.2 with bhyve im trying to install and run openbsd within bhyve ive tried the following configuration when trying to boot the vm loader = "uefi"

8:41 PM <bsdbandit01_> and im still not able to boot the vm i keep getting the pxe ipv4

doing scripted bsdinstall. why do we have to provide a plain text root password in ROOTPASS_PLAIN? seems like a security risk to have the plaintext pw floating around

I'm configuring a FreeBSD install to get X11 working and I'm on this step: docs.freebsd.org/en/books/handbook/x11/#x-config-monitors-xrandr

Title: Chapter 5. The X Window System | FreeBSD Documentation Portal

how can I run xrandr? startx doesn't find the display

sorry, xrandr doesn't find the display. startx doesn't work yet

"no screens found"

rsjw: HW?

+ version of FreeBSD, etc.

not sure if my message got through because my irc client was acting wonky: I'm following the guide to get xorg configured but it's asking me to run xrandr to get the screen resolution, but xrandr can't run without xorg working. how can I get around this?

02:53 < coreystephanphd> + version of FreeBSD, etc.

13.2-RELEASE

okay, and on what hardware?

lenovo thinkpad T450. it has an Intel HD Graphics 5500 I believe, but 5500 is not grepping out of dmesg

is X11 supposed to start automatically?

on a new install, I mean

I think the i915kms driver is working though, because I restarted and the console text got smaller

Yes, you do need i915kms

yes, I followed the directions in the guide to install i915kms

Title: Graphics - FreeBSD Wiki

I have found that xf86-video-intel, libva-intel-driver, mesa-libs, and mesa-dri tend to be less optional than the guides suggest

Start by making sure that you have drm-kmod, mesa-libs, and mesa-dri installed *AND* kld_list+=i915kms in rc.conf

if you have not changed any xorg settings, try startx again after rebooting and see if you get into TWM or w/e xorg ships with

also, make sure that you have full xorg

how do I make sure I have full xorg? I did pkg install xorg

then you do, good

did you try manually tinkering with any xorg settings? have you installed any wm or de?

and I did install those other things/add that in rc.conf, but rc.conf says = instead of +=. should it be +=?

no, sorry, you have it right

and did you try to change any settings, such as in .xinitrc?

I added 20-intel.conf to make a Section "Device" with the BusID like it said before

without that, I got a different error

ah, that normally should not be required and might even cause a problem

I'll delete it and try again

it *should* just work out-of-the-box on the T450 if you install drm-kmod, mesa-libs, and mesa-dri and make sure to have i915kms load in rc.conf

excellent, deleting it worked

yeah

your custom setting messed it up

is it possible to mount encrypted HFS+ on freebsd?

xorg is very finnicky about such things, but it is supposed to work out-of-the-box on supported Intel iGPUs

it does take some what are IMO unnecessary steps in FreeBSD right now, but that is beside the point. You got it working. Nice

qmr: wiki.freebsd.org/HFS -- no

Title: HFS - FreeBSD Wiki

thanks for the help. I did read the box saying that config wasnt normally needed, and I even knew that beforehand, but then I got to the later sections and just started following directions heh

bugger

Any idea why what was a text console is now a blank white screen on my server?

rsjw: FreeBSD's *desktop* documentation is sporadic and randomly spread ATM between Handbook, Wiki, and random forum threads, all of which have varying levels of being up-to-date or not. #freebsd-desktop here in LiberaChat and the Wiki are generally what I consider to be the 2 *best* places for up-to-date info. The Wiki tends to be better on graphics and sound drivers than the Handbook, for

example -- strictly anecdotally/personally speaking.

The Handbook is very sysadmin-centric.

true

qmr: What kind of server, etc.?

anyone seen RhodiumToad?

coreystephanphd: thanks for the tip on #freebsd-desktop, I'll poke my head in there too

rsjw: No problem. Happy to help. #freebsd-desktop can seem like it is all about KDE sometimes, because KDE is kind of the FreeBSD project's 'official' flagship, but it is really a place for people using FreeBSD on desktop to have a separate discussion zone, whereas #freebsd (here) tends to be more about server, enterprise, and other things.

I lurk all around. Join in when I think of it and offer help in those moments.

