i'm saying it's not supposed to be a raid0

oh no. thats not goign to change

ok, if that's the way you want to do it, you're own your own

Now I get it. You proposed raid1

that would be best. I know. better raid 10 but can't have it.

at least for now.

now you get it, after an hour?

board, nvme card, etc

rtprio sorry :)

its my 3rd day with freebsd

and zfs

thank you! appreciate.

In a mirror, if one side is a slower disk(s), then the faster disk will be read/written at max of the lower disk speed? thus making no use of mirror/raid1 efficiency? thus no point in having faster disks  OR in other words, "don't put slower media with fast media in mirrors"

i wouldn't mix slow and fast in the same mirror

will the effect be same as I wrote?

it will destroy the core efficiency of faster medium

Letiute: basically.

regardless what kind of raid you're building, you want the same kind of drives. same type, performance and size

yes ideally

thank you!

you can stack different parts of the system onto different types of media, but it makes no sense to build a single system out of different base components

they've been bouncing between this channel and #zfs all weekend, it's been... something.-

imagine… building a wall, with different bricks, but the bricks don't even almost match up, and I'm talking to myself

meena: and the bricklayer doesn't seem to listen to you

and here we have someone trying to reinvent it, poorly

this is why kids need to play with Lego before you let them have a go at video games

i mean i had a frankensystem with a half dozen mismatched hard drives and did some stupid tricks

but only used it for test data before going out and buying six identical drives

debdrup: you may say that again!..

I was always a read up on theory before touching hardware type, because hardware can explode…

I haven't been in the zfs channel, but that one has been, as you put it, «... something»

i especially love the need to always hilight someone when another thing crops up

i'm fascinated that people have been willing to put up with that.

is this the fabled support vampire?

meena: yes.

there is a risk in using labels like support vampire, in so far as how they relate to highly technical fields

it'd be quite unreasonable for someone only 3 days into a technical field to be expected to give back

what rubs me the wrong way is the constant asking for advice, then throwing away half of the advice because it's not the answer they were looking for, which i fear stems from a notion all too common among experts in some subject matter (which i'm also occationally guilty of myself, i might add)

but, like, are they an expert in their field?

also the bouncing between places to get half-answers

i would be surprised if they aren't an expert in _a_ field

i don't know them, i can't speak to what they might be experts in, but i think i've seen enough to recognize it from when i do it

another thing that rubs me the wrong way is when they refer to using the handbook and manual pages; the handbook is a fine place to start, but we all know it doesn't (and can't) have all the answers - whereas the manual pages are more of a reference work that you use when you need to know the exact incantation to invoke the elder gods of awk and grep, or some such

i _wish_ i'd had allan and mwls zfs and advanced zfs books when i started, i'd have _devoured_ them whole

and i have no clue what do about any of it, but also don't actively wish ill upon the person in question, because they're trying and it's hard to fault that

so now i'm just having a little winge in public, i guess :v

This isn't the smoking gun, but it is the smoke.

On FreeBSD, for some reason or other (while I'm inclined to believe it's my fault somehow, I have also explicitly configured hw.acpi.cpu.cx_lowest=C8 so it SHOULD be letting the CPU drop unless it's some kind of quirks mode), while the cores drop to C7 just fine, the package doesn't go under C3. On Linux, pcm tells me that there are package drops as deep as C8.

(this is still only the smoke)

what do you mean by package drops in this context?

It's just inaccurate language. ignore me

I'm just mumbling into the channel

I'll stop dumping thought process here

huh.

HUH.

 

Hi all, I have a crapton of stuff in all my jails' /var/spool/clientmqueue ; how can I make sure I either a) don't ever hear or read about that ever again or b) deliver mail to an email address that actually can receive email? (postmaster⊙jfd doesn't, conenction refused). NB: sendmail_enable="NONE" in all rc.conf files of all jails

I was searching for utf-8 support on the console @ lists.freebsd.org/pipermail/freebsd-current/2009-July/009351.html. Does "he kernel will use a table

to remap all Unicode characters to CP437 (the default VGA font), so it's

practically useless.

sorry got posted before i could finish type.

moviuro: I reckon it's stuff that cron or periodic produced

meena: peeking into files, it *looks* like periodic (not cron)

Does this statement " The kernel will use a table to remap all Unicode characters to CP437 (the default VGA font), so its practically useless" mean that there is no point following the work around for utf8 support on console in freebsd?

that sounds very outdated, vt supports unicode

moviuro: gist.github.com/c77ed494e102977c9fd06ce9b053cda0

Title: disable sendmail on FreeBSD, completely. · GitHub

yuripv: oh okay, sorry I am new to freebsd. Thanks for the info.

thank you meena , I'll purge those directories and disable everything. I'll be back in a few weeks if my machines still act up

alternatively, you could disable periodic if you don't need it

yuripv: i was just installing zsh and tried powerlvl10k customization and when i set console to use zsh, the icons show as boxes. what could be the reason?

periodic_enable="NO" in rc.conf(5) ?

sozuba: missing glyphs in available console fonts, i'd guess

Missing console fonts

yuripv: parv got it , thanks :)

moviuro: probably. check rc.conf(5)

anyelse can help me to debug www/redmine50 ?

