I'm trying to pause the debug output from a command but it doesn't work with |less or |more, any other methods?

pike, It's probably comming out stderr. Did you redirect stderr to stdout before piping to less?

vstemen, thanks for teaching this n00b some magic :)

pike, No problem :-)

looks like I'm gonna have to recompile ffmpeg if I want the best AAC encoder libfdk_aac :(

pike: It happens. Fortunately, it's a port options so that makes it easy.

I already started using pkg's. Easy to switch or should I re-install?

That's fine. No need to do anything different.

Certainly no reason to re-install.

I'm just not sure if there's a package that has that particular option enabled.

I read somewhere that you should not mix pkg and ports?

If you'd like to continue using packages only (which is certainly recommended), it looks like you'll be wanting to hop on the Poudriere train to build your own with custom options.

pike: For the most part, yes. It is not advised. However, I do it constantly. But, I've been doing it for many, many years.

Pretty good at fixing my own issues.

I just can't put up with the 5-day compile time for things like rust and node. So, I shortcut things like that with packages. Otherwise, all ports.

this freebsd install I use to learn how to use freebsd :)

Sounds nice!

pike: What I would recommend first, is getting familiar with ZFS and snapshots. That way, if you make mistakes, you can always roll back and try again. It'll save a lot of time.

great tip :)

SNAPSHOT ALL THE THINGS

ls -D'%s' is there list of format options?

ignore, i missed strftime

Hey guys, ports download packages pretty slow, is there somewhere I could edit to put a closer url and if there is a list of mirrors somewhere?

wokko: where are you ?

european pkg mirrors are fast, up to 60 MB/s

mzar: Australia

european pkg mirrors are fast near to them

300kbit/s in the uk

If i ctrl c the make and restart, it does go too bad

we are doing dns based mirror selection by default

doesn't*

you could try other mirrors

is there a list of mirrors somewhere? and where to put info

pkg.freebsd.org

cheers

hey ppl, recently I found an interesting problem, I tried to install freebsd on oracle cloud but the problem is that da0 where I put the image (or ISO, I tried both) is read only (not the filesystem, that's clear, but the DISK)

i can't say i've ran into that before

how does it show the filesystm is read only?

a bit offtopic: anyone familiar with drupal?

i can't get my commerce products show up in my commerce store ;/

no

the filesystem is read only, that's normal

byt it shows the `da0` DISK to be read only

so I can not install on it

da0 is probably the install media?

if not, what is it?

oh, it is da0, huh

yes yes, it is the install media

that's the point

on oracle cloud you do not have another way to install so my try was to do dump and restore to a md dev inside RAM and then install from there

but it denies me access then to write on the "install media"

i installed by dd'ing over the running drive and rebooting, which worked

but i seem to recall someone putting effort into the freebsd iso being available

yes

that was my idea too, but it says the media is read only so it can not write on it

however, I did successfully put the installer there by DD

but DD from Linux ... not FreeBSD

as soon as it boots from FreeBSD everything gets stuck

ok, so pastebin some output

so we can see what's going on

well

there is not much output

instead of "access denied"

being root of course

when I tried to add some new partition onto the disk (which has plenty of space) it said "no space left on device"

and there are no errors

I will try that one more time in a virtualbox to see if it is not a oracle cloud limitation

as far it looks like boot volumes are only writable from linux

I will try with OpenBSD now, to see if that one works

how did you dd onto the device?

wsky: really off topic but if they're published there might be a taxonomy term to get them there

rtprio thanks you got me on a track

i still didn't make it work however

OpenBSD seems to work, I am at the installing process but no fails sofar, I hope it boots

rtprio: I did dd onto the device from Linux (default install)

well, it installed but it does not boot because of lacking UEFI support

even though it is the current one, UEFI shell can not find it

lol

Just a short question... I've read that for booting a GELI-encrypted root pool, the /boot directory needs to be in an unencrypted partition. Though, when I use the bsdinstall utility to create such a system, I can't find any unencrypted partition for /boot. I just get 3 partitions: a 512kb freebsd-boot type partition, a swap partition and the ZFS root partition, which seems to contradict the claim

that I need an unencrypted /boot partition to boot into a geli-encrypted zfs pool. So I'm a little bit confused. :-/

are you still fucking with this?

andreas303: you can use an encrypted boot partition. you use the -g flag with geli configure

rtprio: With the boot problems, yes, unfortunately. :( Or, actually I decided to try a different approach - using a GELI-encrypted ZFS root dataset instead of a native-zfs-dataset as root. I'm trying to find information about it in the freebsd handbook and in different manual pages.

however, it will require that you type the password at boot in console

not sure about what bsdinstall does though.

jmnbtslsQE: Ah, OK, I had totally missed the "-g" flag. I will try again. Typing the password during every boot is not a problem. I don't reboot very often.

as far as what bsdinstall does, from what your wrote, i guess it puts /boot and everything else into that ZFS root partition, but this is not strictly necessary

oh, it's also might be true that you can't do geli -g with a zpool partition, not sure. you might need to do UFS

(say, just for /boot)

but then you can have your zfs-on-root on a separate partition

jmnbtslsQE: I could "solve" my problem by just using bsdinstall, but I would rather learn how it's done manually.