Just learning, too :)

coreystephanphd: random craigslist bits. was a text console, after some updates and stuff white screen instead. it's a fancy ish mobo with dual nics and oob. intel graphics. oh, I did turn on something in the bios to enable cpu graphics I think too.

mobo e3c226d2I i5 4440 I think

OS updates?

that might have been somewhre in the mix yes

It sounds like you need to find the specific thing that broken the console. It could be a lot of different things.

[root@frisbee /etc]# ps aux|grep X is empty

polyex: I saw Rhodium lurking as recently as last week. Might just have an ISP outage or something.

ah ok. trying to get this scripted bsdinstall set up and kinda stuck hehe

qmr: Well, since I, a rookie, seem to be the helper ATM, ummm...I suppose what I would do in that situation is try to boot into a live environment from a flash drive, maybe with GhostBSD, and make sure that the screen is actually working. Since...you *did* do a hardware upgrade, too, right? (sounds like)

If you can eliminate the hardware itself as the problem, then turn back to the software. look from different angles.

polyex: I would just ask your question and hope for another pro to respond :)

doing scripted bsdinstall. why do we have to provide a plain text root password in ROOTPASS_PLAIN? seems like a security risk to have the plaintext pw floating around

coreystephanphd: yes but console was working on this same hw previously ( had to change NIC references from em0 to igb0 )

oh just realized ROOTPASS_PLAIN isn't 13.x. damn. so doing scripted bsdinstall, what do we add to the script to set the root password? ideally using the hashed value, not the plain text value.

you can use openssl to get a hashed password value, iirc

ya but how do i use it?

openssl passwd

or use ROOTPASS_ENC in your script

ROOTPASS_ENC is -current only no?

it's in 14

i'm on 13.x so i can't use it

how did ppl set root pw in scripted bsdinstall in 13.x?

does the system have network connectivity?

ya

then drop a ssh key in place and don't set a password

ya, but is there a way to do what i want tho?

jc

run openssl passwd, and write a script to put it in master.passwd

or run 14 >:|

oh so you're saying if there isn't an env var for providing root pw in hashed format, just hardcode it in the script part in a command?

like with a command

yeah

you just have to sed ^root with your hashed full line

i`m out

ty

i set dummy password for root just for the matter of having it and then simply disable the root account on running system. and use doas/su instead

hmm seeing some freebsd or zfs smartness here. rsyncing these old storage SSDs to file server so I can wipe and use for something else, tons of read from the SSD, then brief bursts of full speed writes to the HDDs intermittently

ya me too but i still need to set initial root pw. and i wanna set it with a hash instead of plaintext

trying to figure out how to do that

what's the larger context here? I would say just do that later. you can do it with pssh or ansible or puppet or whatever after the system is provisioned

well rather just make installs right from the start

is it possible to install freebsd "from outside" somehow? I want to create a CF card for a small device (probably using nanobsd), but I don't have any running freebsd systems yet

is there a command to change root's password hash from commandline?

if i have sudo set up, can i just disable root login entirely? (set pw to *)

any downside to that?

Yes.

what're the downside?

None. well, assuming you trust sudo and it's big code base

ya hehe

and all of the normal system level periodic crons and stuff will run?

polyex: one big downside is: what if an upgrade goes awry and sudo doesn't work?

can you think of any others meena?

that's the main point. upgrade goes funny, sudo needs a lib that's in limbo

raspberry pi 5 is coming out

it's literally the main reason we've kept toor around for thirty years

Title: ⚙ D40565 Remove toor backup account

wow it's being debated today still

i wonder if it'll be accepted to remove toor?

it's planned for 15.0

we got around two years

the historic reason toor was created was to get a root user defaulting on bourne shell, nothing more, it got useful later, so the reasoning that some are pushing to remove toor is now root is having a bourne shell

looks like default root shell is /bin/csh

so as long as i don't change that, i don't need toor?

not in freebsd 14

the root shell is /bin/sh i freebsd 14

oh the default ... ahhh

toor gecos being literally bourne-again superuser ;)

bapt do you see any problem with me disabling root login (by setting its pw to *)

?

depends on you, not on me

I really don't see the point of doing it

if i do that then i don't need to set a pw or key for root. isn't that a good enough reason?

then you have your reason, you don't need my opinion on it ;)

there's a few ways to disable login. can set its shell to nologin, can set its pw to *, can also 'lock' the user

apparently setting pw to * stops shell login, but not ssh auth. that's weird

ok i just set root account to a * password AND locked it

didn't touch the shell because it has ignorenologin in login.conf so it would be useless

i miss anything?