Title: PrivateBin

is rubygem-bundler installed?

yes

Does md5 (or sha256, etc) command read a chunk of a file, calculate intermediate result; read another chunk, calculate another result; at the end of reading whole file, spit out the combined result?

parv: given the nature of the algorithm, i would say: 🤷🏻‍♀️

i think, for sha256 anyway, that each chunk is combined

until the end

white-wolf: what's the trouble with building? what options are you setting?

I think I'm going to slo to geep

default options

use passenger en nginx

in my past i link my tutorial based

white-wolf: if you're using default options, then you might as well pkg install

What does `bsdinstall jail /foo/bar` actually do? Just downlads and unpacks selected tarballs?

ccx: essentially… also some config stuff

I want to automate installing jails and I'm wondering if I should bother trying to automate bsdinstall or just script what it does.

ccx: IIRC, bsdinstall can be scripted, but most jail managers I've read the source code of do it themselves

Title: PrivateBin

kthnx; I don't think I care for any of the configuration it does anyway. But having something that fetches and verifies current base.txz could be handy.

with the tutorial, use posgrsql and mginx

with passanger

white-wolf: i think it's mysql-server but `ls /usr/local/etc/rc.d/` and see for yourself

why don't rbenv / rvm work on freebsd

Is there a way i can use a nerd font (meslo nf) on vt console? I tried converting the relevant .ttf to .fnt to be used. But failed. I am trying to us powerlevel10k theme of zsh and when using zsh by default, the vt console tty shows empty squares/

rtprio: nobody cared enough to make it work?

meena: rbenv is in ports, but naturally won't install the version of ruby i need

sozuba: i really doubt it

rtprio: thanks for the response. any idea what could be teh reason?

sozuba: what failed?

meena seeting a nerd font (meslo nf) as default font for tty to diaply the glyphs/icons for the powerlevel10k theme of zsh

setting*

but the conversion worked?

meena: yes it did, but when i set it using vidcontrol, i get "bad magic in font file"

---

'

then maybe it didn't really work?

tsoome has been summoned

yes:)

meena: oh okay

this mini-me is doing his thing:)

Is there a public log of this channel?

not that i'm aware of

Aye

i think i have some logs

like a better connection, or an irc bouncer

rtprio, paste.blues-softwares.net/?9358e20c…FxckPzMVLLr5R4VWvvsrUkFQ3XWW9xJ6iUZ

Title: PrivateBin

oracle cloud has free freebsd vms

small ones, but enough for irc

rtprio, Thanks; will look into that

white-wolf: you need to run bundle in the source directory, and you shouldn't run it is root

is there a package for redmine50 ?

yes pkg

rtprio: what, exactly, are you trying to do?

install redmine 5.0? with what database? how? ports / pkg? (use pkg if you're doing default options)

(i am trying to get my mastodon working again, after a pkg upgrade bumped ruby to 3.1 and all hell broke loose)

at some point the default ruby version was bumped also

rtprio, (I might be following you already) What is your account (when it would be working)?

@elliot⊙mn

isn't there something in make.conf to keep the ruby default back

rtprio, paste.blues-softwares.net/?be6ed2ff…prmQJDeDKaYkhifKULAUZWinmChL9g9moeY

Title: PrivateBin

rtprio, Python & Perl do (DEFAULT_VERSIONS+= python=3.11 python3=3.11; perl has been recent enough for me ). Check ports/Mk if Ruby has similar

white-wolf: you're crossing the streams here; either build it from ports or build it yourself. it's going to get very messy trying to do both

sorry can you explain ?

the port is supposed to do those commands

have you tried installing the package? `pkg install redmine50` ?

yes as you saying

Title: PrivateBin

have you checked our log files for details about the error?

Title: PrivateBin

meena: i think i'll switch.

white-wolf: is thre a gemfile in that usr/local/www/redmine ?

yes

set RAILS_ENV=production

i think you want export RAILS_ENV=production

i do

meena: haha, there's a gap in my timeline from 5 minutes ago to Feb 12th

rtprio: 😬

rtprio, how restart to zero ? purge all pkg and config file and retry ?

white-wolf: i don't think you need to do that

just change 'set' to 'export' and try again

export: Command not found.

are you using csh?

yes i think

i don't know man, i hate ruby; all the cool kids are using gitea or whatever they call themselves now

computers were a mistake, sand was never supposed to think

# echo $SHELL

/bin/csh

csh, is an unusual choice, but ok

bundle exec rake -T

does it print the list of tasks?

