I also have had problems with the gtk toolkit and used the lucid widgets with emacs.

Until I switched back to the emacs-nox version full time. I did that and no longer have widget toolkit problems. :-)

meena, I think I understand the snags I hit if I am not a member of wheel.

I see that the wheel group restriction applies to su as described here. freebsdwiki.net/index.php/Wheel

Title: Wheel - FreeBSDwiki

When I install a fresh installation I have the base /usr/bin/su installed. If I log in as rwp and am not a member of wheel then I cannot su to root.

And if I cannot su to root then I cannot pkg install anything. Cannot install sudo for example.

Can log into a vt console as root and then as the root logged in user install pkg ports and sudo and such.

But I prefer to log in as me, rwp, and then use "su -" to become root as needed. And for that I must be in the wheel group.

But I think your point was that sudo does not have that same wheel restriction that su has.

True. But I can't get sudo installed yet without it (or logging in on a vt console as root) and so it feels very limiting without it.

Also I like to be able to read the /var/log/* files without needing to become root. Those are mostly group wheel.

So again being in the wheel group removes a limitation that otherwise is limiting by design, but feels frustrating.

rwp: you're right about wheel and su(1).

The su(1) man page says: "In particular, by default only users in the 'wheel' group can switch to UID 0 ('root')."

This may feel frustrating in some sense, I guess, but on a multi-user system, you probably want some way to restrict which users can become root.

One simple way to do that is to require membership in a particular group, in this case "wheel."

The man page further says: "This group requirement may be changed by modifying the “pam_group” section of /etc/pam.d/su. See pam_group(8) for details on how to modify this setting."

So if you'd like it to work some other way, you can probably arrange that.

jld, The wheel group is a designed in restriction that is debatable if it is needed or not.

I mean, we are debating it right here and now! So clearly it is debatable. :-)

I am not debating it ;-) I am just describing it.

The System V "I grew up on" didn't have a wheel group policy. And so I don't have this baked in feeling that it is needed.

I no longer remember how things worked on unixes I used before freebsd

Mostly I was describing it too. Because earlier in the scroll-back we were discussing if being a member of wheel when installing a new system is needed or not.

well, I remember how some things worked, just not whether, or if so, how, the ability to become was or wasn't restricted

I will say that when installing FreeBSD and at install time creating a user "rwp" that for me it is needed to add rwp to wheel or I won't be happy after I log in and try to "su -" to become root to "pkg install $bunchofstuff"

that's true

long-time habit for me is to add my account to wheel during installation

anyway, you seem clear on how it works so I'll leave it there

jld, Thanks for jumping in! Relatively speaking I am a newcomer to modern FreeBSD. But have used BSD in the past many years before.

rwp, happy to chat. maybe we'll cross paths here again sometime

daemon: i deliberately wanted to set options with my package

a binary wont suffice

so i installed gcc for something, but it didn't include header files to compile something. now i deleted it, but the port i was building (emacs) still sees a shared object file

i used pkg delete gcc

but.. it didn't apparently completely delete

i noticed /usr/local/lib/ had a gcc directory, no clue if that was there before

trying to mount windows disk on freebsd 13.1

ntfs-3g not working

is this safe to remove? is it a "bug" that i could fix maybe?

fuse: failed to open fuse device: No such file or directory

nekobit, Do you have another gcc installed? Check via: pkg info -x gcc

gcc11-11.3.0_5

hmm, i see

oh, is that an old version?

concrete_houses: what full command line did you use?

concrete_houses: kldload fuse

add

fusefs_enable="YES"

to /etc/rc.conf

nekobit: one of the ports you build probably required it

nekobit, Could have been installed for some package/port

nekobit: one of the ports you built probably required it; probably safe to remove.

hmm, so

`pkg install gcc' reports that i can install it (11_4)

... until (that version) is required again

`pkg install gcc11` reports that it's already installed

sudo pkg upgrade doesn't report gcc being upgraded

im on FreeBSD 14 fyi

nekobit: are you trying to figure out why it was installed?