Is there a place where I can see all the packages: pkgmir.geo.freebsd.org/FreeBSD:14:amd64/latest/All gives me a 403 forbidden, same for pkg.freebsd.org

you can use pkg rquery

doing scripted bsdinstall how can i provide config files in the installer iso that the packages installed by the bsdinstall script need?

for example nginx.conf that will end up going into /usr/local/etc/nginx after pkg install nginx in the script part of the bsdinstall script, after its preamble

i don't wanna use heredoc and dump ALL of the configs into the bsdinstall script

actually wanna put as little in bsdinstall script as possible, and leave as much in files that'll just go into the right spot

re: zfs disk partitioning, how can i tell on a running system if GPT (UEFI) or GPT (BIOS + UEFI) was selected during bsdinstall?

sysctl machdep.bootmethod

polyex ^

NICE!!!

Hey guys, we have a build host running 13.2, which we use, to build the kernel & world from source with a custom kernelconfig and patches to /usrc/src. When then usually mount /usr/src & /usr/obj via nfs on our "productive" servers and run make installkernel installword. That worked without a problem until now, as we want to upgrade from 12.4 to 13.2 - which fails with undefined symbol errors. Do we need a 12.4 build host, where we build

13.2 so that we can install 13.2 by source on the 12.4 servers?

daniel1 do you do scripted bsdinstall?

danel1

danel1 what is the exact error?

polyex: check if you have just efi (gpt/uefi selected) or efi+freebsd-boot (gpt/bios+uefi selected) in gpart show output

polyex: No, we simply use make.

antranigv: ld-elf.so.1: /usr/obj/usr/src/amd64.amd64/tmp/legacy/usr/sbin/make: Undefined symbol "regcomp@FBSD_1.6"

running newer binaries (native make built on 13.2 is "newer") on older host is not supported

it can work, or it can break

thats what we tought.. so we'd have to build 13.2 on a 12.4 build host and then it should work i guess?

yes, that *is* supported :)

danel1 oh yeah, seems like a versioning issue indeed. if you're upgrading from 12.4 to 13.2, then your build machine should be 12.4 :)

well then, i guess we gonna create a new build host... :-)

danel1 good luck!

polyex I do scripted BSD Install, a very complex one.

polyex I think you had a question?

ya tyvm

bsdinstall manpage says for zfs i gotta declare zfsboot_vdev_type and zfsboot_disks to create the pool. but i don't declare zfsboot_vdev_type because its default stripe is what i want. is that ok?

lemme check my scripts

polyex I have `export ZFSBOOT_VDEV_TYPE=stripe`

but that's the default

so why include it?

polyex just to be safe, I guess ¯\_(ツ)_/¯

hmm

I also have all the other defaults set as well, such as ZFSBOOT_BOOTFS_NAME and ZFSBOOT_BEROOT_NAME

why?

you know never what could go wrong. shell scripts are not very predictable, you know :))

i got freebsd running zfs. it's a host for bhyve vm guest that's also running freebsd and zfs using a 40G dev volume the host makes for it. there any reason to set copies=2 in the guest vm? can a vol created by the host zfs ever fail and need healing in the guest vm?

antranigv

nah it would be useless to set that inside the guest. use the ZFS features in the guest for better management, i.e. snapshots, clones, whatever you wanna do. but for other things use the host, i.e. raid, copies, etc

yep ok makes sense

doing scripted bsdinstall how can i provide config files in the installer iso that the packages installed by the bsdinstall script need?

do you tar them all up and add them to DISTRIBUTION?

as just another tar.xzf

tgz

antranigv

polyex yup!

that's what we do

we have company.txz

that we ship

and we added that to distribution

that's pretty much it

can that overwrite other files that will be there by default, like /etc/rc.conf with ur own custom?

put your stuff in rc.conf.d/ and don't touch rc.conf

wow, no kidding?

no rc.conf.local either or?

polyex it can, but don't do that. instead use either rc.conf.d or post-install scripts

