based on what i just read here wiki.freebsd.org/WiFi/Rtw88#Currently_known_issue

it sounds like i can get my inspiron's wifi card working if i just put a 4GB stick of RAM in it (it currently has 8)

how can I initiate a zfs send/receive, from the receiving machine, also using netcat

can I do something like: nc -l 9999 | zfs receive -F ssd_bkp/mail_home && "ssh host1 zfs send -v zfs/mail_home@mail_home-20241023182500 | nc -w 1800 host2 9999"

why netcat ?

no encryption limitation

guess I could use mbuffer too or some other variant

ZedHedTed: you just limit your config for it. but the rtw88 is buggy on 14.1

FreeBaSeDGirl: seems I misread. we have a different issue. the freebsd servers were not involved.

last1: look for transfer file with netcat.. then test with a file.. and then replace the file with the send/receive

last1 why netcat? If the problem is speed using ssh, you can use arcfour cipher. The transfer will take place quickly.

hello my php-fpm dissapeared and I cannot get it to work anymore not sure why

or what happened?

sixpiece: disapeared how ?

I don't know

it seems like magic

but seriously service php-fpm restart stopped working

it started with all kinds of warnings and then it suddenly dissapeared

sixpiece it was changed. use php_fpm instead

service php_fpm restart

I am getting a lot of warnings

all of a sudden

sixpiece nothing wrong now. Its not a warning

it's not working anything :(

php_fpm was started

see the status: service php_fpm status

php_fpm is running as pid 64323.

Well, it seems nothing related to PHP FPM, but maybe php version or module. Did you upgrade major php version (eg. php8.1 to php8.2)?

I think so by accident but there was no warning

:(

it just removed it when I added this package but no warning that it would remove it :9

fpdf

If you upgraded major php version, maybe there some plugin incompatibility with this version. Let me see your packages (please termbin): pkg version | grep 'php'

it's not responding

just wait

is there a way to get back 8.1?

Yes, but wait.

The command returned to you something?

You need to calm down, run the command above that I mentioned.

strange. Create a file phpinfo.php with content: <?php phpinfo(); ?>

that was the problem it erased php83 and replaced it with php82 now it's working thank you

ober: i don't think i can change the config on a live system

thank you devnull

sixpiece you're welcome, but you have misc php versions installed: mod_php74-7.4.16, mod_php80-8.0.30, php81-libawl-0.64 , etc....Remove unnecessary packages.

And before upgrade major php version, always test your website on development environment (creating a subdomain like dev.hydroponique.shop). Updating in a production environment without testing first is a risk.

But I'm glad that helps.

hi. git 2.47 added support for git-prompt.sh in most shells (including ash-based like in freebsd), while up to 2.46 it only worked in bash and zsh. i was wondering if anyone tried it, and if yes, were there issues? (i wrote the compatibility changes)

the source file is at contrib/completion/git-prompt.sh but distros install it in different places, typically someplace in /usr/share/ or /etc/

(no idea whether freebsd installs it)

sixpiece: if you're using zfs then you can create a "boot environment" and mount it into a jail, upgrade packages etc, and if it works, then switch it over.

here's a good place to ask about advice if any of that sounds useful / needs more explaining

jails + zfs is the super power

avih: use the plist? %%CONTRIB%%share/git-core/contrib/completion/git-prompt.sh ? (%%CONTRIB%% is the conditional for the CONTRIB option)

dvl: for some reason freshports fails to show plist of devel/git

nimaje: not sure what you're asking. i was asking if anyone tried it out and if there are any comments.

(i know how to use it and where to get it from. i was asking if ither tried it)

others*

your comment about the file location looked like the question where it is on freebsd

last1: gist.github.com/dch/3333118#using-mbuffer-for-speed and use the `-c` option on send, to send blocks that are compressed on disk, also compressed over the wire.

last1: mbuffer helps a lot as zfs is bursty on both ends, keeping the pipe full is important

avih: its in `/usr/local/share/git-core/contrib/completion/git-prompt.sh`

dch: oh, thanks. so it's installed together with git?

avih: yup, AFAICT in all flavours (CONTRIB is in OPTIONS_DEFAULT in the port Makefile)

although before git 2.47 it was not usable with the freebsd /bin/sh, but now it should be. previously it only worked with bash or zsh

erm

apart from git@lite

cool. nice.

did you by any chance try it out?