yeah

but im also wondering what the difference is

`pkg info gcc` reports nothing, so i assume it's an alias or something

(it is)

what is going on here?

the package name is gcc11; the one named gcc could be a diffrent version

gcc doesn't exist according to pkg info

grep pkg /var/log/message and look at the order that the packages were installed

something picked gcc11 as a dependancy

Dec 3 20:46:18 toomuchram pkg[36027]: gcc11-11.3.0_5 installed

Dec 3 20:46:18 toomuchram pkg[36027]: gcc-11_4 installed

Dec 3 20:50:35 toomuchram pkg[42932]: gcc-11_4 deinstalled

so i did lol, idk whats up here

or go to the port you just built and run make build-depends-list

or just delete gcc11 and get it out of your mind

/usr/ports/lang/gcc11

oh yeah, well i was compiling gcc11, never finished it. i scratched it

ok, deleted it

rtprio: okay... its gone

i assume that the gcc11 installed multiple little packages?

it was still compiling stuff for gcc11

it considered gcc11 installed

i don't think so

weird, i guess i can't provide a full answer now

i have it too; i wonder why it was installed

why was gcc and gcc11 a thing?

oh, now it's not

the port is called gcc11, the pkg is called gcc

is there a reason for that ambiguity?

if you look in /usr/ports/lang you'll note there are many gcc versions

ok well, motif was buggy for emacs too

trying Xaw3d

speaking of, does `make config' usually force a recompile whenever i do `make' again?

this is my first port ive compiled so far...

it does not; you'll need to uninstall, make clean and make again

alright, thanks

i didnt do the uninstall part, hope that's fine. thanks for answering my questions mate

are u a dev?

well, it won't you install the new one until you remove the old one

no, i've just used freebsd since 1998

fair lol

or 1999

you mean it wont uninstall, say, the motif toolkit? makes sense to me.

I have 2 drives pulgged into usb 3 and transfer running about 25MB

seems slow both drive rated 400M

nekobit: also, yes, it won't uninstall the motif toolkit. but it also won't uninstall the emacs built for motif until you tell it to

i just ran `pkg uninstall open-motif' for now

probably didn't cleanup the open-motif build stuff, but its fine

seems Xaw3d as exactly what i wanted :-)

da1: 400.000MB/s transfers

da0: 400.000MB/s transfers

why am i only getting 25MB

using rsync?

with fuse. right?

yeah

fuse is just slow on fbsd

fooookkk

usb drives are just not great

I want the speed!!!

enjoy the wait

:(

I torrented 24 season 1-8

aw yeah

I want same show but with swirds or lightsabers

you write driver

or naginatas

concrete_houses: cool. welcome to 2001.

I should convince room mate to get rid windows

and watch movies in sitting room all bsd

nekobit: the emacs port selects gcc11 as a compiler, fyi

I wana go abck to 2001BC

rtprio: yeah i saw that

when i switched away from gentoo to freebsd recently, i used ext2 on a drive for backups thinking "oh well it should support my drive, ext2 might be common", the native ext2 module didn't work at all, had to use fuse, took 2 days

so if I have another usb drive with UFS

I will get great speed?

you should yeah

sweet

I am comign from archlinux

wich is freebsd biggest rival I thnk

congrats

slightly more up to date binaries

i don't know about that

?

bsd needs let fookery with audio adn xorg screen size i notice

less

archlinux has very up to date binaries in is pcakcger

xorg screen size works fine, i don't know what you're on about

and supports smalltlk liek pharo.org

now i think freebsd did beat it on a latest postgresql binary one time so i looked bad sayign that one time

bsd is better than arch with x and audio

way better

archlinux is pain in nads

but once its confige

bsd != freebsd fyi

rolling release adn little more upt to date software

good point

fbsd

only survived non-feature-emphasized bsd?

