yes! everything about this page inspires absolute confidence! freedesktop.org/wiki/Software/xlibs

Title: xlibs

dammit, I hate it when I forget to mount large data drives with nosuid

(or noexec)

I wish there were a persistent attribute for that

is it possible to setup the following: partition a drive in 2: 1 part for OS , second part as special device

oh, in zfs, and it has to be mirrored

special device?

yeah, you can use it to store the metadata

aka: data on spinning hdd, metadata on ssd. gives huge speed boost

Do not see the point of partitioning the same disk to separately store the metadata

there is a point. the os and metadata are on ssd, data is on hdds

*metadata for the hdd pool

So the partitions are on a SSD with OS & metadata for the data on a separate disk?

last1, I see you had already mentioned that. Ok then

how many hdds are involved?

RhodiumToad: I just alias mnt to mount -o noatime,nosuid

44

wow

the only part I'm not sure about is how mirroring of the special device would be handled

my question was more like: can I partition ada0 in 250GB for the OS, rest for metadata and then install FreeBSD on ada0p1 with ZFS

I would do the same with ada1 so the OS would be zfs mirrored on ada0p1 / ada1p1 , and the special vdev would also be a mirror on ada0p2/ada1p2

in terms of partitioning, you can do almost anything, because zfs is basically ignorant of the underlying geom layout, it just consumes whatever providers you give it

my problem is that in the installer I can't choose ada0p1 , just ada0 in the Auto-ZFS installer

you have two separate ssds to use for the OS/special bits?

no, just one

anything non-standard and you have to do manual partitioning, obviously

last1, you would need to do that manually

ugh, a manual zfs zroot install, I wanted to avoid that :)

last1: I think you'd have to do manual partitioning from the shell. which is a PITA with ZFS if you want the default dataset layout that the installer creates (and you should because they're good) because you have to create them by hand.

you should be able to do the partitioning in the installer

yeah, that's what I had started on, I created the partitions with gpart

but then I wanted to jump back in the installer

I would install just the OS partition manually, and set the rest up later

and have it use ada0p1

but then I haven't used the installer in anger for a while

last1, You could parition via the installer. Then choose the partition to install ZFS (there will be no separate datasets, bugs.freebsd.org/bugzilla/show_bug.cgi?id=262461 ) Later make the mirror after the installtion.

Title: 262461 – bsdinstall: After manual partitioning the disk, installer installs "ZFS on root" directly on the pool not under separate ZFS datasets

ugly business, I might just get 2 dedicated mirrored OS drives

but I wanted to save on 2 SSD slots and do it all on those 2 drives

what's ugly about it?

ok, let me try this one more time.

so in the Installer, I go manual route, create just one 250Gb partition as freebsd-zfs and try to install on that

and after, I add the second drive as mirror

yup

Title: Imgur: The magic of the Internet

if I try to set mountpoint as / , the installer crashes and restarts

parv: how did you manage to do auto-zfs on just part of the disk ?

unfortunately I don't have a VM spare to test it in right now

any way to know when this page dates from and if it's still to be used ? wiki.freebsd.org/RootOnZFS/GPTZFSBoot

Title: RootOnZFS/GPTZFSBoot - FreeBSD Wiki

ah, it says at the bottom: last edited 2022-06-12

so fairly recent, but this line in the beginning: newfs_msdos -F 32 -c 1 /dev/ada0p1 gives me an error: 1542 clusters too few clusters for FAT32, need 65525

is that for an ESP?

apparently it's for the UEFI boot partition

I'm doing it by hand

yes, that's what an ESP is

1542 clusters of 1 sector looks very small

eh, I'll just do legacy boot instead

are you trying to make it just big enough for loader.efi? that's a very bad move

well, I was just following that document: wiki.freebsd.org/RootOnZFS/GPTZFSBoot

except on install media, you should make it at least 64 megs

Title: RootOnZFS/GPTZFSBoot - FreeBSD Wiki

it's apparently wrong

you can make the partition larger, or use fat16, or both

try -c 8. I *think* it's better to stick to FAT32 for the ESP...

-c 8 will make it worse

that'll push the minimum size up to ~256MB

RhodiumToad: maybe I remember wrong but I think it's what I did not long ago. it gave me 4K clusters which seems fine. the error message is very confusing.

I've never had any issues with fat16 for ESP

oh alright

4k clusters is fine in itself, it's just that newfs_msdos wants the device to be large enough for 65525 clusters to use fat32

why is iocage making me update every single file in /etc/ by hand for every single jail just to upgrade from 13.1 to 13.2?

so the larger the clusters, the larger the minimum device size

i didn't change weird things like ssh moduli

surely something has gone wrong?

The following file could not be merged automatically: /etc/mtree/BSD.debug.dist

Press Enter to edit this file in /usr/bin/vi and resolve the conflicts

manually...

huh, that shouldn't need a manual update, it should just be replacing it

yeah

but I have no idea why it would do that

but there's no option to replace it

what are the conflicts?

almost everything in /etc/

from where debug files

weird

to sshd_moduli

no, I mean if you edit the file, does it show any conflicts and if so, what

oh

yes

basiclly all the stuff you'd normally expect from the new release

version identifier changes

small additions to hostapd's init script

most of the merges are things that should be totally doable automaticlly

like inserting lines where these was none before

is it invoking etcupdate to do the work, or using its own code?

i don't know i'm just running iocage upgrade -r 13.2-RELEASE ALL

IIRC etcupdate on recent enough 13-STABLE (& 14-CURRENT) had handlled those files

you could look at the process list and see if etcupdate is running

alright i'm restarting to upgrade on the jail to check if it's using etcupdate

but if it is what would that change?

just whether it's an iocage bug or an etcupdate bug

one second

/bin/sh /tmp/tmp_onnvkjc -b /zroot/iocage/jails/mumble/root -d /zroot/iocage/jails/mumble/root/var/db/freebsd-update/ -f /zroot/iocage/jails/mumble/root/etc/fr

sweet, I've done it: imgur.com/a/LNBVH4v

Title: Imgur: The magic of the Internet

eebsd-update.conf --not-running-from-cron --currently-running 13.1-RELEASE -r 13.2-RELEASE upgrade

i don't know

i can't find etcupdate

so now zfs is on ada0p3, and I would just add a mirror to ada1p3 to have it redundant ? don't I also need to mirror ada0p1 and ada0p2 ?