12:06 <bapt> put your stuff in rc.conf.d/ and don't touch rc.conf <<= Puppet for instance does this

post install script meaning something in the company.txz that you run from the script part of the bsdinstall script?

polyex yes

ok!!

so i got freebsd running zfs. it's a host for bhyve vm guest that's also running freebsd and zfs. in bsd when i select the disk to use, looks like bhyve created vtbd0 and ada0 for me. vtbd0 worked but ada0 didn't. that make sense? what's ada0 then?

vtbd is for sure the hard drive. what is ada0? you can check in the `dmesg` of the guest

or you can run `gpart show ada0`

or `geom disk list`

hm only the vtbd0 shows up

weird

the ada0 description was something like bhyve SSD

and i could look at the disk info for it through the zfs configuration dialog

looking now

ok got it

disk info option in zfs configuration dialog, shows 2 disks

ada0 bhyve sata disk

vtbd0 virtio block device

antranigv

what devices show in the guest depends on the bhyve config

ada0 is an emulated sata device, vtbd0 is a virtio-blk device

if you boot an installer image, you might get the image on ada0 and the intended destination volume on vtbd0

i see in my bhyve command virtio-blk, but i don't see anywhere i init an ada0

ahhh

how did you run bhyve?

that's p;rolly it

bhyve -AHu -c ... -m ... -w -s 0, amd_hostbridge -s 1, virtio-blk, ... -s 2, virtio-net, tap25 -s 3, ahci-hd, ...

gets pretty long heh

that ahci-hd is the ada0 one

that's what i was thinking

tyvm!!!

good to have you back btw

the vmrun.sh example script adds an ahci-cd or ahci-hd device if you use its -I option, choosing which to use based on whether the filename is *.iso or not

i have a few dozen bhyve vms. i make a tap# interface for each 1 and add it to cloned_interfaces in rc.conf. it's kinda tedious. any way to eliminate that?

So .. the ARC is/can be compressed; I presume this means that there's no additional compression/decompression step to benefit the ARC - it's already compressed on/for disk, so it's "free", in a sense?

Question then is this also applies to encrypted datasets?

My brain hurts.... if I want the same ZFS dastaset mounted rw in two different jails, how?

antranigv yuripv - well now the build on 12.4 fails :-).. cp: /usr/src/sys/modules/irdma/../../dev/ice/ice_rdma.h: Function not implemented

did you install any 13.x binaries on the 12.4 host?

from scrollback, the problem with trying to installworld from an NFS /usr/obj is that the "host" tools in the built world are built for the build host's arch and version, and won't necessarily run on the target host

We didn't. We tried cloning the 13.2 source and run buildkernel / buildworld on there.

on there = on a fbsd 12.4 host

dvl: mount_nullfs ?

What RhodiumToad said, we hit that problem many times.

danel1: When upgrading a 12.4 host, you must do the build on a 12.4 host

thats what i am trying

Ltning: That seems to be the way. In the past, all my nullfs has been readonly. In this case, I want rw.

Ltning: But I see now, that's just history. nullfs can do rw.

dvl: this is not a split-brain situation since it's still the same brain (kernel)

danel1: Good. I've been through this several times and it's bit multiple times.

Ltning: ... we're still talking about nullfs? Yeah, same kernel, so no different than two processes trying to write to the same place (that will be controlled by clients, given the situation).

danel1: and you started with a clean /usr/obj ?

danel1: and no ccache?

Ehrm.. maybe its my fault.. i tried running buildkernel before buildworld.. buildworld now seems to run smothly..

right, buildworld goes first

when you've built both, remember that you have to do etcupdate -p, installkernel, reboot, installworld

RhodiumToad: For 12.4 to 13.x, does the etcupdate need to be boot strapped?

I recall doing that at work

dunno offhand. I tend to update by building entire new images

antranigv: thanks for your Devuan jail article, there's surprisingly little thorough documentation on the subject

antranigv: reading it a first question comes to mind: why openrc & not sysvinit?

secondly, I love that you mention using /rescue, I just do by nullfs mounting /rescue into /bsd in the Linux jail

veg anytime!. I hope everything worked fine?