Title: PrivateBin

... that's not what i typed at all

slow down

Title: PrivateBin

you might need to db:create the database first. assuming you've alrady configured database.yml

ccx: man.freebsd.org/bsdinstall(8)#SCRIPTING i think this is still pretty neat no matter how limited

Title: bsdinstall(8)

Title: PrivateBin

freshports.org/security/pam_rssh this looks quite useful

Title: FreshPorts -- security/pam_rssh: Remote sudo authenticated via ssh-agent

does that work with our OpenSSH?

rtprio, after ?? in localhost:3000 i have always the same error

meena: it uses ssh-agent, so i'm not sure i understand the question

ah, right, so, uhm

yes? go on

full disclosure, i've not used it at all, i just saw it mentioned on mastodon

so when sudo wants a password, it asks pam, which in turn asks the socket and then you have to press your Yubi key, and everyone is happy?

romain linked z4yx/pam_rssh #8 there, which is how i found out about it

Title: Ready for distribution / production ? · Issue #8 · z4yx/pam_rssh · GitHub

meena: that does seem to be the i dea

meena: presumably it requires an openssh new enough to support yubikey

that's what i meant, i don't think we have that in any release right now

we don't?

and if we do, i think we stripped the fido libraries for some reason. my memory is foggy

is it just FIDO or is it U2F and FIDO2 too?

10:44 <meena> my memory is very foggy

Title: FreeBSD / src / d0f1f38202c4 - FreshBSD

it looks like it's also something that's actively changing

freshbsd.org/freebsd/src/commit/a613d68fff9af037 has been merged to stabe/13 this febuary, so i think it should be in 13.2-RELEASE?

Title: FreeBSD / src / a613d68fff9af037 - FreshBSD

Title: PrivateBin

debdrup: ooh, i missed that! cool

meena: i did too, i only just saw it now

I did see this bit the other day tho: "devd integration is not included in this change, and is under investigation for the base system. In the interim the security/u2f-devd port can be installed to provide appropriate devd rules."

rtprio, paste.blues-softwares.net/?f65d41ae…BzoEhLCFmNyBuXjBVaz8HhtKWReiWAVCSwb

Title: PrivateBin

rtprio, how can i restart all to zero ? i think when i do mistake need restart... may be need juste install redmine50 and servce onestart but do anything

someone told me here it is possible to write my own installation script for freebsd

i did it as sh script

but someone told me it is possible to script an installer itself

how can i do that

?

see the man page man.freebsd.org/bsdinstall(8)#SCRIPTING

Title: bsdinstall(8)

thanks

hm, seems like it doesn't fetch distribution files in scripted mode

well my sh script extract base and configure disk

i am looking for just gui probably

to choose some options and press install

to install os my way

you can use poudriere to build an image that contains the software (and configuration) you want using poudriere-image, the packagelist functionality, and the overlay directory functionality

it can build a zfs image which can then be updated using zfs replication streams of boot environments that're built later

switching to vt console and back to wm (hyprland) from vt console, causes a series of "QQQQQQQQQQQQQQQQQQQQQ"s printed on the terminal in the dm, does this happen to anyone?

oooks ike it happens only with kitty

looks like*

Is this correct? `zfs create zroot/pg-lg /pg-lg/`

Letiute: correct how? for doing what

want to create a dataset  for /pg-lg/ path and treat it with different compression level

meena ^

you should directly set options you want to set while creating then, see -o

ok

and it is the mountpoint property, not some path you give zfs create on the command line

it was zfs create -o mountpoint=/pg-lg/ zroot/pg-lg

I get default compression on. with zstd-3.   Is that correct?

is zstd-3 good for databases? i thought to make it zstd-1

oh, inherited from zroot

thanks

I throught you wanted to change the compression property?

I did. zfs set compress...

looks good now

btw, I heard default compresion level is zstd-3

so is that a sweetspot?

is that good for databases data?

or I should do zstd-1

or just try and see/observ

the default should be fine if you don't have a reason to use another, it is the default because it should work fine for most usecases

also, it's only inherited at creation time, IIRC, if you change zroot's compression after creating datasets it won't cascade down

I might be wrong, somebody correct me

nimaje so zstd3 is the default?

meena ok

Letiute: if you read the man page, it might have that info

it's hard to remember everything, and things do change over time

its 3

# ls -la /pg-lg/ -> drwxr-xr-x 2 postgres postgres 2 Mar 20 16:25 ./ -> postgres is the owner

create tablespace pglgtablespace owner leadgen location '/pg-lg/';

# alter database leadgen2 tablespace pglgtablespace; -> WARNING: could not open directory "pg_tblspc/12054613/PG_15_202209061/12054615": No such file or directory

Did I missed anything?