sufficiently freebsd is the bsd

have you ninjas run freebsd on live sites with heavy transactions?

yeah yeah, history n stuff, i can't keep up with that

I heard some of the busiest sites run fbsd

but nowadays it seems applicable to separate them since it can be several things is what im saying

uhh, with the babble: I hopped over to FreeBSD for development purposes since I like UNIX-y and POSIX-y stuff, i've always used BSD-like distros. I've used FreeBSD around 3-4 years ago when i was younger, so glad to be back again

would also be cool to contribute to FreeBSD once i get more experienced

concrete_houses: iv never had heavy traffic

concrete_houses: i was asked once by intel junior engineer in the interview. I told him I would then look around aws pre-designs for stuffs like load-balancing

concrete_houses: I think he said something like I am wrong

concrete_houses: I smelled weird so I asked about LAMP

concrete_houses: He said he does not know what LAMP is

concrete_houses: I got rejected. old memory

intel interview?

yes this year

>old memory

yes because I think I applied hundreds after that

makes sense

it does not make sense. I think it is recent memory you're correct

i mean it make sense why you said old memory lol

such are the cases for job interviews

in terms of bsd, maybe it is more difficult for initial interviews

I was never asked any single thing about bsd even if I put it in my resume in the past

I think they really don't know what to ask. this is what i guess

what is the package for hald called?

Fudge: hal for embedded?

Fudge: what is that

hald is needed for xorg

as in /etc/rc.conf hald-enable="yes"

is that right? dbus, hald are needed to start xorg?

I do have not "hald" for Xorg

perhaps something else needs that🤷‍♂️

oh right

Firefox cannot work without dbus

Fudge: this? freshports.org/sysutils/hal

Title: FreshPorts -- sysutils/hal: Hardware Abstraction Layer for simplifying device access

nekobit: actually I didn't get rejected right away. I was asked to make a presentation slides for all my projects in my resume and send it to them. I did it and sent them. Then, I got rejected.

parv: hello

mictty, Hi there

I am sick of being a janitor for retard MBA

MBA really are scum of the earth my grandpa is right

may he rest in peace

its so insulting when a tard who doesnt know how to do stuff is put in contorl of a proejct doesnt plan it well then calls you in to work some magic is the huge crap plastered mess they have crated

concrete_houses, Could you go to #freebsd-social for your ranting needs?

i am convinced more than ever that usury is the culprit and sql database is just a thing to anchor you to a product even postgres its hell to update later I think files and whatever alnguage win

oh ok

is there any way to speed up chrome?

like is there any kind of cacheing i can turn on?

I remember running squid one time.... like html file cache

bsd has a better one wahts it called again

varnish?

concrete_houses: if you do similar things over and over it goes fast due to jit's nature

concrete_houses: if you do similar things over and over it goes fast due to jit's nature

concrete_houses: I think I leared today

concrete_houses: I don't have MBA person near me but it sounds no good.

is it fine to use a system i installed as a jail with chroot?

considering i deleted the jail

jail was overkill and i was having issues configuring it, i just need it to run a project for development

i installed with `bsdinstall jail /to/path'

what does the `jail' bit actually do here? man page just says "suitable for use"

hmm, according to /usr/libexec/bsdinstall/jail, looks like it does a full on installation

would've thought it was more of a regular install to a directory, with some `jail' commands slapped on

umm, there isn't actually a mention of the jail command here at all...