The following file could not be merged automatically: /etc/mtree/BSD.debug.dist

Press Enter to edit this file in /usr/bin/vi and resolve the conflicts

manually...

last1: there's no sane way to mirror the ESP, but you can make a copy of it on the other drive as a backup (nothing on it should change in normal use)

what's ada0p2 ? swap?

RhodiumToad, is there some other way to tell?

if you want to avoid crashing if ada0 dies, you can use gmirror to mirror the swap partition

yes, swap, that I can ignore I guess

oh, good idea

and set the swap to the mirror

you'd probably want to set it to 'prefer' and give the ssd the higher priority

well shit, i'm just going to roll it back

thank goodness for zfs snapshots <3

I left some unpartitioned space on a nvme to install Windows later on. I assume that this just works (and that it'll use the existing ESP without issue) but if this wrecks my freebsd install I'm going to be a little bit annoyed...

it may overwrite the bootx64.efi, you may want to keep an extra copy handy

if you have multiple OSes you probably want some form of EFI boot manager anyway

RhodiumToad: ah right. well if it's only that it's something that can be fixed from a USB key.

RhodiumToad: can setting up some EFI "boot menu labels" (or however that's called) be enough to dual boot them?

heh I haven't bothered trying to dual boot in forever. small SSDs are so cheap. but with nvme it makes sense to try to share them again.

I've not tried dual booting from EFI

but I know in theory how it works

alright. well, I'm just gonna hope that messing with efibootmgr(8) can make it just work.

i.e. in theory it's just a matter of having all the various OS boot loaders on the ESP and either telling the firmware which one to boot via an EFI environment var, or having a menu program on the ESP to handle that part

efibootmgr works by setting those EFI env vars

if you want a boot menu, there's things like rEFInd

RhodiumToad: that's an intermediate boot loader right? you boot into it and then it chain loads some other ESP boot program?

yup

I'm hoping that I won't need anything like that. If I have to press the "boot menu" key while booting to pick from the EFI's built-in boot meny that's fine. if that works...

hm, my buildworld looks nearly done

after 130 minutes

hm, not as nearly done as I thought

hi. is there a package that reports pcie 10gb nic temperatures?

it looks like at least some high-speed NIC drivers report chip temperatures via sysctl?

I'd look at IPMI/SMBus as well.

ok, I'm checking things right now

can pcie cards be on smbus?

tried sysctl dev | grep ix | grep -i temp but nothing relating to the temperature

ixgbe driver?

yes, i think so.. Intel(R) X520 92599ES

looks like ixgbe adds sysctls for temp only on certain device IDs, specifically IXGBE_DEV_ID_X550EM_X_10G_T

sorry, i mis-typed. its a 82599es

RhodiumToad: Not sure, I thought I remember seeing one of my servers show me NIC info via the embedded management interface, and I thought it was getting it via SMBus, but not sure. Couldn't hurt to take a quick look though, should be easy to eliminate if it doesn't.

FETCH_DEPENDS+= ${GO_CMD}:${GO_PORT} \

fucking hell.

how am I supposed to fetch all my distfiles in advance if any go port needs to build go just to fetch itself?

sfox: iocage hasn't seen updates in quite some time

why is that?

it's not abandoned is it?

hopefully this is just a matter of rolling back and waiting for a proper fix upstream?

sfox: iocage/iocage #1289

Title: Last Accepted Pull Reqest was Oct 2021 · Issue #1289 · iocage/iocage · GitHub

1289 – errno breaks in thread-safe c++ compiles bugs.freebsd.org/bugzilla/show_bug.cgi?id=1289

ugh

i just migrated from ezjail too

am I really supposed to just go use linux?

i switched to FreeBSD cause I couldn't stand Linuxisms any longer

now this new update even implements /etc/machine-id

sfox: Linux is now the new standard without a standards body

if we don't follow some of directions or sets you won't be able to run any applications, because people who pretend they program for the POSIX platform don't test on anything than their macOS they developed on, and the Linux ci

anyway. ezjail hasn't seen updates even longer than iocage.

if you want a thing that's actively developed, please do the research or ask for recommendations

So OpenBSD (sans ZFS)?

wiki.freebsd.org/Containers and there's also github.com/illuria/jailer

Title: Containers - FreeBSD Wiki

parv: oh, yeah, OpenBSD will definitely stay what it is. I think

cue OpenBSD integrating a kernel bus

tho, honestly, i'd love to see that, because i think they'd do a fantastic job

btw, jailer might be a bit volatile weblog.antranigv.am/tags/jailer but the work coming out of jailer (lol) and landing in base is there to stay

Title: Jailer | Freedom Be With All

it's almost orwell

sfox: anyway, to summarise: the real reason to join a BSD community isn't the technology, so much, although we have some cool tech, it's the process behind how that tech comes together, and the kind of communities that make that possible

I like the tech.

the fact that i can get a one line patch into the base system, without having twenty decades of kernel hacker credentials tells you something about our process: github.com/freebsd/freebsd-src/pull…is%3Apr+author%3Aigalic+is%3Aclosed

Title: Pull requests · freebsd/freebsd-src · GitHub

GitHub doesn't tell you that only one of those wasn't merged

I love that morse(6) is a game (section six) but it also implements an ITU standard: freshbsd.org/freebsd/src/commit/82a…6d86acaac75b1ff6d4f8b767bbb117694f1

Title: FreeBSD / src / 82a036d86acaac75b1ff6d4f8b767bbb117694f1 - FreshBSD

Nice.

Freaky: what'd you think of displaying the subject line in the title when linking commits?

the commit hash isn't as useful to the casual observer as the commit subject line, imo

seems like games is also meant for stuff that was made for fun and not for some serious need, like caesar too, totaly usable and implements frequency based guessing of the right rotation, ah wait that has a serious need, how else would you solve #ircpuzzles if caesar wasn't in base?

no rc.d script for if_wg in 13.2?

does this mean that wireguard-tools should still be used?

Hello again. Looks like there was an power outage during package installation. Now pkg complains about malformed local package database image. Is there any way to recover it without re-bootstraping pkg?

Do you have a list of packages that were installed? (I do wonder how it got corrupted though, is there a *.wal file in /var/db/pkg somewhere?)

...also, anyone know if there's a setting in Firefox to get it to release /dev/dsp when audio isn't actively playing?

mage: I think it's in ports

Replaced the cable and hasn't happened since, but would still like to avoid the sound device being locked and unable to release/re-create without reboot.