3.

I don't know half of what you're doing, so i don't know what you missed.

also, the only database I use is SQLite, because it doesn't require me to wear a ceremonial brown robe and dim the lights when using it thedailywtf.com/articles/Behavioral-Deficiencies-

Title: Behavioral Deficiencies - The Daily WTF

ok

that is to say: while we probably have a large overlap, this isn't #postgresql

meena: but the robes and the chanting are the best part

if you gotta be in a cult, at least make it one with robes and chanting

I found that this is a freebsd directory permission issue. Despite owner being postgres user, it does not work. Bu it does work if chmod 777 is there. Maybe more chanting will help

debdrup: I prefer white robes. looks a lot better with blood sacrifice once it inevitable turns that way

that seems like a lot of effort to get a red robe

Letiute: if you have to chmod 777, and it's not chmod 1777 /tmp, something is broken

what does /tmp has to do with it?

it's the only directory that should be world write-executable

so tmp was the issue? termbin.com/0qij

`1777`?

meena ^

riddle me this, batman.

maybe a new kind of permission is invented that I was not aware of

what's software raid methods are their in freebsd? Googling isn't helping, or I'm not understanding. I'm trying to avoid ZFS because of CoW

what is your problem with cow on zfs?

don't need it, and it causes fragmentation if using iSCSI and go above 50% ish full

Some would argue fragmentation on a multi-user OS is a good thing, as it provides an interleave.

my storage isn't made of IOPS

Hopefully a simple question, but is anyone able to suggest an editor or utility for freebsd that will present output without wrapping lines? Maybe there's a setting in vi to prevent it, but I guess I'm looking for low-hanging fruit for now.

gnuself: Here's a fairly low-hanging one: less -S

oijark: Wonderful, thank you very much!

gnuself: As for vi, you can :set nowrap

vkarlsen: I tried that, but it's not an option in mine when listing "set all"

gnuself: I see, the bundled vi doesn't have it

maybe an improvement from vim and not a vi feature

vkarlsen: no worries, I started there when beginning my web searches. Just trying to read rows of data on a vertical monitor... *the struggle*

hello folks

I'm trying to compile something and its telling me 'storage size of statbuf isnt known' when creating a `struct stat' structure for a `stat(...)' call. I am assuming its missing the correct structure definitions so i'd like to ask where its normally located under FreeBSD-13.1-p7

The version is also a RELEASE version

I think /pg-lg/ is a dataset itself. That is why this error is coming? imgur.com/NnA6Lli.png

hm, the man page tells me sys/stat.h, but how do you get stat(…) without that?

gzar: which function are you using> look at the manpage for that an inclide the headers

seems like the code missed a sys/stat.h include directive in a specific file

it works now, thanks for the help

Letiute: did you configure your database in /etc/rc.conf and then did a service postgresql initdb <your-db> && service postgresql start <your-db>?

nimaje do I have to do that everytime for every new db?

Letiute: does PostgreSQL know that it should be using /pg-lg/ to store its data?

meena tablespace should tell

nimaje I have a db already running right now. this was a new db though

nimaje postgresql_enable="yes"  is in rc.conf

I already have 2 databses runing in pg

nimaje do I have to initdb everytime for every new db?

nimaje do I have to initdb everytime for every new db?

sorry for repeat

have you tried #postgresql ? they might know

or else, their documentation. I hear PostgreSQL has really excellent documentation and a really friendly community

no idea, the error message read like it is missing some files and it seemed like you wanted some new seperated db, that's why I suggested setting it up in rc.conf and using initdb

pg people blame fs maybe but I am trying. looks like it just worked.. I restarted pg..

is it possible to run a program in 32bit mode ?

gzar: if the programme is 32 bit, and you have the 32 bit libraries installed (default for most supported platforms) then, yes, it's possible

where can i get a 32bit JVM ?

I think they stopped making those about 12 years ago

or maybe I'm wrong, freshports.org/java/openjdk11

Title: FreshPorts -- java/openjdk11: Java Development Kit 11

too much sun / oracle propaganda about 64 bit is better

that is available for i386

im asking because some companies still use 32bit jvm and i need to use their software

lib32 should make it possible

so java programs, shouldn't they just don't care if the jvm is 32bit or 64bit? (except for the cases they need more memory than a 32bit jvm can handle)

Hi I have problems with my probook 650 wireless card: 802.11 broadcom (couldnt find anything more specific). I have loaded the if_bwn.ko kernel module yet sysctl net.wlan.devices doesnt show anything. Is this just a unspported wireless card?

for some reason this application complains about not being able to use 32bit libraries on a 64bit jvm when it tries to load the SWT library

ElectricJozin: is it listed as supported on landonf.org/code/freebsd/Broadcom_WiFi_Improvements.20180122.html ?