(or any words that say "jail", fact of the matter)

isn't this just a "install to directory" step? why is it called a "jail" script? is this just a bad design choice?

bare in mind im a new user, so idk if there is a reason or not truly

the 'jail' option just has it not do disk partitioning or install a kernel

otherwise it's a normal install

i get that, i'm mostly peeved at the naming here

this is perfect for chroots

in most cases you should use jail rather than chroot

in most cases indeed, im leaving out i want a regular chroot

don't forget that you need to mount a devfs inside the jail/chroot tree

i used a jail initially, but i found it was overkill when i had troubles configuring it for networking. i really just need a chroot. im doing local development

and you should in general make it a very limited devfs, using the provided set of rules for jails

i agree for other use cases, but im really not itching security at all here

this is to locally deploy a package i regularly develop for without installing postgres and all that shtuff

onto my main system

well, the devfs isn't optional either way, and adding -o ruleset=4 is very easy

oh.. yeah that makes sense

whats the -o ruleset=4 for?

see /etc/defaults/devfs.rules

didnt know about all this

basically mount -o ruleset=4 devfs /path/to/chroot/dev

yeah yeah

silly question but

add path stdin unhide...

how did i manage to mount the chroot while this is still set?

that exposes /dev/stdin

yeah i get that

while what is still set?

i mounted the chroot, stdin exists no if im running tcsh in the chroot?

the rule refers to the specific name /dev/stdin

most stuff runs most of the time when devices are missing, but the failures from things like missing /dev/null can be very weird

does... say `printf()' not use /dev/stdin?

am i getting that right?

(it's not like a certain OS I could mention when missing /dev/zero would stop you running any dynamic binary)

printf uses the already open stdin, it doesn't need to look up the name

(i mean stdout)

that's hurting my brain a bit...

having a /dev/stdout just allows programs to create a duplicate of their existing stdout descriptor by doing open("/dev/stdout"...)

don't worry about them, just note that some programs rely on things like /dev/fd/* or /dev/null

i know, its just a curiousity thing while im here

i would assume, say i create a virtual terminal with a terminal emulator (obviously?), that would have to poke /dev/stdout?

right?

not usually

in fact, the most common use of it is probably interactively, where you want to pass stdin/stdout to a program that expects a filename

ah that makes sense

thx for answering unrelated questions mate

I think bash wants to use /dev/fd/* for command redirects like <(somecommand) but it needs fdescfs mounted for that, just the default /dev/fd/{0,1,2} isn't enough

i think that will require extra reading for me

anyway. maybe though, a good first contribution to freebsd would be to make the "jails" command for bsdinstall an alias...? that really bothers me for some reason

and change the "jails" to something like "install-path" (or remove it entirely, and add a special case if it detects a path i.e. bsdinstall /my/path)

haha

i do appreciate the mounting devfs bit, almost would've forgot and had some troubles

off to sleeps

im glad i hopped over to the bsd's, feels nice getting assistance on the tighter things and not just "how 2 install a package?"

er, well i assume thats more of a supporting-role issue, so nevermind

we were all once noobs

though admittedly in my case that was back in the days of floppy disks

the availability of online help has improved massively since then :-)

yep yep

it was seemingly a gateway drug for me, i _only_ used all the bsd-like distributions on Linux

chrom esp youube is sloooowww

facbook and twitter also slooooowww

you might look at the memory usage

00:52 <jld> This may feel frustrating in some sense, I guess, but on a multi-user system, you probably want some way to restrict which users can become root. ⬅️ even a single human user system will have multiple users, daemons, who do jobs for you, run services for you, and hopefully don't have any big bugs so that when they run rogue, they

can exploit your system

Isnt that what wheel and sudo is used for?

I am not too familiar with Freebsd

Its how it works in Arch

hi all

Can I install previous versions of FreeBSD? FreeBSD 12.1 for example.

Get the right image or use the source

ftp-archive.freebsd.org/pub/FreeBSD-Archive/old-releases/ISO-IMAGES/12.1/FreeBSD-12.1-RELEASE-amd64-bootonly.iso

is ok?

I want to Compile the DRM Module.

at last I executed the #startx. But failed.

(EE) open /dev/dri/card0: No such file or directory

(WW) Falling back to old probe method for modesetting

parv: thanks

Seems like the correct 12.1 image from here

What king of GPU do you have?

s/king/kind/

Lenovo B50-80

Mobile DualCore Intel Core i5-5200U, 2700 MHz (27 x 100)

In other are you saying that you have a Intel (integrated) GPU?