(...or at least, that was the only way I found to fix it, though I found a couple of blog posts that seemed to confirm that's the case)

_xor: Unfortunately, neither. There is only some shell history, but would it be really enough?

No idea. You mainly need to know which packages you explicitly installed were, the rest of it can be installed as automatic dependencies.

I know it's "normal", but seeing a bunch of "pid 12345 (conftest), jid 321, uid 0: exited on signal 11 (core dumped)" messages in my logs still feels...odd.

meena: that's a surprisingly invasive change

Freaky: hmmmm… I figured it might be

Title: FreeBSD / ports / 23612e19c7ea25fc1a94c70aa3da2905df7da6af / www/onlyoffice-documentserver: Fix path in documentserver-jwt-status.sh - FreshBSD

maybe not *that* invasive :)

shortened the ids too

Title: FreeBSD / src / 77f0e19 / procctl: add state flags to PROC_REAP_GETPIDS reports - FreshBSD

Nice!

now my only complaint is that some diffs are too hard to read on mobile

the line numbers take up way to much space im.eena.me/uploads/8b910d392321b60c…reenshot_2023-04-16-16-37-13-23.jpg

turn phone landscape?

_xor: I reported the reason for that conftest segfault years ago, it could have been fixed but has not been

Freaky: why does it show older commits to main as being on e.g. releng/13.0, related to git switch?

RhodiumToad: What's the actual fix for it?

(brb, picking up lunch)

Title: 236165 – crash in malloc with ld.lld and -Wl,--export-dynamic -static

yuripv: branch names in git are a clusterfuck

that answer works :D

with svn you just extract from the filenames in the commit, with git you.. shrug

branches are just names for tips on the commit graph

"older commits to main"? you mean ones from before releng/13.0 was branched?

I use git name-rev and it tries to pick one vaguely sensible branch name for the given commit, which is far less than ideal

git just has no concept of what branch a commit was made to, you can only vaguely infer it from walking the commit history from the tip of a branch and going back through history until it seems to merge with main or another branch

RhodiumToad, _xor: reviews.llvm.org/D42748

Title: ⚙ D42748 [ELF] Don't create a .dynamic section when linking with -Bstatic

hey there; for an 8 x 22tb disks NAS storing movies, would I be better off doing a zpool of 2 x RAIDz1 VDEVs (containing 4 disks each) or 1 x RAIDz2 VDEV (8 disks)?

I've consistently read that the more VDEVs the better performance wise, but wouldn't using 2 RAIDz1 VDEVs reduce redundancy by only allow a single drive failure per VDEV?

also, I'll be using 10gb NICs soon enough, hence my worrying about having better IOPS

I really should try Nandos sometime. I've heard good things.

if you have Moe's Southwest Grill, it's like Chipotle but a slightly better. I like both, _xor

Yeah. Fast food chains are, oddly enough, one of the things I miss most since moving to a small village in South America where those options don't exist.

meena: tried removing line numbers for small displays but Chrome told me to die in a ditch

Freaky: that's okay, i don't use chrome

omg, it's a stupid global nodejs dependency that's causing this

Looked like some stupid race condition or possibly intentional DDoS (sadly, not uncommon with nodejs ecosystem).

Freaky: a quick look at GitHub, gitlab, and codeberg tells me they all do that. it's just your site that i mostly look at: codeberg.org/FreeBSD/freebsd-src/co…c2b136841b84b8ffdc703a1afeee5d0e268 / freebsd/freebsd-src b1a00c2b13 /

Title: Quiet compiler warnings for fget_noref and fdget_noref · b1a00c2b13 - freebsd-src - Codeberg.org

Title: Quiet compiler warnings for fget_noref and fdget_noref (b1a00c2b) · Commits · FreeBSD / FreeBSD src · GitLab

meena: I'll sort something out, one way or another

and then the others will follow suit I'm sure ;)

veg, Both of your proposals allow for two failing devices but the first one only if the second failure is from the other VDEV while the second allows any two.

Both are good configurations. It depends upon your own judgement as to which is better for you.

If you expect to be able to react to the first device fail by replacing the redundancy of the failure before a second one in the set of four fails then it is fine.

That would be typical in a corporate enterprise datacenter situation with spares handy and someone on call to make the replacement every day.

The raidz2 configuration allows any two devices to fail. Allows a little more slack in your schedule if one fails and you don't get to it before a second one fails then things are still okay if you then replace the failed devices and restore redundancy.

At the cost of somewhat less performance than the striped VDEV configuration of the first proposal.

In my home setup I decided I might be away on vacation for a couple of weeks and I did not need that performance. YMMV.

Also remember that if one has an identical collection of storage devices and they are all running identical hours in identical environments then systematic type failures are more likely to cluster together.

Twice in my career I have had two sibling spinning disks die within a few days of each other. One I caught okay.

The other failed before the client decided to replace the redundancy and I had to do a full restore from backup.

rwp: oooooooooof

oooooooooof?

that just sounds like no fun.

For the one that I caught I copied the data to a different NAS and switched over to it. But left the original running since it was remote.

And then two days later the second drive in the mirror failed causing the loss of the entire array. But the data had already been moved. So all okay.

For the one that needed full recovery from backup the entire tale was just snafu because that client needed to be convinced to do something about it causing time to drag on. Could have saved it. But not after the second drive failed too.

Experiences like those are why I am a huge fan of raidz2/raid6 which gives a little more safety. Sometimes just enough more. :-)

I wonder how drives from different vendors that have same specs… fail

.epi

Failures modes from different vendors should be completely decoupled. Which is what you want.

For my own stuff I often split lots of drives. I mean I have two drives that are identical and been running mirrored for a year.

I buy two more identical drives. I then split those lots up so that each system has one new drive and one experienced drive.

Hoping that being a year apart that any failures will not be coupled failures.

Corporations though often have SLAs with service vendors and those vendors will often require arrays to have identical drives and identical firmware. Because they don't want the client to be complaining about weird issues that might be due to drive firmware or behavior.

But they might have a 4-hour service agreement to replace any failures very quickly. Which makes up for the potential problems.

Since veg mentioned 22TB disks I will mention in passing that I sure hope they are not SMR Shingled Magnetic Recording. Because those are completely unsuitable for purpose in a RAID.