Title: Landon Fuller: FreeBSD Broadcom Wi-Fi Improvements

ah, yes bundled dependencies

its hitachi pdi (pentaho-data-integration) just for the record

maybe you can install swt and get it to use the system one, instead of their bundled one, but if it doesn't work by just installing swt, then it is probably easier to install a 32bit jvm

debdrup, nope its a BCM943228HMB

just opened up the laptop and doubled checked it and it is the one

ElectricJozin: better hope it can be replaced with one that is supported, then

otherwise, you'll likely need a wifi dongle

debdrup, what chance is there that I can replace it with a wireless card from another laptop

I have another elitebook

ElectricJozin: no idea, it depends on the interface (mini-pcie or m.2?), and whether the firmware accepts arbitrary devices (not all firmware does).

debdrup, its minipcie and about the firmware - just try it out is the way to figure it out?

ElectricJozin: typically, yes

typically firmware locks hardware access by pci device ids, so it's also possible to modify the firmware (this is what's often done for thinkpads)

dch: howdy!

debdrup, and for something like this: Intel® Centrino® Advanced-N 6200, Dual Band

where should I look for driver support

ElectricJozin: iwn(4) should support that

I dissasambled two laptops and this is the second I found (the first had another broadcom)

debdrup, a guess?

ElectricJozin: it's listed in the manual page

wiki.freebsd.org/dev/iwn%284%29 also listed here

Title: dev/iwn(4) - FreeBSD Wiki

hrm, my connection seems terrible. I’m going to drop, set up a bouncer for myself, then I’d love to talk 13.2 on ampere.

debdrup, I see ok btw since I am kind of new to freebsd (10minutes new). Do kernel modules load automatically or do they all have to be loaded manually?

ElectricJozin: that depends, /etc/rc.d/devmatch is a script that matches every device against device identifiers and loads the module if one is found - but if you want to load something that isn't loaded according to kldstat -vv, then there's a few ways of loading it (which option you use depends on what you want to load).

the devmatch script gets run automatically as part of the system startup

debdrup, just replaced it and it works! thanks

rtprio  I think #2 (changed) will not have boot records so it won't boot despite being a mirror to #1? jamboard.google.com/d/18hni1wEORDsO…f8y7Mjur4in1ecJMuM/viewer?pli=1&f=0

Title: Storage topology - Google Jamboard

debdrup, one last weird question: for you personally if windows is the top of the iceberg and next is linux freebsd etc whats deeper than that. Whats more performant, more bare bones than freebsd. Whats the gentoo of the bsd world?

ElectricJozin: i switched away from windows in march of 2000, because windows sucked. used linux for about a week, then decided it sucked because of lack of documentation. i've been using freebsd ever since.

i'm not the right person to ask that question, nor do i think it's a question that belong in #freebsd

debdrup, right thanks anyway

as for your specific example, the only thing i know about gentoo is that it took its best ideas from freebsd, so if that means something then freebsd is probably the answer to your question.

i'm also really suspect of the iceberg analogy, since it mostly comes up in connection with conspiracy theories, and that's a whole can of worms that i'd rather not engage with.

Letiute: stop using raid0

yes sir!

how many times do i need to tell you

0

raid0 or not raid 0, my point is that if OS disk is mirrored, the bootlaoder won't be mirrored. If primary drive fails, The mirror won't boot. Correct?

debdrup, I keep getting wlan0: no link ... giving up

is there any other kernel module I am suppose to load

did you read the manual page?

I did

did you read the part about needing to load the firmware too?

yes I have if_iwn_load="YES"

in loader.conf

that's the driver, not the firmware.

the word firmware literally isnt mentioned in the whole handbook

yes it is.

wait, why are you talking about the handbook suddenly?

....where else should I look :|

hopsies

in the iwn(4) manual page, like i told you

it instructs you to load not just the driver, but also a set of things that include the letters "fw"

debdrup, so I included all of them now I get: iwn0: iwn_read_firmware ucode=xxxxx

still the same issue though

try running `dhclient wlan0` as root

debdrup, same

so the dhcpd on the other end is too slow to respond?

i assume you can use dhcp on other machines?