(1) install gptboot or gptzfsboot into your freebsd-boot partition depending on whether your /boot is on ufs or zfs (2) make a geli partition that has /boot (either ufs or zpool, again not sure if zpool will work) (3) configure your loader that's in that /boot to boot from some other partition as needed (you set the "mountroot" variable, can be a zpool or ufs)

geli partition from (2) has -g

prtition you boot from in (3) probably has geli configure -b, not -g, and you load your keys from the loader in your /boot (2)

(if you want your main partition to be encrypted)

(you probably can see this but the freebsd-boot partition is the small, first bootloader, not your /boot)

you c ould also have an unencrypted partition with a /boot, then a separate encrypted partition with a different /boot, and configure the first loader to load the second one (which will ask for password), but i don't know the details to make this automatic

but, i think these days the loader can do a lot, and now lua is involved somehow so you can also program things, if you want to go down that rabbit hole

jmnbtslsQE: Sounds interesting! I will try again and see if I can get it working. Thx!

OK

hey folks, can someone explain why if i run service restart via ssh it hangs?

for instance, ssh root@freebsdbox /usr/sbin/service httpd restart

using ssh -n and nohup on the command don't help

Generally if you're passing a command like that, you need to wrap the command in ''s

no, so i just found a forum post describing it too

if i add >/dev/null 2>&1 then it runs instantly

so it's only if i allow output to the terminal it hangs

this sounds like a file descriptor issue

What if you skip the ``>/dev/null`` and just add the ``2>&1``?

You should be able to get the command output back to your terminal.

yeah, that's the key. i just updated the forum post

so "ssh root@box service httpd restart" will hang, but if you redirect BOTH stdout and stderr to files it works.

"ssh root@box service httpd restart >/tmp/mylog 2>&1" works

but that's very unusual

hm

the forum post the person was redirecting to null, but you can use a flie

file

That's really an unexpected behavior! (And I have been using ssh for decades.) I must try this and try to recreate that strange behavior. It's so curious...

Demosthenex, I am running freenginx as a binary pkg in a jail and I cannot recreate your reported behavior using it. What httpd web server have you installed? Was it using a binary pkg or source compile? How might I recreate your environment?

I am guessing that the service script associated with your installation is having a problem in some way. Sounds like a bug to me.

which httpd is it?

Hey there. I am currently installing freeBSD for the firt time and I currently experience a little issue. Basically the ISO didn't include the driver for my realtek network card.

I loaded the driver with a firmware blob, but it still won't work.

dorian1229, welcome

is this a wireless NIC?

The installer says that wlan0 failed to inityialize the driver interface when I go through the setup.

Hello :) yes it is wireless

due to licensing and size constraints, the wireless firmware(s) don't ship with the base image.

fwget(8) will try to install them from binary packages

this means you need an internet connection

I al9ready got the driver loaded with kldload and got the binaryblob from elsewhere

rwp: httpd was an example. i was starting zrepl

what does "elsewhere" mean? you want to get the wifi-firmware-*-kmod from ports

OH. it must be zrepl! lighttpd didn't hang when i restarted it

ok, so zrepl is doing something with the tty... sshd and lighttpd dont.

the driver I pulled of the repository called realtek-re-kmod-1100.00.1403000_1.pkg unpacked that and loaded the driver inside manually. the firmware blob I got from the kali linux repository

but shouldn't service be isolating that

the firmware was rtw8852b_fw-1.bin the driver did load and found this firmware the network now also shows up it just has an issue later when it comes to actually connecting

there might be a better way of doing this, but I have never used any bsd before and this is what I came up with

dorian1229, I'd recommend to get the one from freebsd ports.

you mean the firmware or the driver?

because internet is currently not available on the system

yes

transfer it with a USB stick or whatever

I did that

I have both the driver and the firmware on a usb and loaded

get that package instead

ill look into it thanks for the help

looked at it, the firmware file from kali linux is idential down to the hash

so it seems to be the same blob

which makes sense as this litterally links to kernel.org probably the same source which kali also used

dorian1229: if it's the same firmware it's fine, but installing the package (once you have a working network connection) ensures it'll be automatically updated in future

yeah I will 100% do that, I just want to get past the installer

dorian1229, personally, I would just skip the wifi stuff and deal with it once your system is installed/running.

so you can actually see what's going on.

I was told that wifi was needed to install both by the iso and another guy. Does the bigger iso do installs offline?

Because the one I got doesn't.

rwp: yeah, it's clearly only an issue with zrepl. it sounds like it could do some stuff with file handles (zrepl stdin server option?), and so i have my workaround.

Demosthenex, So the problem wasn't service with httpd but service with zrepl? Okay. But that does seem like a critical detail.

Regardless I still can't guess why redirecting to /dev/null would cause it to work. Because ssh without -t with a command will still be a pipe in the output either way.

For another data point you might try running with ssh -t and without the /dev/null redirection and see if that blocks wedged or not.