If I were _given_ that much SMR storage I would probably still use the drives. But not in any raid. I would use them only as singles.

hello I was trying to make the sound work in a user account and then I followed some instructions from the BSD chapter on multimedia now for some reason the sound seems like it works on the user as well as root

but xorg stopped working in the user account but it still works in root

any ideas what to do to make it work?

Xwrapper.

not sure why it stopped working but that was working before

Ok no Xwrapper then

sixpiece: what error do you get from X when starting as non-root

not*

I just get the 3 screens

it's like 3 terminal screens that are small

I've had it on these issues

how are you starting X?

startx

it's supposed to automatically start however and it works as root

and what is in your .xinitrc, and did you change that file recently?

you have it enabled in /etc/ttys? or where?

no but I might have made a command that screwed it up

Title: Chapter 8. Multimedia | FreeBSD Documentation Portal

I did commands in there

sysctl hw.snd.default_unit=4 an extra one or two of these

# kldload snd_driver

-=- and this command as well that could have interfered the other thing that I did was I added the user to several groups not only pulse but pulse-access and _sndio I then pw groupmod pulse-access -d kiwichap to try and repair the situation

ah, did you take the user _out_ of the video group by mistake, perhaps?

try the 'id' command to see

I added the user to video group but it shouldn't matter when it goes to sddm since it could be any user of the system I think

I mean they don't know specifically that I am that user until I login is what I mean to say

you're using sddb started from where? rc.conf, or ttys?

*sddm

rc.conf

and sddm starts ok and prompts for the user?

termbin.com/cihg for the id command

no it does not start ok

what happens?

good observation I will try

cannot mix incompatible Qt library (5.15.5) with this library (5.15.8)

i had the same issue recently

i did `pkg update -f; pkg upgrade -f` and it got resolved

somehow you have 2 qt libraries core versions or something

> if you want a thing that's actively developed, please do the research or ask for recommendations

yes, i'm looking for a reccomendation of iocage development

i mean yes i'm looking for a reccomendation for a replacement of iocage since it's no longer developed

right I installed chromium browser

rwp: thanks a lot for the thorough reply! My disks are 22tb WD Gold, AFAIK they do not use SMR technology

doing the upgrade now , although it doesn't explain why the root still works

i had the same issue with screengrab

found the solution on the freebsd forums

after forcing and update and upgrade it worked for most people

its like your binary is compiled against an older version of QT then your current library

it says notice: this port is deprecated you may wish to reconsider installing it uses python 2.7 which is EOLed upstream

it must have gotten updated and left the other packages that still relied on the older one unupdated

you have some python2.7 installed, pkg upgrade -f upgrades all your packages

chromiumm is still stuck on python 2.7

ok sounds good

stupid chromium

thank you

no problem

or was when I last looked, anyway - not sure if it's fixed

hm, no, that should have been fixed apparently

so I could go up to 6

?

qt6

does anyone know how to do the secure boot is it really not possible or very difficult?

it is not that bad

if your bios will let you enroll your own key

then it is just some commands to sign the loader.efi etc

helped a customer with in last month

it has a key already it's like a trio boot of windows, linux and freebsd

signing the freebsd loader with microsoft's key would require having microsoft's private key

I'm not an expert on this to say it has a key already

vs, you can add a 2nd trusted key, and sign the freebsd loader with that

correct it worked for ubuntu so it must be able to make a 3rd key

ubuntu has a shim loader signed with microsoft's key

so it doesn't require loading your own key

but yes, it is doable

ok

sounds very interesting

these instructions are rough, and a bit old, but, they talk about the process if you are interested: freebsdfoundation.org/freebsd-uefi-secure-boot

Title: FreeBSD UEFI Secure Boot | FreeBSD Foundation

thanks I believe I tried something like this and got stuck or something maybe that one is updated

I'm just upgrading the laptop to 13.2

Allan: one day, when PkgBase hits, we can just add WITH_BEARSSL as a standard, imagine that

when is "one day"?

yuripv: my reading is that it's planned to go beta in 14.0

yay

hackmd.io/JczFDHtiQYSeEyeK9182jw?view needs more work tho

Title: FreeBSD 14.0 Planning - HackMD

any way to query pkg info for only packages that are not dependencies (i.e ones that were installed with pkg install)?

yes

pkg alias --list has something, which i cannot remember

i got it i think

I think it's leaves?

i have no aliases

nvm i need to copy the sample pkg.conf

noauto

sweet, thanks! meena

btw i made a forum post about my uaudio problem

no one cared to comment yet

i'm about to stay up all night and dig into the kernel driver and start hacking it up

ah, to be young again…

i'm old

yes, but, is anyone actively leeching the marrow from your bones, every waking hour of the day, and sometimes at night, too?

yeah kinda

i don't know what kind of child eats bone marrow though

unless i'm missing the poetic reference

it just feels like it

it's poetic, alright. except, in utero: if a foetus doesn't get enough calcium, it does leech it from the mother's bones

congrats on the news

it gets worse

anyway, our kiddo is four, and, yes, i know

oh nevermind then

thought you had a marrow-eater in the oven

I've just updated the base for several jails. all went well except for one jail which is no longer able to update packages from a remote repository due to: 'Certificate verification failed'. All other jails are happily pulling from the same repo. I am not sure how to resolve this. I tried to workaround by bootstrapping from pkg-static but that didn't work either.

where does one usually go from here?

why is it failing, exactly?

not sure. it seems that the jail is lacking ca_root_certs tho.

can't install it easily tho as that would happen via pkg :p

huh?

is this FreeBSD's repos, or your own?

my own. but shouldn't matter. all other hosts & jails are happily working with that repo.

yes,

it's just that FreeBSD's base system supplies the key for the FreeBSD repo

yeah I have my own cert installed. same as on all other hosts & jails

i would use a downloaded/ cached ca_root_certs from one of the other jails and manually install it into the jail, and try again

Certificate verification failed for /C=US/O=Internet Security Research Group/CN=ISRG Root X1

34378686464:error:1416F086:SSL routines:tls_process_server_certificate:certificate verify failed:/usr/src/crypto/openssl/ssl/statem/statem_clnt.c:1921:

can curl/wget/fetch verify it?

sure, I just successfully fetched the HTML of the poudriere dashboard form that jail

actually, it's ca_root_nss and that is installed (brainfart with ca_root_certs)

wtf, I am getting the same error when doing a 'pkg install' on a .pkg

pkg-install -U did the trick. just updated pkg from 1.19.0 to 1.19.1 on the jail but still getting the same error :(

pkg add would skip some things that install does

but that doesn't explain why you're getting those errors

I've manually installed ca_root_nss from the jail's host in the jail and now things are back to working :s

weird

not sure what happened there

meena, also thanks for the pkg-add note

jbo: does the log tell you anything?