I just discovered it, so I haven't given your approach a try

veg sysvinit works fine too, I'm just more familiar with OpenRC because I used to run Gentoo for years

oh, thanks for the clarification

have you encountered issues in running non-vnet Debian jails?

veg that's a good point, we can use nullfs indeed for /resuce -> /native

I haven't gone through the trouble of VNET'ing anything yet

I never run non-VNET Jails, but one of our employees tried it using my guide (but without VNET) and reported that it all worked fine. the only exception is that you'd need to configure the networking via jail/jail.conf instead of /native/ifconfig

right, totally fine for me

veg I can report that, however, with VNET Linux Jails, you can have pf inside Linux :))) NOW LINUX HAS CLEAN WATERS!

hahaha

I just need to add DTrace inside Jails securely, and we'd be on-par with illumos features.

well, just the uname -a kicked me off my chair the first time

I was very interested by Debian kFreeBSD when it was a thing

looks like Debian jails come closest

have you found Devuan jails to be stable & production ready, antranigv?

I was very impressed with speed & performance, seems like the linux compat code is efficient, yet I'm slightly worried to involve more moving parts, and possibly moving parts that won't attract much love from core devs

veg for my use case, yes. but I don't think that I'd recommend it for everyone.

I may be mistaken, and there may be a big crew of FreeBSD devs who are obssessed with Linux emulation,

but it seems somehow unlikely

simplest example: modern epoll is not implemented, hence we cannot run modern nginx. unless we compile it ourselves.

at the moment I just wanna run rtorrent/libtorrent on 10k torrents, I figured out to lift the maximum files limits thanks to compat.linux.default_openfiles, that's a start

but I wonder how network performances could be affected?

what's the problem with running natively?

RhodiumToad: I can't get rtorrent-ps-ch to compile

it has some linuxisms, some patches depending on random_r from glibc

I need to migrate the exact config from a running Debian server, and I can't afford to downgrade to vanilla rtorrent for the time being

veg yeah something like that would most probably work fine.

altho, you don't need a jail for that, maybe try a chroot?

(if you're okay with running it inside tmux :D)

huh. we have code for random_r but it's not exported from libc

I need to learn tmux first, heavy GNU screen user here, but I know "tmux is the way"

true, a chroot might suffice

what's an additional jail amongst friends, though?

especially with zfs snap && zfs clone

interesting RhodiumToad, the other one is initstate_r

but my c++ is non-existent, so aside from using patches from net-p2p/libtorrent to add the proper includes, I haven't made any headway to making rtorrent-ps-ch compile

it's probably not much

I'd be happy to learn how to make a proper port, since I assume FreeBSD is being used as a seedbox/NAS by quite a few people who could see it handy, but I seemingly lack some building blocks

all of random_r, initstate_r, etc. exist in libc but without them being exported, there's no easy way to get at them

veg I created my own, because everything else was way "not my way"

veg you can check it out here. in case you like it too. github.com/illuria/jailer but it forces stuff. like you have to use ZFS

Title: GitHub - illuria/jailer: Minimal, flexible, and easy-to-expand FreeBSD jail manager.

that's unfortunate, but thank you for the clarification, RhodiumToad

interesting, antranigv, thanks

I actually decided to not use any jail manager for the time being, as I wanted to properly understand what was going on under the hood

bastille looks nice enough, but I figure it will stop being maintained sometime in the future

pot looked like an interesting contender too

but plain jails.conf (& the occasional deboostrap) have been easy enough so far

runj too maybe

interesting polyex, more reading 🤓

looks like your script is "vnet all the way", antranigv, it may force me to try :)

I was reading about some heavy performance penalty for using vnet though?

antranigv what's the key reason to go all in on vnet?

with some workarounds documented on a klara page?

you guys using pf or ipfw?

I want to use pf, polyex, but I haven't started writing my firewall

this leads to a question, however

i use pf and like it but i'm gonna try switching to ipfw. it's the native freebsd fw

I was under the impression that having jails using an IP within the same range as the hosts, on the same interface (bge0 in my case) allowed for not configuring NAT?

may I be mistaken?

vnet or non-vnet jail?