M =8G

yes

intel

I found something like this on the forum and I want to try it: forums.freebsd.org/threads/freebsd-…rd0-no-such-file-or-directory.78157

Title: Solved - FreeBSD 12.2 startx failed '(EE) open /dev/dri/card0: No such file or directory' | The FreeBSD Forums

lol

yo9fah, Is a drm-kmod package not available your OS version?

Damn! Reverted to previous version

Is a drm-kmod package not available your OS version?

pff

sorry

I ran #startx but failed. after installation.

I installed freebsd 13, 14... and I get the same result

I remember seeing a mention of needing to add a small Xorg configuration file

So, I try the method from the forum. Thanks again for help

I will come back later

You may be better off with 13 or 14 in terms of support from kernel modules for GPU from drm-kmod port

Is (START)TLS now the default in 14 for sendmail to send out mail (on to relay host)?

I ask for I had to set "Try_TLS:<relay host> NO" in 14 in mail/access but none has been needed in 13.

This helped: unix.stackexchange.com/questions/14…tls-for-internal-relaying-of-emails

Title: sendmail - How to turn off STARTTLS for internal relaying of emails? - Unix & Linux Stack Exchange

parv: I installed freebsd 12.1, but towards the end it doesn't download my packages. It says it can't access the mirror.

Then I installed freebsd 12.4. The installation went well until I started to compile drm-fbsd12.0-kmod.

It stopped the compilation because it could not find the kernel source

So, I give up. I haven't found anything to solve it...

That version of FreeBSD in unsupported.

12.1 I mean.

Have you collected FreeBSD source, stored by default in /usr/src: If not use "git" (or "gitup") to fetch it

Ahhh, I have one more try. I found something like this on the forum: drm-fbsd13-kmod is gone (in favor of drm-510-kmod)

And yes, drm-kmod will require source headers in /usr/src/

s/:/?/

Don't use drm-fbsd13-kmod or drm-510-kmod, just use graphics/drm-kmod.

It's a meta-port that can detect which version of FreeBSD you're using and install the right one.

i installed pkg drm-fbsd12.0-kmod , but it did not work

debdrup, thanks.

I will try this too

See "git" use at docs.freebsd.org/en/books/handbook/mirrors/index.html#git ; use it so : pkg install git && mkdir -p /usr/src && git clone git.FreeBSD.org/src.git /usr/src

Title: Appendix A. Obtaining FreeBSD | FreeBSD Documentation Portal

Thank you for your kindness.

After that make sure you have right branch for your setup: cd /usr/src && git checkout releng/12.4 (or "stable/12.4", etc)

s/stable/stable/ # yikes

... wait that was correct but seemed wrong🥴 time to go it seems

Doooh, at the end it stops and asks for linux: pasteboard.co/HgcbJujJlhsn.jpg

Title: Pasteboard - Uploaded Image

I think I have to change the laptop. :D

Babaj, (meena) > "You can specify specific user names in the sudo configuration file." Yes, but, you can't use su to install sudo or to edit files until you are in the wheel group. It's a "Catch-22" circular dependency.

Setting the wheel group at install time allows the user to log in and then use "su -" to become root in order to "pkg install sudo" and then to modify files.

After that point one can modify sudoers in any way they wish and the wheel group is not needed for sudo.

rwp: This is possible. I'm not sure. It seems that the "su" used to work without adding a user to the wheel group.

Without being part of the wheel group the only option I can see is to log into the vt console as root.

Babaj, freebsdwiki.net/index.php/Wheel

Title: Wheel - FreeBSDwiki

Babaj: i think you're thinking of linux; you don't need to be in group 0 to su on linux

it has always been this way

It's humorous but I keep remembering a talk by rms given back in 1986 that I read more recently where the wheel group is mentioned as part of his motivation.