Having fun updating python before moving on to FreeBSD 13.2. I used Poudriere to pre-build the new packages. However, pkg won't admit that Poudriere had built nextcloud-php81. Now manually (re-)building.

Schamschula: what does poudriere testport tell you?

hi all, how can I use lock(1) [lock -npv] with zzz(8)?

write a wrapper script?

im going to switch from debian to freebsd soon

lockzzz() { lock -npv & && zzz }

concussious: something like that ⬆️ ?

puddinghead: cool

any advantages of freebsd over linux? aside from the whole approach to freebsd as well as the so-called customizabilty

from what ive noticed everything i run on linux should also be running on freebsd, even the emulators

but im still a bit worried about running freebsd alone as opposed to dual booting for non-emulated games and photoshop

(not like i havent used photoshop in a looong time)

@meena thank you so much

puddinghead: my favourite part is that i know where to find the source, cgit.freebsd.org/src/log (and how that makes contributing back to it very easy)

Title: src - FreeBSD source tree

concussious: thank me when it works. that's untested, written on my phone

@meena, ok I'll go test now

@meena nice

im also a bit worried when it concerns hardware compatibility

mostly because i have an nvidia card but still

does the 3070ti work on freebsd?

i mostly use FreeBSD on servers, and virtual machines. I had one Lenovo laptop that i used for FreeBSD development which I bricked (well, not really bricked, just upgraded into a non-working state)

either way, I'm not allowed to touch real hardware, so you'll have to ask someone else

i see

id also like to move my vps from debian to freebsd tbh

kind of tired of how debian forces me into dependency hell and whatnot

There is no escape from dependency hell

for real

we also just cook with water

based

i hate oil

(ELF, but we have slightly more symbol versioning in base)

puddinghead: yes, there's binary nvidia driver for FreeBSD

nice

how good is it?

(not as much as Solaris or Windows, but definitely more than a standard Linux distro)

it's lacking cuda support, otherwise should be the same as in linux

poudriere testport seems to indicate everything is in order. I installed lots of other packages w/o any issues. Everything is back up using manual builds.

puddinghead: the nvidia driver is provided and maintained by nvidia, as far as I know the only long-standing defect is lack of CUDA, and I saw that there may have been progress on that

meena, sorry, was busy doing laundry.

speaking of ports, I need to upgrade the GoToSocial port

i see

meena, haven't checked logs. it's a burner jail so also didn't care too much.

i think that my biggest issue with switching from linux to freebsd is the whole ai thing

i sometimes work with ai on my gpu as a hobbyu

not too often, though

you could dual boot

(or more :-) )

afaik bhyve recently gained the ability to pass-through GPUs

kinda sucks if you only have one GPU tho :p

or have a vm and pass-thru the gpu, oh

it's worth checking on whether progress has been made with nvidia in that respect

I'm just glad that now I can finally have a VM with more than 16 vCPUs. that was a real bummer.

yeah, i only have one gpu

What web/gui/tui/cli do u use for manage bhyve's VMs?

jbo

Tried ClonOS - missing many must-have features

VVD, I use cbsd (the guy that also works on clonos)

from the guy* (I'm not that guy)

I spend months testing different CLI tools to manage jails/VMs and eventually settled with cbsd - been happy ever since.

VVD, if your user info is to be believed the cbsd author is one of your commrades

jbo, do you know how to connect iso from different place to vm?

without any repositories and etc

and create VM without any templates - just create vm

I know about author and planned to contact him

sure, you can just dump ISOs wherever you want on your host and register that location as a local repository in CBSD. you can also mount them from an NFS share or whatever if you feel fancy

jbo, ye, I have NFS share with a lot of iso files - use for VirtualBox and VMware

> register that location as a local repository in CBSD

every file or just root of place?

once i switch from linux to freebsd, what should i use for my backup?

i generally use btrfs for linux backups, but i think that for freebsd i might as well use zfs as the backup format for my external hdd

VVD, I think you can do both. I remember having spawned some VMs with weird images I didn't want in my library so I just attached that ISO directly to the VM upon creation.

puddinghead, have a look at zfs-send and zfs-receive. that allows you to do a lot of stuff.

VVD, have a look at the 'media' subcommand of cbsd

puddinghead, I'm using zfs snapshots send/receive via network for backups 5 years

VVD, cbsd media --help

that sounds nice

my pc is pretty decent (16gb ram + 1tb nvme) so i should be fine just using zfs right

definitely. been running ZFS regularly on 8gb RAM machines and afaik 4gb works too (as general rule of thumb)

thanks a lot

just go zfs - don't even look at UFS for a desktop system if you're new to freebsd

zfs is epic

i guess thats yet another reason to switch to freebsd in addition to my grievances with linux and being able to keep a stable base AND installing the (almost) bleeding edge versions of programs at the same time

for servers better

:-D

yeah freebsd for my media server was what i was originally planning to do with it

puddinghead, we've all been there :p I too switched from linux to freebsd about 12 years ago. first only for servers. now also desktop.

yay!

once I saw how much nicer the freebsd experience is on my servers I decided to also go freebsd desktop and I have not looked back a single time.

using FreeBSD for servers from 4.x

for desktop from 7.x

I think I was still wearing dipers in 4.x times :s

my first freebsd was 10

that's a loooong time

my first freebsd will be either 13 or 14 depending on how late i end up switching

2.1.x installed from floppies :-)

first need to get rid of that pesky bios password

nice

puddinghead, a word of warning tho: hardware support is not as nice as it is in linux world. not a problem for servers, generally also not for desktops but if you plan to go freebsd on a laptop choose carefully.

jbo: thats what puts me off, tbh

yeah thats why im planning to get an old thinkpad if i do get a laptop

im on desktop rn

main issue for me are the peripherals tbh

ixmpp, I was initially concerned too - but haven't looked back :)

i cant even have it on my laptop because elan touchpad

i use everything wired but im stil worried if my mouse/kb/speakers dont work

just don't use crappy peripherals you don't need but marketing makes you think you doo :p

yeah

guess i'll rip out my touchpad :p

rip touchpad

who uses a touchpad anyway o.O

xD

anyway, my touchpad works just fine (synaptics).