I do

I never noticed the wifi was that slow

ElectricJozin: can you try configuring the interface with static addressing?

you know, my wifi button is blinking kind of weird

:>

debdrup, the router surely doesnt allow that

i'm not sure what that means

the router doesn't get to decide that, i'm pretty sure

but it does get to decide if you can talk to any devices right>

?

no, not really

debdrup, so 2.4GHz works

the card is called "dual band" but hey

its atleast something

it might be dual-band in the sense that it's 802.11a that uses 5GHz

it seems the firwamre got automatically loaded btw

debdrup, what does that mean

802.11a is the first wireless standard, and used the 5GHz ISM band

neovim uninstalls obs-studio and vice versa. imgur.com/lxP8kAt.png

802.11b, g, and n uses the 2.4GHz ISM band

Letiute: the answer is in the image you posted; look for the integrity check with conflicts

How to check integrity?

it's something pkg does, and involves conflict checking.

it's happening because obs-studio hasn't been converted to USES=luajit

 pkg check -B obs-studio

Checking obs-studio: 100%

integrity checking including conflict detection is not the same as pkg-check

Title: 267656 – x11-wm/enlightenment multimedia/obs-studio: luajit-2.0.5_6 conflicts with luajit-devel-2.1.0.20221004_1 on /usr/local/bin/luajit

and bugs.freebsd.org/bugzilla/show_bug.cgi?id=267834 for the specific issue

Title: 267834 – multimedia/obs-studio: Update luajit to luajit-devel

no idea what to do here until wiat for a resolution from them

Q: freebsd does not have a system I think, What is the best way to make something like it where I run an an a a specific user, in a specific priority (nice level), rerun the app automatically if it fails/exits, run at system boot of course and turn  on/off via serivce (linux way: systemctl) ?

freebsd does have exactly that, in the form of daemon(8)

oh

and it had it long before systemctl existed

i don't remember if it can renice, and you'll need to integrate daemon(8) with rc.d to get it to run at system boot and be usable via service(8), but it's absolutely doable

if you're using rc.d, you can definitely renice it

yes, you'll have write a rc script

In linux we had builtin command like user=  nice=  srestartInterval (if killed) = etc I think (don't rmeember exact syntaxt though)

yeah, read `man daemon` for the options

here we have -R and -u

Someone's in the process of retooling daemon(8), after which it's possible that rc.d can be modified so that a restart interval can be added as a function.

Someone will have to write that, though.

having a cookie cutter rc to start daemon's would be extremely handy

It'd be better to have rc.d invoke daemon(8) if a restart interval is specified.

It'd involve a lot less rc script rewriting for one.

bsd.to/wSjb like that

Title: dpaste/wSjb (Plain Text)

would be fantastic

Hi

it's not possible to pass a disk device to a jail, right?

I was thinking more along the lines of ${name}_restart=N in the rc script itself, and documented in rc.subr(8).

yes that would also be helpful, if it knew how to back off

megaTherion: it can be done with devfs using devfs.rules(5) and a ruleset in rc.conf(5)

debdrup: ok, thanks

megaTherion: what issue are you trying to solve? I suspect you might be trying to fix X by doing Y..

well originally I wanted to pass a whole usb3 controller to an VM, but bhyve has a hickup with my HW *shrugs*

(thats however a different problem obviously)

rtprio: a lot of the current rc script that use daemon are pretty shite. they don't use daemon in supervision mode, so they couldn't even restart the service

pci passthrough can be used to pass the entire usb controller through

meena: right, that's what integration into rc.subr would hopefully fix

yes but FreeBSD doesn't see any DMAR tables, despite everything is properly enabled

i know

debdrup: that's a fine idea. but my plan, tho this might only evolve on a cosmic time scale, is still to burn rc to the ground

yuripv: do you have any idea how many years it took systemd to run in containers

hey i like rc

"A is somewhat like V, which is the default on Linux, which is why changing C to A is exactly like introducing systemd"

i will never forgive systemd for it's posix war crimes

i'd like rc a lot more if it wasn't 12653 lines of sh (rough estimate)

how many lines is systemd

rtprio: nobody complains that SMF or laubchd don't run on other Unix systems

the problem here isn't systend's bad opinions on POSIX. it's simply that it is badly designed and that critical application facing components aren't actually components

yeah, what stuck with me is i was tasked with changing ports of mysql; somehow systemd didn't get the message so it did it's own health check (!) and when it didn't respond on the default port it killed it and restarted until it timed out

systemd is just terribly scoped, designed, and implemented. also its documentation isn't great, and its community management was abysmal. but other than that, it's an okay idea

what sort of brain worms came up with that idea, seriously

Will never leave as long as it has rc.

The systemd (tm) is: It just works, iuntill it doesn't.

rtprio: SMF, launchd, s6, Windows Service Framework, and a billion others

AIX probably had something, too

meena: I just fear that with all the rc replacements that've been proposed and none of them having been adopted, it's hard to see how anything can replace it

the email guy wrote a program called svscan, which s6 riffs off of extensively

(the dissident email guy specifically)

SMF in Solaris doesn't really work without SFM.

that would be nice to have, too. but, baby steps

My point is you can't have SMF without SFM.