Search this for "wheel" and you will see the part I am talking about gnu.org/philosophy/stallman-kth.en.html

Title: RMS lecture at KTH (Sweden), 1986 - GNU Project - Free Software Foundation

rtprio: No, I was talking about FreeBSD. Perhaps you are right. You've convinced me.

In say Debian the gid 0 group is "root" not "wheel" as in BSD and in Debian et al has no special significance.

In Debian et al /var/log/* files are group "adm" so adding one self to "adm" allows reading /var/log/* without root similar to being in wheel does for FreeBSD.

In Debian et al the default sudoers file uses a "sudo" group is used more similarly to wheel for sudoers as a default to give sudo access.

but that's just something the sudo debian package maintainer thought would be a good idea

Agreed.

It's called "negative transference of learning" when things are similar but different among different systems and one applies something learned on one to another but it is different on the other.

the group name "wheel" goes back to at least TOPS-10 (DEC, 1970's)

Those people were big wheels then.

But it isn't just the util-linux package su as Unix su and System V su did not use wheel, and so commercial systems like HP-UX did not use wheel either.

jgh, It was probably someone at DEC then that originated the concept of a big wheel group. Probably.

Bringing that back to today it means that installing FreeBSD and setting up the user account it is good to add that account to wheel in the installer.

In order to gain access to "su -" in order to be able to do further things.

if i creat a single drive zfs pool and replace the drive's interface (say from a usb sata interface to a sata sas interface, can i simply use the same physical drive like i was doing?

CCFL_Man: I can imagine you might need to re-import it after a scan.

CCFL_Man: The only real obstacle I can envision is if the new interface presents it as a 4K drive where the old presented 512K or vice versa.

i suspect that, but say if i expend my pool from a single drive vua usb to multiple via sas

oh, this is abn old 3GB sas controller

CCFL_Man: You might be able to just do the swap and see if it presents any challenges in particular.

I believe you can always plug a SATA disk into a SAS/SATA controller. The reverse is blocked by the keyed socket.

I think I had something like that happen once when moving from USB to non-USB or vice versa.

Yeah, SATA into SAS is fine.

USB interfaces are mostly awful and there is much buggy USB hardware. It's often the source of weird problems.

I moved to a SAS shelf largely owing to the inadequacies of USB. And then the system moved from Debian to FreeBSD thanks to bugzilla.kernel.org/show_bug.cgi?id=209177

Title: 209177 – mpt2sas_cm0: failure at drivers/scsi/mpt3sas/mpt3sas_scsih.c:10791/_scsih_probe()!

I should say for USB Storage. USB Storage is awful. Strangely but USB Network devices have been very solid and reliable for me.

but in the end, the drive should be seen the same whether on a usb or sas interface?

CCFL_Man: It could present with a different native block size.

because i wanted to move some data to a new drive that would be part of a single device zpool

ahh

It will have a different /dev path to the new device.

yes, and a different /dev path

If things were set to use a 4K sector size then it will work for either. That's why we universally set things up with 4K sector sizes now for everything.

Good call.

so it should be set up to 4k sector size and i can move from usb to sas and still keep my data?

Yes. And you can ask zfs what sector size is configured. (If I or someone can remember the command for it.)

rwp: ashift, isn't it?

The current system's default can be queried: sysctl vfs.zfs.min_auto_ashift

And set: sysctl vfs.zfs.min_auto_ashift=12

CCFL_Man: You can look at the current ashift on a pool with zdb.

i don't have the pool defined yet

If you have not created the pool yet then best is to create the pool on the SAS/SATA controller and then you won't have to worry about exporting/importing it.

If you are running a recent FreeBSD 12 or later system I think the default is now ashift=12 4K sectors. (Since FreeBSD 10.1 my archeology says.)

Meaning that now we are compelled to check that it is 12 out of paranoia that we don't accidentally create a pool on a 512-byte sector disk.