puddinghead, what CPU are you running? Intel 12th & 13th gen support is pretty so-so. My last laptop upgrade was intentionally an 11th gen

ryzen 7 5800x

i am not buying recent intel processorws

not running any myself but from what I remember from forums/mailing-lists that should work fine.

yeah I think my intel times are also over... :s

yeah amd has no issues iirc

amd has had plenty of issues

yeah that seemed like a pretty non-solid statement

true

was mostly referring to cpu though

don't worry - we're used to it from folks switching over from linux :p

yes, amd cpus have had plenty of issues

RhodiumToad, care to enlight us/me a little? I am considering going AMD soon.

yeah

jbo: used to what?

one of the most glaring ones from a few years back is that some amd cpus would break if you used the top page of memory

😂

lmao what

RhodiumToad, that sounds like the AMD I used to know :D

anyway, not running shitty software usually goes a long way in avoiding hardware upgrades

if your text editor requires a web server you don't deserve to have a good experience IMHO :p

that i can agree on

all the things i want from a desktop experience are already on freebsd, happily, because i keep things trimmed down

but then hardware fails me

heck, I am not even running dbus :d

amazing

i think i'd have been running it on that laptop for ages if not for that driver issue

ixmpp, no luck with porting the linux driver(s) via KPI?

that's what is currently going on with iwlwifi too

heh. tried for so long to avoid dbus, but its just not reasonable to, on linux, anymore

huh, i dunno actually, i never dug far

yeah, once your init system requires a DNS implementation any argument falls short I guess :p

oh at least that is still avoidable

for now

I was amazed the other day when working on the hardware design for an embedded linux project that systemd won't do with 128MB total system RAM out-of-the-box

NICE

ive actually used linux without systemd more than with systemd on my desktop

rn im on devuan for desktop and i actually started with artix

i think i only put up with systemd briefly

seems about right

i like the convenience on my server but on my desktop? not as much tbh

systemd on the desktop is kind of bloat tbh

so what you want is just a good service manager? or, what draws you to it on a server?

ease of use, really

as well as good service management, something that gets stuff out of the way

Super hard to avoid dbus in my experience. Most login managers depend on it. Chrome depends on it (afaik).

systemd uses more ram than id like to but its the price to pay on my server

on desktop doesnt matter as much since i dont prioritize memory management as much as on my servers

ox1eef_: if you use a light login manager, you avoid that first one, and obviously yeah don't use chrome. my problem was every small app under the sun these days assumes dbus, even if for dumb small things like libnotify

its whackamole

i dont even use a login manager these days tbh

it seemed smarter to cave and have one controlled dbus broker, than every app starting one

Yeah, I wanted to avoid it as well but I eventually gave up. I was using Slim at one point. Now I just use startx.

same, except i actually stopped when i got lightdm replaced with slim

then i tried tbsm a bit

but uninstalled because i noticed i just had xfce and therefore had no need for anything more than startx

i think there's some pam-based stuff that ends up easier with a login manager - console sessions don't create a runtime dir, graphical ones do. though, on freebsd i just loaded pam_xdg so maybe that doesnt apply there

On OpenBSD I use xenodm. It would be nice if that was ported to FreeBSD. I'd most likely use it instead.

i see

i reccomend greetd, or Ly

both are lovely

yeah ive been considering greetd

text-based login manager, every time

yeah

Ly is nice indeed. I used that at one point too.

unfortunately, i use pam_yubico, and Ly doesnt support that

so yet another pain

(greetd isnt ported to freebsd)

i tried ly

prefer tbsm tbh

but idk if its ported to freebsd

doesnt seem so

All of it is candy anyway. startx is all you need.

yeah lol

xdm ftw

* puddinghead has used xfce since she first switched to linux

i am pretty set on wayland. i do think it was a good move

if i wasnt on novideo id prob be using wayland tbh

though i take it thats unpopular here

puddinghead, the way to do that is by using /me followed by whatever message

i see

so itd be like this

hah, found the amd last-page thing

"Lower the amd64 shared page, which contains the signal trampoline, from the top of user memory to one page lower on machines with the Ryzen (AMD Family 17h) CPU."

ixmpp, wayland is just one of those linux things that tries to solve a problem that doesn't really exist and hasn't solved it yet. plus terrible support from nvidia's side.

jbo: luckily, Val is very interested in it, so it might become more usable in time for us

granted, i don't use nvidia, but i still appreciate it

and frankly, nothing wrong with competing standards, right?

we'll see. I am not opposed to new things in general. but until they actually work better than what has been there for decades I see no reason to switch. leave the hype to linux users :)

meena: it's already pretty usable, from what i've tested

ixmpp, let me guess: you just installed a desktop and used it normally?

yes?

hardly what matters to the industry

all I've seen, be it Linux or FreeBSD, was pretty poor

that ^

wayland already is and has been used by the industry some time to be fair

it's like a second system effect in reverse

phones, car screens, netbooks

ixmpp, same story as with systemd, pulseaudio and all the other "crap" (excuse my language). a lot of things are being used because the workarounds have become more painful than just accepting the degrataion of quality.

(and actually from what i understand, in those situations it was used because it actually was better)

something that FreeBSD luckily handles very differently.

heck, I remember when debian had to gave in and switch to systemd... they didn't do it because they perceived systemd as the better system.

how does freebsd handle X differently?

it doesn't. the "handle it differently" was in regards of not jumping on every new hype emerging from the linux world.

linux changes fundamental system components more frequently than the average european changes cars.

or the brits their PM :D

i get it

ha, ouch

do europeans even use their cars?

too close to home

the ones in the cities at least

puddinghead, we try to avoid it when possible. I live in a remote mountenous area so live without a car is pretty difficult

i appreciate that it is a pretty loud linuxism, but of all the ones of late, its the only one i'm actually sold on

but my car usually doesn't see action for more than once or twice a month

i see

ixmpp, I can see how wayland is definetly one of the least affected - indeed.

again: I am not agains change. I am against changing things constantly just because some community needed another hype to fuel their self perceived value rather than actually improving things.

it seems to have become a trend to replace prooven things rather than extending and improving them and then eventually figuring out that it's way too much work after replacing stuff that has been working well for decades - and then moving on to the next thing doing the same again :D

Yup. That's my biggest issue with Linux generally. They want to replace everything that came before with complex, and complicated alternatives that are so far from simple that it is almost sad.

for real

linux stop switching audio systems every 5 years challenge (difficulty: impossible)

yeah that one hurts

sometimes you do need to replace something, but you need to make sure the replacement is actually both better and simpler

jup

look at geom. still going strong :D

and pretty much anything else in base

and then we didn't even start talking containerization technologies...

meanwhile, jails still rule.