non-vnet jail shares the host's networking, so you just put some alias IP on the host interface and have the jail use that

non-vnet, I have host:bge0(10.30.20.30) & jail:bge0(10.30.20.31)

for vnet jail, you can bridge the jail's epair and the host interface

that should work fine

only networking conf for jail is ip4.addr & interface in jail.conf

well, it does work fine for ping, ssh in/out

but I don't manage to transfer any data using the previously mentionned rtorrent

tried it in a regular jail with rtorrent from ports, same

so I figured I may be missing some pf action

though I fail to understand which

I do launch have rtorrent bind to the jail's IP

I'm guessing it needs inbound connections? I haven't used torrent in years

yes, inbound connections

outbound works fine, ssh'ing from the local network & the host works fine too

sockstat -4 shows outbound, but inbound is a no go

so since you're on a private IP, you must be doing nat _somewhere_, and that's where you need to map inbound connections

I do it on a Linux firewall, yes

the ports are mapped

I'll go through the conf again, but the very same lines work flawlessly when redirecting traffic to the Debian server I need to migrate

once I swap the IP for the FreeBSD jail's, no action

is there a sysctl trigger I may have forgotten?

doubt it... did you check with tcpdump what packets if any are getting through?

I see traffic but length 0

I haven't used tcpdump in a while, I need a crash course I suppose

Title: rtorrent to jail - Pastebin.com

(this is an excerpt trying to retrieve a Debian image, should be safe to share)

every 3 months, ansible's acme_certificate module bites me in the arse

and every 3 months, I swear I will replace it with a tiny shell script

and every 3 months, I forget about it

the reason it's not straightforwards is multiple internal & external services, needing DNS-01 auth and some automation

veg: there's definitely packets going both ways there? e.g. line 23 has an incoming SYN which leads to an established connection which is then closed cleanly

veg: however, the initiator of the connection sent 68 bytes of data and whatever was running in the jail just acknowledged that and then immediately disconnected

veg: so you might want to check your logs and config for what's in the jail

thank you, RhodiumToad

logs are empty, but I'll double check the config file indeed

vanilla rtorrent (compiled from ports) is what's running

this test is in a plain FreeBSD jail, not the Devuan one running the patched version

OMG RhodiumToad

MANY thanks to you

I don't know what exactly, but it was something in rtorrent's config

thank you so much for pointing me in the right direction

I'll dive into FreeBSD's networking later on as a result :)

yw

RhodiumToad antranigv yuripv Thanks again for your help - it's working as expected now :)

how can I get the volume up/down keys to work?

rsjw, i solved this for myself.. let me see how i did it x)

the keys are recognized by xev so I'm not sure why they're not having an effect

although I'd probably want the volume to change even without x11

rsjw, for me to make multimedia keys work on freebsd, i added usbhid_load="YES" hw.usb.usbhid.enable=1 in /boot/loader.conf

two lines

if xev sees them, then it's a matter for your window manager to adjust the volume accordingly

yeah xev didnt see mine so thats a difference

if you want the volume to change without x11, then the next question would be whether the keys are detected by ACPI

how would I check whether the kays are detected by APCI?

Test

Test was successful. You are now authorized.

Greetings! I am lurrrrrr, leader of omicron perciay eight!

Lol

On the Internet no one knows if you are a dog or not. Note there is useful information in the channel Topic. That may be truncated. Most IRC clients will repeat it with the /topic command.

antranigv: 10 seconds to look at a weird error?

on 15.0-CURRENT *and* OTP26.1+ I get this cat vomit

Title: Snippet | IRCCloud

no function clause matching in :prim_tty.cols/2 (kernel 9.1) prim_tty.erl:980: :prim_tty.cols([{:ansi, "\e[G"}, "\e[92m" ...

if I use OTP25 its fine. If I *don't use 15.0-CURRENT*, it's fine.

I'm starting off in OTP src but yuck anything with termcap is The Dark One's territory

ok, it's to do with my .iexs

I think its actually erlang/otp #7631

Title: kernel: fix prim_tty xenl glitch by frazze-jobb · Pull Request #7631 · erlang/otp · GitHub

it seems this `{:ansi, "\e[G"}` is enough to break OTP26

wow

so, not that prim_tty issue, but possibly similar.

antranigv: thanks from earlier!

im using poudriere-devel, for a given jail/ports combo i have a bunch of "queued" ports in the web interface but they are already built (i think from a previous run that failed cause browser made cpu go a bit too wee)

whats the odds its just a little ux/ui artifact and not something i should "worry about"

dch hey! sorry, just got back