Linux and BSD are apples and oranges. One is a kernel, and nothing else. The other parts are developed elsewhere, and with no single standard. BSD is a complete operating system under one roof. And I'd argue that systemd has made Linux far less UNIX-y. At least in my eyes.

i see these comments often. does anyone use only the base system? seems like with linux or freebsd, you're going to need 3rd party software. or am i missing something?

I think you're approaching it from a different perspective, I don't mean we don't need third party software, I mean that the development model is different. Linux distributes its responsibilities among others, then a "distro" picks up the kernel, and mixes different parts of software from different players.

thanks, and i get that. i guess i don't understand why it's so emphasized. is it purely philosophy, or is there a tangible benefit i get as a user?

The benefit you get is FreeBSD xD

lol, fair enough

i really feel like FreeBSD should tweak how it markets itself. the coolest stuff i've found that sets it apart from linux isn't mentioned so much, which i find odd.

Obscurity has its benefits too.

nah, it should tout the good stuff. for example, i have a server with 6 luks encrypted disks and systemd boot. a near identical geli + zfs freebsd system boots like a full minute faster. ... so many small items like this that together make it an interesting alternative.

If you look around you'll finding people promoting FreeBSD (eg on Twitter), and I think people look past it. Or somehow don't see it. I don't think boasting about being better than the other guy brings much but bad blood.

it's not boasting. it's highlighting actual advantages.

linux smokes freebsd in several things, and that's fine. don't have to win 'em all.

but highlighting the wins help people fighting those same battles

then they can choose

That's already the case. The choice is there.

what's a front page +1 link that highlights a solid reason to choose freebsd?

"Its advanced networking, security, and storage features have made FreeBSD the platform of choice for many of the busiest web sites and most pervasive embedded networking and storage devices."

that's the closest thing, and it's so fuzzy

Look at Void: voidlinux.org

Title: Enter the void

Much smaller project

Right on the front page, 6 highlights

i wouldn't say any are boasting. they're just true things that might help me make a decision to use it.

I see it differently. I think FreeBSD does a good job at explaining what it is (eg freebsd.org/about) and how to use it (freebsd.org/handbook). I know Void has a handbook, but it's almost empty. And doesn't teach you all about Linux.

The about page is like buzzword 101

"advanced networking" "uses memory efficiently" etc

too fuzzy to be useful

"runs a huge number of applications ... over 33,000". nixos has over 80K. it just makes it feel out of touch to me. anyway, don't get me wrong, i think freebsd is cool. i don't the the website does it justice.

one exception, i love that the security notices and errata notices are on the front page. it highlights a feature: security matters.

And there's "pkg audit" too. I don't recall xbps having that.

agreed, thats a good highlight, so many more

hi there

is there a way to make ifconfig output more readable ?

4 interface listing > 2 pages ... 99% useless information

nerozero: what info are you looking for

(the first answer to this is: the output is this way, because it has always been that way. changing it, but default, would break a lot of people's scripts)

yes I do understand, this is ok to me to have this outout

what output do you need?

but when machine has > 2 interfaces, some bridges, some tap, ... the output becomes completely unreadable without less

i didn't understand meaning of goups

is there a way to list only all physical interfaces ?

only bridges, ... etc

is it possible to output table only interfaces which has assigned IPs

ifconfig -g bridge

should list all the bridges

it's annoying that physical interfaces are marked by their lack of a group - on FreeBSD

on OpenBSD they do

meena, -g will output only interface names

nerozero: github.com/canonical/cloud-init/blo…oudinit/distros/parsers/ifconfig.py

Title: cloud-init/cloudinit/distros/parsers/ifconfig.py at main · canonical/cloud-init · GitHub

nothing else

I wrote a whole ass parser for this shit

WOOOW :D

epic !