RhodiumToad, would you happen to know whether there is any progress on root-on-zfs with boot-from-encrypted-dataset?

progress from where?

I thought that was a spanner in the working

afaik there was some plan of adding dataset decryption support to the boot loader to encrypt zroot

oh, with zfs native encryption?

yeah

dunno. haven't been looking.

I ditched GELI one zfs native encryption became a thing but can't boot from those.

bug*

anyway, jbo do you have resources for attempting to port a driver with linuxkpi? maybe i ought to try

at least on that laptop, thats the only thing holding me back

ixmpp, have a look at drm-kmod and iwlwifi. they do this successfully (with lots of KPI work happening every week)

ixmpp, just don't get confused by the linuxulator thing. there's a difference between linuxulator and LinuxKPI. the later is targetted for interfacing linux kernel drivers

i see

yup, already aware

great :)

im now wondering if sound will work too, because even linux struggled with audio drivers for that laptop until i ripped the firmware out of chromeos

what abomination of a laptop do you have?! :D

acer chromebook flip C433 (shyvana) :p

alright :D

chromebooks are great for value for money

audio card is MAX98927

ixmpp, grep on source shows results in the device tree. no idea whether it actually works tho

heh

the device-tree text file lists it as "compatible: should be one of the following [...] MAX98927"

ixmpp, you can always boot from a usb drive and test everything that way

yeah, i was thinking that, i have a usb mouse

well, shouldnt need one i guess

surely RhodiumToad is still using a PS/2 mouse

actually I am

knew it <3

gee i havent seen one of those in over a decade

but I do currently have a usb keyboard (not entirely by choice)

there's a ps/2 keyboard plugged in as well (I think the ps/2 mouse port doesn't work without it) but I'm not using it

did you require n-key roll over for more efficient blogging? :p

no

that was an unnecessarily snarky question - sorry. I'd be geniuenly interested in learning why you switched to a USB keyboard involuntarely.

mostly due to having to switch keyboards around without rebooting the machine

I wish I would have been able to experience the glory days of computing.

plugging in a usb keyboard was less disruptive than trying to get at the ps2 ports (which are not exactly hot-plug-friendly)

reasonable

ps2 peripherals sound so nice

kinda want to switch from usb to ps2 mouse noww that i think about it

mostly because i want a mouse without rgb lightning but whatever

the best mouse-related thing I did in my life was switching to a trackball

I believe the latency is still lower on ps/2 ports

i c

this, im intrigued about. i love my fancy keyboard and mouse

puddinghead, just buy peripherals without rgb o.O

having 12 extra buttons on a mouse is something i wouldn't go back from

keyboard is just fancy enough to have a backlight, its a very cheap reddragon otherwise

and im fine with it rn, gonna switch to a more expensive keeb if it ever dies on me

just seems like ludditism to go back like that

when it comes to peripherals, I just use my general strategy of "buy once - but buy well".

i can live without the rgb part, but the buttons no

i mean, what is your rationale for preferring ps/2, it seems we already detailed one reason it's a pain?

fwiw I can't find evidence of support for the MAX98927 on freebsd

damn.

RhodiumToad, it's only mentioned in some text file in /sys/contrib/device-tree/Bindings/sound/max9892x.txt

it seems to be an i2c device

would make sense, touchpad is also i2c

RhodiumToad, most audio codecs & amps have an i2c sideband interface to control stuff like volume etc.

and then usually SAI for the actual sound interface (or USB or whatever)

(also, trackballs break so often, do you really think that's worth it)

lord i don't wanna go back to the ps/2 trackball era

what breaks about them so often? just buy a decent one. same as with a mouse.

I'm using a USB trackball - just use whatever you like.

moving parts

the disappearance of mouse balls is something I very much appreciate (not all innovation is bad :-)

its inevitable, eventually

the only moving part on mine is the trackball itself which is a solid sphere. everything else is static.

oh, it uses cameras or something?

yes.

interesting

it's literally like any modern optical mouse but then just tracking the movement of a ball rather than its movement relative to the surface you use it on.

now that i think about it

that i understand, then

does any virtual drive software exist for freebsd?

puddinghead, define "virtual drive"

some program that allows you to mount iso files

sure

mdconfig

fuse exists, so

i see

ofc

is mdconfig the default mount anything program or is it just for images

md = "memory disk" (but it actually does files as well)

it's a memory disk - anything you want.

mdconfig creates disk devices which are backed by wired memory, swappable memory or files on any filesystem

bonus points for doing a memory device over rclone

the created device works like a disk for all purposes

just found this in some of my code:

y -= 7.0; // No idea why but it's 3am in the morning and fuck this shit.

not good.

i see

also, do appimages work on freebsd or are they linux only?

pretty sure its the latter but i still wanna check to make sure since a thing or two might be appimage only unless i try to compile 'em from source

appimages are a linux thing

and pretty much anti-unix in every perceivable aspect

seems about right with all the alternate packaging methods currently trending on the linux world

snapd and its suckstemd dependence, flatpak's lack of quality control, etc

the way to install 3rdparty software on freebsd is via ports. anything that has an existing port can be installed via pkg install foo

yeah i am indeed considering ports

you don't get a choice. FreeBSD tends to be concise

i like being concise

I mean sure, you can build everything locally and install it that way but that is pretty much what the ports framework does for you.

yeah, thats what i like about ports

well, some ports do not have packages, but those are a minority

pkg install installs from binary though, doesnt it?

generally due to licensing issues

or does it install the port with the exception of some big packages

and makes sense

pkg install installes a package.

the ports tree defines how to build a package from the original source.

i see

puddinghead, by default, pkg installs binary packages built on the official builders. however, you can also build your own repositories (where you compile everything yourself). I one wrote a high-level beginners tutorial for that

nice, where can i read it?