...at least from what I remember, though it's been WAY too long.

only in zones, tho i reckon that a zone SMF might talk to the system's FM

which from a design point of view makes sense. there's no hardware in a zone that can fail

fault management

fault management also tries to manage software failures i think

that's kinda what i was getting at, yes

yes, but it only aggregates them from SMF

meena: the docs for gotosocial are kind of shite

I recently got gotosocial working nicely on FreeBSD, had to write a bunch of rc scripts though

there's a whole _thing_ surrounding fault handling that needs to be implemented cf. handling different error codes differently (ie. some indicate that a program can be restarted, while others indicate the exact opposite)

if only there was a standard

it's one of the many things POSIX doesn't cover

the other big one being user accounting

the closest we get is errno(3) but that doesn't cover all the kinds of errors you can get from software iirc

amigan: is it worth using postgres with it?

Is sysexits.h known the world over?

rtprio: I certainly did, but this was for a production-ish setup (our local computer museum's AP server)

i don't know what it is, but that doesn't necessarily mean anything

It doesn't conform to any Official Standard

rtprio: you don't have to but moving from sqlite to postgres is hairy. I did it but I wouldn't recommend it

oh, sysexits(3) is what you meant

i haven't set it up yet; wondering which would be easier.

yes

just about to tootctl self-desctruct

rtprio: just do postgres off the bat. I love sqlite but for this, it is a testing database

sqlite is obviously easier just to try out

So I'm having a super weird issue. Setup: pf nat box with I210 NICs with that is also an nginx ssl termination proxy for a bunch of apps inside. Another box with I211 NICs that is running grafana. Only grafana has ever shown this issue, but I think it is exposing a bug lower in the stack. Originally, saving a dashboard in grafana would hang. It was caused by TSO mangling packets. I turned off all

hardware offloads on the I210 gateway. Many, many months later, saving is broken again, but the packets are arriving at the application server, which is doing nothing with them. Turning off all offloading (TSO, LRO, VLANHWTAG) does nothing.

I still suspect that the I211 is a f'ing me, but it's really weird seeing seemingly well-formed packets arrive at the app server (tcpdump shows) that are being ignored, and then you see the retransmits from nginx

i should probably install stock FBSD

AmyMalik: you absolutely should

because that'll help me chase this weird power virus issue - the question can be phrased as: is 1 watt the price of exploit mitigations?

On mobile, I say no

¿

1 watt is a lot on battery and not worth the lost utility. If you have a fixed power supply, things change.

Oh, so all this time you've been talking about FreeBSD, you haven't been talking about FreeBSD.

lolololo

I don't even know why I bother helping people when they lie by omission.

debdrup: at what point do i get banneh

+o damnit

AmyMalik: I wasn't planning on it.

oookaaaay

I really just wish you'd stop lying.

Well, i'll install stock, and let you know where I get.

I do agree, ^FREEBSD != FREEBSD

Please feel free to not do that.

welp

I guess I aren't long for this chat

rtprio: how can I help (re gotosocial, which, at this stage is still alpha, BTW)

appropriate hostname

i was annoyed that 'to build, see docs' then the docs say 'to build see contributing.md'

amigan: what's wrong with the rc script in the port?

rtprio: if you want my rc scripts, let me know. I have been considering making a port for gts, mostly for my own benefit

Wait, there is a port?!

Since when?

i would take the rc scripts, plz

Wow I feel like a moron right now. I could swear that I didn't see one when I first installed.

rtprio: use the port's scripts, I say tentatively

fuck i didn't even think to check

Title: FreshPorts -- net-im/gotosocial: Golang fediverse server

I made that

meena: thank you

I wouldn't tell you go thru this painful process if i haven't had paved the path already

👍

Lmao I just paved the path for myself 2 weeks ago

The museum dudes are gonna laugh at me

Oh well, got a small project for the weekend to move things over to the port installation

it was a lot of work getting that thing working on FreeBSD, and that has set the stone rolling for other VSDs

BSDs, too.

Yes, I can attest to that

even on esoteric platforms such as i386 and aarch64

Wow, what was needed there?

Figured golang would take care of all of it

the SQLite library translates the C code to go somehow, so i have no idea what. but basically: port the systems' libc to go, essentially

What the actual fuck

don't ask

Are they using cgo for sqlite?

Or something else

Title: cznic / libc · GitLab

ty. I am a go dev by day, so this is highly interesting to me...

Jesus christ dude, you did all this for the port?

Title: cznic / sqlite · GitLab

no, i asked people kindly

Lol fair

Also, not a dude

My apologies

I call everyone dude though, fwiw

amigan: don't do that. It's not gender neutral in most of the world.

I did look at the code, considering for ten seconds to contribute the effort, and then i looked away

AmyMalik: point taken!

old coworker was also a proponent of the gender neutral dude. It's hit/miss tho.