WELL DONE !

will take a look, thanks

I wonder if I should have spent my energy on adding libxo to ifconfig

but, i feel like it's gonna be easier once melafarino@ is done with the netlink refactor which cleans the code to real nice

u did a huge job!

thanks

melafarino, lol

nerozero: you can also take a look at the ifconfig(8) man page. it really gives some good options for formatting output

I have tried some, but that doesn't give the output expected .. I guess I need to spend more time understanding that

test

complete

thanks - was intended for a single server. Oh well, live and learn.

meena: i have no idea what that does, but nice work! lol

is there a standard (base) utility that will write a date-timestamp before each line of a file

Sed and date

Possibly cut as well

cpet: sed is a powerful turing complete language. is there anything less powerfull? ;)

Awk date

if its scriptable i usually use DATESTAMP="$(date "+%Y%m%d.log.%H")"

And possibly cut as well

or something similiar

as "now" when the line is output or what does that mean for a "line of a file"?

cpet: awk is a powerful turing complete language…

Oh for ducks sake Google it

i am. I also asked just in case someone knows better than my skills

maybe logger is doing what you want - i assume, this is to enrich logging

$nobody knows your skills

If you have to ask her you have no skills

\o/

there's ts from moreultils… which is written in Perl…

ridcully: good point, lemme check

cpet: my skills vary wi(l)dely

Could care less about your skills

kinda rude

*couldn't. insult with proper grammar please ;)

Grammar Nazis are evil

jawoll

Egg rolls ?

anyway. logger has -s which sends the output also to stderr, but it *always* logs to syslog. And it also doesn't log the time/date. That's syslog's job

cpet, don't seem to be yourself today as if its influenced a wrong side of the bed wakeup

Thank you

could be anything. don't really care, but every time i see this behavior i see a +q happening. including myself

so, yeah, after all, sed/awk might be the best way to go about it, no matter how much i don't like it

just take that as a friendly *ya good* dude!

you just made me look up neovim

+1 for neovim, i use it everywhere

editing a 50 line shell script without syntax highlighting makes me think of the good old times of being woken up in the middle of the night by a broken Java App and editing XML in vi on Solaris

must be talking about github.com/neovim/neovim

Title: GitHub - neovim/neovim: Vim-fork focused on extensibility and usability

$EDITOR is my favorite.

same!

FCEDIT="$(which vim ||which vi)"; export FCEDIT

VISUAL="${FCEDIT:-vi}"; export VISUAL

EDITOR="${VISUAL:-vi}"; export EDITOR

Woosh.

didn't intend for that to be 3 lines ;)

what's FCEDIT?

another shell va

r

i lost its meaning.. i ref it as "fucking command edit"

meena: FCEDIT is for the sh built-in 'fc' command, for editing previously entered commands.

I dislike that the moretutils ts name conflicts with the task scheduler ts name. Sigh.

meena, Try something like: ...command here... | while IFS= read -r line; do printf "%s: %s\n" "$(date "+%F %T")" "$line"; done

Adjust the +%F %T format to whatever is preferred. I like that format for me.

rwp: is that doing post-processing, or am I just imagining that?

(this is a giant log file)

I was responding to this: ":43 <meena> is there a standard (base) utility that will write a date-timestamp before each line of a file"

You mentioned the perl moreutils ts program. Which operates exactly the same way. So I suggested the shell one-liner as an alternative that does not require anything but /bin/sh to work.

If you need to post process a large log file I don't know how one would know what timestamp to put in the front of each line if the time were not saved there originally.

rwp: no, i thought your method was post processing!

I thought that the loop would buffer

Has anyone ever run into the issue where `php82-pecl-gnupg` fails to import files that both `gpg --import` *and* a custom gpgme-using test program import just fine?