a port is basically just a recipie from which the port will be compiled

a bunch of machines at freebsd.org spend all their time building ports and putting the packages into repositories

puddinghead, blog.insane.engineer/post/freebsd_poudriere_guide

Title: Poudriere: A complete guide

puddinghead: ftr, appimages are just packaged binaries and libs, you can extract them if youre desperate, or just build it?

ixmpp, and end up with a bunch of executables that make system calls that aren't available?

i think the only thing i still use appimage for is ledger live, because it's proprietary and there's bugger all to be done about that

jbo: i meant on linux

or, linuxulator

RhodiumToad, I changed my poudriere config the other day. I was a bit done with waiting for some ports to compile on single-core while all other builders are idling

now I am looking at load averages of 20

@ixmpp only think i really need appimages for is pcsx2 i think

pcsx2 is open source, isn't it?

yeah

retroarch is on ports, dolphin if i ever need to use it is on ports, rpcs3 has official freebsd releases

mame is on freebsd too and i think ill need that soon

jbo: on that topic, am i losing out on much by using portmaster and not poudriere, i've yet to comprehend what poudriere offers entirely, but it seems more complex to set up

poudriere is the bee's knees

and I say that as a long-time portmaster user

bee's knees?

+1 for poudriere

its a british idiom :p

fair enough, will try, on my server at least

I never used portmaster so I can't tell but everybody I know that used portmaster switched to poudriere eventually

ixmpp: poudriere's big win is better parallelism and clean builds

jup, poudriere does all the building in jails

neat

very neat.

also better handling of dependencies, portmaster notoriously keeps recursing through the same things repeatedly

whereas poudriere sorts out the dependency list once at the start

while im at it, is there a way to globally enable debug symbols, with e.g. poudriere config?

I think poudriere is one of those cases where something newer actually solved problems :p

ixmpp, yeah, you can have a global make.conf

WITH_DEBUG in make.conf

nice

gotcha

i've wanted to work with jails since forevers

seems like a better docker (especially given the recent controversies)

just do it.

note that poudriere has its own make.conf files, since they may need to be different from the host ones

I'd argue that jails are a core concept/feature of FreeBSD.

certainly much cleaner than lxc

really? that sounds cool

also, poudriere can do cross-building nicely

nice

RhodiumToad, although dealing with qemu-static is apparenlty at times painful :D (if you want to do aarch64 builds)

oh, I know

currently reading the tutorial and they say that pudriere should be ran in a separate server... i dont have any

just my desktop

it doesn't need to be separate.

really?

puddinghead, my tutorial? that should just be a recommendation. you can do it well on your local machine and still get all the benefits for free.

thanks

I know plenty of people running poudriere on their desktop because they don't have a dedicated server

however, if you don't disable or restrict poudriere's use of tmpfs, it'll use a shitton of memory

how much should it be compared to actual memory

yeah, that's why my poudriere server runs 128GB RAM. but that is certainly not a requirement at all.

(it's an option in poudriere.conf)

like, if i have 16gb of ddr4 ram

what hould the limit be?

the limit would be 16gb

really?

i expected like half that

if you only have 16gb then the limit is 16gb, yes :p

I mean the base OS uses like what? 300MB?

my first poudriere server was running happily with 16gb

if you let poudriere use tmpfs for workdirs, which is the fastest way, it can use a lot more than 16GB

yeah, mine is currently consuming 103 GB :x

but I set tmpfs to 'all'

puddinghead, the takeaway here is: you are not restricted. poudriere will run just fine on your desktop. just go with the default config when you start out.

definitely DO use ZFS tho.

er, I'd say turn off tmpfs use rather than using the default.

yeah im 100% gonna use zfs, been wanting to try it out and might as well try on desktop before i get my media/datahoarding server

puddinghead, you can also put your poudriere server in a VM (eg. when you run your desktop on freebsd just spawn a native bhyve VM). that way you can easily limit memory usage.

i see

so i could limit it to like, 8 gigs if i wanted?

sure

I'm currently running it in a 4.5GB vm

and then how would i transfer my poudriere builds to the actual pc

puddinghead, my tutorial shows all of that.

my issue with vms is disk storage just as much as ram, ive been starting to hoard data so its become an increasing worry (and im planning to get a second drive soon)

cool!

the normal way is to setup an nginx webserver on your poudriere server (which very well may be a VM). then just setup your client (your host) to consume that repo via HTTP(S) and you're done (also shown in that tutorial).

i see

do i need to do any fancy stuff like open up router ports or nah?>

I'd recommend you just to get started. things will make sense soon.

is `ln -s /etc/make.conf /usr/local/etc/poudriere.d/make.conf` a bad idea?

unless you want it to be accessible from outside your host no.

ixmpp, I'd say so :D

damn :p

host is the hostname right?

as in the local network?

ixmpp: it's not quite what I do

ixmpp, you'd most likely want a different make.conf for your ports than for your host.

ixmpp: what I do is this: /etc/make.conf has only .include lines for make-common.conf, make-default.conf and make-${MACHINE_ARCH}.conf

that's the pro setup. generally I'd just recommend to get started. you can always tweak & improve later.

ixmpp: then in poudriere.d, default-make.conf -> /etc/make-default.conf, etc.

ah, cool, clever

that way I can have poudriere.d/amd64-make.conf -> /etc/make-amd64.conf, and likewise for i386 and armv7

i c

ixmpp, puddinghead you might also want to consider using the poudriere-devel port as the development version has a setting which allows to fetch packages from the official binary repos IF they match what poudriere would build otherwise. this can safe a lot of time & energy if you just run a small setup on a desktop.

oh, really?

i just have to pkg install poudriere-devel instead of the regular one then, right?

yeah, the real annoyance with poudriere builds is having to build things like rust and llvm, which take ages

all my desktops that run poudriere locally (for ports development) use that feature. only my build server doesn't (because I just want to build everything myself)

@RhodiumToad can relate. compiling firefox is like hell.

jep. and changes are that you just want to use rust, llvm and so on the way they are in the official binary repos anyway. in that case poudriere will fetch those instead of going through 16 hours of building them.

honestly all of this makes me want to just upgrade ram before storage

just gonna go from 16gb to 64gb

worst thing is that you often end up having to build multiple llvms

oh yeah, my poudriere is currently building llvm12, llvm13, llvm14, llvm15, llvm16 and llvm17-devel.

so you'd want to fetch those from the official binary repos instead in your average beginner case :)

looks like this build I'm doing will include llvm10, llvm12, llvm13 and llvm15

and I have no idea why llvm13 is in there