But that it only needs to be set once before creating one or more pools and all will be 4K sectors then.

CCFL_Man, Hint: List attached disk names: sysctl -n kern.disks

CCFL_Man, Hint: List disk information: diskinfo -v /dev/ada0

That /dev/ada0 is likely /dev/da0 for your SAS/SATA controller.

CCFL_Man, Hint: I prefer smartctl information though: smartctl -I /dev/ada0

Drat my "i" -> "I" macro got me. Should have been: smartctl -i /dev/ada0

CCFL_Man, Hint: FreeBSD also supports this which is useful: camcontrol devlist

The smartctl command is in the "pkg install smartmontools" package.

rwp: i don't have the sas disk shelf yet, that's why i was asking. i have a drive that is failing and i want to copy the data to a new drive but it'll have to be a usb sata adapter for now until i get the disk shelf

Filesystem Size Used Avail Capacity Mounted on

/dev/da0p2 5.3T 787G 4.1T 16% /

6t disk

says size 5.3T

using 787g but avail 4.1T?

wtf

where is my 5.3 - (4.1+.787)

ufs

welcome to basic computer filesystem operations :-) everything has overhead!

concrete_houses, more detailed, your 787g of data needs an index in order for the filesystem to keep track of the different files and the blocks, that they contain. Also you are probably running journaling etc. All of which takes up space outside whats directly allocated by the files you store

concrete_houses, If that is a fresh install then the installer will have set up swap as well. Hint: swapinfo -g

Also: gpart show da0

Oh, "swapinfo -h" prints nice human readable sizes.

concrete_houses, Also I think you are seeing the file system "minfree" settings.

File systems have a minimum amount of free space required to operate. Non-root processes can't consume that minimum free space. So df is not reporting it.

The disk becomes reported as 100% when the non-root consumable space is consumed.

The root user is allowed to consume even the minfree space. But this causes the file system performance to become awful.

do we have something like GNU's free, for showing RAM? i find top's memory output awfully hard to read

See "man newfs" and "man tunefs".

meena, I like "htop" very much for the bar graphs it presents.

concrete_houses, Also note that computers use powers of 2 sizes (2^10 = 1024 = 1K) but disk vendors report in powers of 10 sizes.

Which for small quantities is little difference. But for large 6,001,175,126,016 bytes for a vendors 6T disk will be only 5.5 in 1024 powers of 2.

rwp: I find htop bad on not-linux systems

it relies on `/proc` or did so until very recently

I do not have /proc mounted.

Try this: sysctl hw | grep -e ^hw.physmem -e ^hw.usermem -e ^hw.realmem

I also see it reported in the boot log: grep memory /var/run/dmesg.boot

FreeBSD top shows ZFS ARC and other data but htop does not.

yeah, I just find that hard to read while everything is hopping around

I find reading hard on general, but if everything is moving too, it's not why easier

maybe vmstat(8) might be easiest

vmstat! That's an option I had forgotten about. I always use it for rates of paging (such as when swapping).

meena, If the htop colors are annoying (I don't know) then try it with TERM=vt100 or with htop --no-color.

Example of how I read the bar graph though: Mem[|||||||||||||||###############******************************* 753M/3.82G]

The "|" chars represent pages used for userland processes, "#" for I/O buffers, "*" for file system cache, and " " for unused memory.

And of course the color version has them in different colors. I find the htop memory bar graph very useful to quickly see what a system is doing.

That's actually from a Debian system because my FreeBSD system looks extremely boring with everything as file system buffer cache.

But I guess that means I don't know if it is really functioning on FreeBSD.

meena, Is the "sysctl hw | grep '^hw.*mem:'" data useful? Or still much too much of a raw data dump?

it would be nice to have regex support in sysctl output

I rather like using grep on things even if they have built in regex. Because *grep works with everything*.

I almost never use "pkg search -x '^foo.*bar'" and almost always pipe that to grep.

i've become addicted to ripgrep lately