(gpgme being what the PHP extension uses internally)

meena, It's line buffered because it reads one line at a time. That's a good impedance match to text log files.

rwp: aye, thanks

I am trying to install enligthenment in freebsd - I wonder if I still need xorg or I can do it with wayland - if yes - is any pointer for it

acu: enlightenment.org/about-wayland not quite done yet

Title: Wayland

Enlightenment is the only DE that ever really made me consider using a DE.

meena: thanks - so I need to install xorg....

debdrup: yes - I installed Enlightenment in Debian 12 - it works - but some prerequisites where there that were fullfilled ----once I installed gnome.... ---- my intention was to install only enlightenment

s/win 22

Hey, you're in #lisp too?

How is enlightenment? I think I only tired E16 like 15 years ago, man I had forgotten about it.

it didn't provide any enlightenment to me

pity

so.. we can't just download and install enlightenment is what you're saying

Is it possible to delete a logged-in user in FreeBSD? As part of a packer build, I have this as the shutdown command: ""echo 'vagrant' | su -m root -c '/usr/sbin/rmuser -y vagrant; /sbin/shutdown -p now'" I see on stdout that it's removing the user vagrant, but it doesn't actually delete the user. /home/vagrant is empty, and the password is locked, but the user still exists. Is there a way to delete an active/logged-in user?

"still exists" in what sense?

I think he means that the user is still logged in. What you can do is delete the user, then kill processes owned by said (former) user.

rmuser is supposed to do the process killing thing

Anyway, as for my php-gnupg issue. I've narrowed it down to "PHP CLI works, PHP-FPM does not". But can't quite figure out what *exactly* is the issue, so far.

Is there just not enough time between the rmuser, and subsequent kill signals, for processes to die before the shutdown?

RhodiumToad: Sorry, user is still in /etc/passwd. master.passwd shows the account locked, so that's good, but still an entry in /etc/passwd.

pedahzur: what was the output from rmuser?

RhodiumToad: Sorry, closed the terminal, but it was something like "Removing user vagrant" or some such. No error output. Give me minute to run the build again, and I can tell you. Hold on.

add -v to the rmuser command

What's the purpose for echo'ing "vagrant" into the pipeline? rmuser has the name on the command line and does not need it, right? shutdown does not need it.

As a debug thing, not suggesting to do it all of the time, I would put a sleep 30 (long time) in between the rmuser and the shutdown and if the problem goes away then would know it was a race condition problem.

if i don't have any specific build options i want to change, is there any reason to compile packages from source vs. just using pkg? i assume no

markmcb, Right. The pkg binary was built from source on the FreeBSD build system and is "just as good" as one you built identically yourself.

thanks rwp

However you may have some strategy overall which includes compiling your own version of things.

rwp: echo'ing the password into sudo.

pedahzur: you mean su? that does not accept passwords on stdin

(nor does sudo)

markmcb, Traditionally (as far as this newcomer can tell) FreeBSD people compiled everything from source. The binary pkgs are new and save a lot of compile resources across the planet. But people who previously compiled from source continue to do so. Meanwhile... That does burn a lot of cpu time for everyone to compile everything. And if we don't need any customizations then the binary pkgs are more convenient.

rwp: thanks, that was my assumption. just wanted to make sure i wasn't missing anything :)

markmcb, It also allows newcomers to start easy with binary pkgs. Because they are so convenient. Get familiar with the system. Then later as they do have customizations it is a gentler transition to compiling from source because we are already familiar with how things work by then.

pedahzur, Unless you are already root then su will ask for a password interactively on /dev/tty so the echo into it won't work. If you want that set up sudo for NOPASSWD for this particular action.

Weird...because it seems to work. I'll dig in more. rmuser gives me this:

==> qemu.vm: Gracefully halting virtual machine...

qemu.vm: Removing crontab for (vagrant):.

qemu.vm: Removing at(1) jobs owned by (vagrant): 0 removed.

qemu.vm: Removing IPC mechanisms.

qemu.vm: Terminating all processes owned by (vagrant):

Sorry for the splat.