Something in that service script is probably writing a prompt or something and waiting for input but that would be stdin not stdout and so it is very confusing.

dorian1229, There is ability to do a wired ethernet installation and then deal with the wifi after having the system installed and running on a wire? That's certainly the easier way if there is.

rwp: yeah, i was summarizing and picked a random service. my bad

but yes, -t works

wild. i do ssh all the time, and i've often needed -n, i've actually never used -t ;]

learned something new, ty. wonder if that'd fix the postgres issue the other gentleman had

however, i'd also expect service to isolate any processes it starts, which is why it confused me so badly

ssh -t forces a pty allocation even if a command has been specified. Normally specifying a command means no pty allocated for a batch mode action.

The service command isolates a lot of things but I imagine there are historical scripts that expect to be able to interact with the admin running it.

yep, so whatever it's starting is making assumptions about the tty status instead of checking.

i'd have thought -n would have disabled the behavior

Or maybe it is simply buggy.

yeah, well zrepl is written in go, so i haven't a clue how to check ;]

ssh -n closes stdin but does nothing with stdout

sure, but stdin closed normally indicates batch mode w/o a terminal

zrepl may be in go-lang but if it has an rc script I imagine that rc script is written in shell.

seems like zrepl always assumes a terminal and doesn't check for batch mode

i saw nothing unusual (to me) in the rc script for it

but no protections either

rwp I cannot make a wired ethernet installation, as the system doesn't have ethernet. the only wired connection I could get to it was usb tethering which was so unstable that it got me nowhere

Then it is probably a problem in zrepl and probably worthwhile to make a PR problem report for it.

yep. thanks for your input. i learned something new ;]

ssh normally passes stdin through it so that "echo ls | ssh me⊙ec /bin/sh" and similar works as expected. But ssh can gobble up input which we need later and so ssh -n avoids that problem case.

"echo ls | { ssh example.com somescript ; ssh example.net sh ;}" will fail because the first ssh goggled up the stdin and the second one misses the input. The first ssh needs a -n to avoid it so the second ssh can read stdin. Just for a contrived example.

dorian1229, Bummer! Well then you must struggle through the wifi setup. :-(

just downloaded the biggest iso that was available over 4 full GB and it seems to install without any internet for now but yeah, in order to procede a bit of trickery will be involved

dorian1229, why not just skipping the wifi step in the installer and dicking around with it after you have a fully installed system?

I'm not saying only that works - that would just be what I'd do.

that is what I am doing

now

and it turns out I could have just gotten the bigger iso

it ships with the wifi stuff... thats's a huge face palm

i had a $20 usb nic for this exact purpose

yeah I should probably get one as well.

thanks to you all for the help though

probably the most active IRC I have ever been in

never forget the highest trough: wireless stuff in consumer electronics is the biggest pain.

s/trough/truth

and it works

great

thanks again for the help couldn't have done it without

we're all happy to help

just in general: be sure to consult the FreeBSD handbook. it's an extremely helpful resource

jbo, I'll see your "wireless stuff" and raise you a printer.

hrr hrr hrr :P

Why not double your pain, and go for a wireless printer?

CrtxReavr, I haven't yet had the stomache to try to print on freebsd

would be nice to use the office network printer

Is it wireless?

nope, ethernet

If the printer supports PostScript, it might be as simple as adding an entry to /etc/printcap and enabling lpd.

that's the thing: I know nothing about printers

I wouldn't even know how to check if it deos support postscript

it's an OKI MC873

The printer has a model on it. . . google it.

hmm, they seem to offer Debian and RedHat drivers. not sure if that is a good or a bad sign

PostScript is a generic printer languge.

Driver not needed.

Though. . . application support for it gets spotty.

hence wondering if them offering linux drivers is a bad sign :p

You're probably looking at implementing CUPS and all its fuckery.

Also, I'm quite sure the "Debian and REdHat drivers" are nothing more than the same CUPS printer definition file.

cups is that apple web thing, right?

I remember that from two decades ago on arch

and I remember pain

CUPS = Common Unix Printing System.

they missed their opportunity for naming it Common User Pain System

Okay, their docs suck, but:

{

"@type": "PropertyValue",

"name": "Page description languages",

"value": "Epson FX, IBM ProPrinter, PCL 5c, PCL 6, PDF 1.7, PostScript 3"

},

(I had to dig that out of the page source - wasn't actually visible in the browser.)

what am I looking at?

The "Page description languages" your printer supports.

Also, that printer lists telnet as a supported management protocol.

It's like it's 1998 again.

will I actually need that?

(the PDL thing, not the telnet thing)

Well, yeah. . .

IP is a network transport.

But to print, you gotta speak a language your printer understands.

so according to your descriptor above, it supports PDF. so I could just push a PDF over ethernet?

surely there's more info such as paper type, double sided printing, color and all that stuff

(as I mentioned, I never even remotely looked into printing)

You'd need a piece of software that essentially allowed you to pipe PDF syntax to the printer. I'm not sure what that would be.

I've piped PostScript via the CLI though.

pdf2ps ?

pdf2ps is in print/ghostscript10