I don't use it often, but I have rarely strictly gendered it

Same goes for "guy", I think.

It seemed ironically cool back when I was young, but even I can see how it wouldn't age well

can confirm at least in the north east of the USA, "you guys" is gender inclusive lmao

other things i did for go on FreeBSD even tho i am not a go Dev: github.com/golang/go/issues?q=author%3Aigalic

Title: Issues · golang/go · GitHub

benaldo: yes, that is also my region and seems utterly anodyne

Used by all genders

I just use "potato"

or "bean"

I am mostly a potato recently, but aspire to become a bean again

meena: a sincere thank you for helping improve the go ecosystem on FreeBSD and everywhere. I use it both to feed myself and to run cool modern apps on my favorite OS, so this is really from the bottom of the heart.

Alright, how do I tell "make" to use llvm14 instead of the base llvm package on freebsd? (Cant believe I never asked this question before.)

Dimonax: env CC=blah CXX=blah make

If the makefile is written correctly

Of course...how did I NOT know that?

amigan: it's really simple: I run bleeding edge freebsd with slightly nonstandard settings, and boom, bugs everywhere

meena: lol, fair. I just run FreeBSD in a really complicated and nonstandard setup and boom, networking bugs everywhere. panics, hardware NIC offload bugs.

triple boot - begin

Title: Mina: "i just realized that this is simply part of my ve…" - Cathode Church

meena: i want a refund

go: RLock /usr/ports/net-im/gotosocial/work/gotosocial-0.7.1/go.mod: no locks available

rtprio, what arch are you running on?

amd64

I have never seen sync.RLock fail... wow

mutex rather

Title: Mina: "love to test basic functionality of software in a…" - Cathode Church

could it be nfs?

No, that is literally sync.Mutex#RLock() failing

I mean I guess in a very indirect way it could be

But that's not an NFS lock it's complaining about

meena: good insights indeed

forgot to change my login shell when i blew away /usr/local

my own sordid fault

And this is why I still use tcsh on FreeBSD but zsh everywhere else

so log in as root and chsh amymalik. we don't need a running commentary

that.. was root's shell.

why would you do that

anyway I know what to do

log in as toor?

toor is usually the /usr/local one

Though on my system it has an empty shell field in passwd....go with god AmyMalik

don't have toor

Looks like boot -s for you

yeah, that's what i think is next

forgot to rebuild the passwd db lol

Getcha every time. Or at least the first one.

There's always ln /bin/sh /usr/local/bin/bash

ah

Lol don't really do that

amigan: don't

Don't stare at wg-quick?

we have a nice rc script in ports now

What do you mean, even the rc script calls wg-quick

Which has a bug I have been trying to get zx2c4 to accept my patch for for months

Title: [PATCH] wg-quick: freebsd: save mask of interface in save_config

false start with etcupdate

anyway I think i'm about to get banned

meena: very interested to know if there is another script. Sick of having to maintain a delta every time it upgrades

meena: figured out the documentation thing

the left side bar was hidden and i didn't notice the hamburgre menu to expand it at the browser width

so i only noticed the first page

amigan: too tired to find it, sorry

rtprio: ✌️

meena: no worries, thanks for the tip!

i bet i can really scale back this vm now

Was it running full fat mastodon before?

oh yea

I decided on gts because the only machine I had available was a 4GB machine

And we also wished to use it to image SCSI discs since it has a PCI slot for an ahc(4) controller :)

still dont know if ill lose my followers in this process

would you look at that, basically the same power consumption on stock as on the latt branch

I would think you will, but idk

how are you measuring this power consumption?

Title: GitHub - intel/pcm: Intel® Performance Counter Monitor (Intel® PCM)

although it's possible I'm having ... is it called a zeno effect, or a schroedinger effect, when measurement changes the datum

1w difference you said?

AmyMalik: what are you measuring? Idle or load?

1w difference from Linux

amigan, nigh idle; the load is displaying the VT and scrolling the output of intel's utility

wow; lucky you to have time to waste with such a stupid differenc

Interesting

I have made this my pet issue

I should probably leave all the freebsd channels and only study it from the perspective of the code base

Instantaneous consumption measured by intel tools is hard to scientifically compare. You would be better served by a datalogging plug

And an average

they are different kernels, how would you expect them to draw the same

true tbh

Exactly. And you need to be running in single user mode

i mean, duh

And actually idling. Nothing but timer interrupt service

still different kernels

Well, yes

is this job market crazy for cloud architect and tek manager or what?

But you can definitely empirically say "x kernel draws more than y at idle". Just sure if that is the question being asked

Err, not sure

this is pointless folly but have fun

rtprio: many things in life are. Some say it is pointless folly to run FreeBSD when you can just use docker on linux for everything

also whats the resolution of the measuring instrament

Another valid question