Maybe it's already sudo'ing behind the scenes...

so i have a jail template i installed with bsdinstall. i just did a freebsd-update -b on it, and it says it stopped and started sshd. Does that make sense? It's not an active jail. And the rc.conf file doesn't have sshd enabled.

pedahzur, The ordering there seems odd. First we see that it is halting the virtual machine. I assume that is from the shutdown. Then after that we see it removing the crontab, at jobs, IPC, and processes. That seems to be out of order. Probably asynchronous action.

I would debug by putting a long sleep time between the rmuser and the shutdown to ensure that rmuser can complete and if that works then it means the rmuser is not operating synchronously but is completing work in the background and that is the actual problem if that is the case.

rwp: Sorry, that is packer saying it's sending the shutdown command (which in my case has the rmuser before it). I shouldn't have included that line.

ok, so setup samba in a jail. i think i should setup per user zfs filesystems with quotas, that likely means allowing zfs into the jail. worth the effort?

Sam a in a jail?

markmcb, I would "ps aux | grep -v grep | grep /usr/sbin/sshd" and see how many sshd processes you have running and if it is the expected number.

markmcb, And then the rest of your question depends so much upon your jail configuration that "it all depends". But probably okay once the details were discovered.

there are two. i would only expect sshd running on the top-level host

If there are two then could the second one be in the jail? (Or a different jail?) I think it is actually running. And hence the message you saw.

If you stop the jail does that second sshd go away? I would think that it would. I am sure that when you dig into everything the details discovered will make it all make sense in the end.

Demosthenex, I don't use samba so am unqualified to speak but if you are using zfs isn't the jail already on zfs? So "allowing zfs into the jail" seems like a strange question in that case to me.

zfs has special support for jails

yeah, i'll dig around. i just wouldn't expect that message at all when updating with -b on a template (i.e., not defined anywhere as a jail)

(specifically, if allow.mount is set in a jail, a privileged user in the jail can mount (only) filesystems that have jail support)

zfs is so useful that I of course would recommend using it. But if you don't like zfs then it will appear invasive like kudzu vines that get into everything.

it's not quite that bad. I build without it, and rarely run into issues

rwp: well, bastille gives the jail it's own zfs filesystem, yes. but i'm thinking i may want per user storage quotas

It also depends upon if you need quotas and all of that. If I were setting this up for a school then I would definitely use zfs and quotas. If for a small office of a couple of people then no I would not.

today my devuan (at least it's systemd free) samba server has zfs per user with quotas ;]

If you are already using zfs with quotas then you are already familiar with it and I think you should continue doing it.

it's just the jail interaction i have to study up on

if i can adminster sub-filesystems inside the jail only, that's perfect

Assign a dataset to the jail. Allow the jail to create sub-datasets in the jail if that is desired. Seems perfect! "Bob's your uncle!" as they say.

so looking into that, it restarted my top-level host sshd. is that expected?, i.e., if i did bsdinstall jail /my/jail/template, and then later did freebsd-update -b /my/jail/template fetch install

hrm, sounds like if i assign a dataset to the jail, i can't mount it on the main host

that'd suck because i'm doing all jail backups from the main host

so maybe i made the datasets from the host, and just let the jail use them

markmcb, I would not expect it to restart the top level sshd and so that does sound strange to me.

Demosthenex, Huh? The host gets to see into the jail. At least that's the way mine are set up.

Or are you making raw volumes or something other than datasets?

rwp: no, the docs say you can't mount a dataset from a jail, on the host.

because the jail admin could set the mountpoint to something bad ;]

Demosthenex, Hmm... Well... No one told me that and I just listed a jail dataset on the host with ls no problem!

I am not any kind of an expert on jails however. Just muddling through myself when it comes to jails.

rwp: did you use zfs jail to assign it to the jail, and set the jailed attribute to allow the jail to manage it?

No. I created a zfs dataset and configured the jail to use it in /etc/jail.conf and started the jail.

yeah, that's just mounted on the host.

if you give the jail control, you have to unmount it on the host, according to the docs

but... i think it makes sense to me to have the host admin the zfs datasets anyway