I just use whatever fish shell does by default

so not yet

yeah, it doesn't work with fish AFAIK. it requires posix-ish sh plus support of "local" (which all ash derivatives have, like freebsd /bin/sh)

it already does this stuff fishshell.com/docs/current/cmds/fish_git_prompt.html

i see. by the identifiers names, i guess it tried to follow upstream git-prompt.sh to some degree

tries*

(which makes sense, because upstream supports also global or per-repo config values via the bash.* names, although again, now it works with almost all posix-ish shells and not only bash

so the fish git prompt uses the existing config values)

avih: the (pkg-)plist is a file* that lists all files a port will install, so you can look into that if you want to know if some optional file from a project gets packaged and where or if some project gets split into multiple ports you can search each ones plist to figure out which brings the file, freshports can also search plists for you (*well, the final plist is build while building the port,

as for a small number of files, having it in some variables in the makefile is more ergonomic, then having a seperate file; but some ports use that to dynamically building the plist, which makes searching it from just the ports tree impossible)

nimaje: right, thanks. though currently i don't have freebsd installed. i mainly wanted to check if anyone tried it out and get some feedback if possible.

(i guess i could check the packages online as well, though for now it's enough that you confirmed it is installed together with git excepr for the lite package)

except*

so it's a good start! :)

how can i add uefi freebsd to grub for boot

to gpu

sdd

for mbr is menuentry "FreeBSD" { root=(hd1,7) chainloader +1 }

e.g. 1,7

in grub.d/40_custom

this is for non-uefi for mbr sdd/hdd

but for uefi sdd/hdd gpt how

gpu=gpt

i mean

should be like you chainload any other uefi program, the boot loader is simply named loader.efi

then should be menuentry "FreeBSD" { insmod part_gpt insmod fat set root=(hd1,gpt7) chainloader /efi/freebsd/loader.efi } ?

i think

So I ordered a server from Hetzner, also a 16-GB usb stick, burned FreeBSD iso in linux rescue mode, then requested KVM, so far, fine.. However, the problem now is, FreeBSD installation says;

"No interface specified for IPv4 configuration" it seems, FreeBSD does not recognize the NIC interface on that server? :/ Am I wrong? Any clue would be much appreciated.

The mainboard is: ASUS System Product Name/PRIME B760M-A D4, BIOS 9017 02/07/2024

And one more info from rescue mode regarding network: RTL8226B_RTL8221B 2.5Gbps PHY r8169-0-500:00: attached PHY driver (mii_bus:phy_addr=r8169-0-500:00, irq=MAC)

r8169 0000:05:00.0 enp5s0: Link is Up - 1Gbps/Full - flow control rx/tx

tercal: might be you need to find the device name and put that in your rc.conf.

It's Realtek, but bootable FreeBSD (13.x and 14.1 I tried) does not find any NIC. Would simply adding "re0" with a DHCP configuration let's say, in rc.conf, really work?

tercal: if you drop to a shell in the installer, does pciconf -lv show a driver attached?

The re driver supports RealTek RTL8139C+, RTL8169, RTL816xS, RTL811xS, RTL8168, RTL810xE and RTL8111 based Fast Ethernet and Gigabit Ethernet adapters including:

i was pretty sure its supported.

my current server uses re0. i havent installed freebsd as a new install on that particular server since ...2014ish. :) its gone thru a few motherboard changes and I have changed the driver, in rc.conf. . VMs i install are always the virtio driver.. so yeah toss a dhcp config for re0 and fire it up

mrelcee: in case of going fully manual, nothing to load in terms of kernel modules for re0?

is it as simply as adding related lines in rc.conf? some kldload thing?

re should be included in GENERIC kernel, i doubt adding anything to rc.conf could fix this

you could try the net/realtek-re-kmod port, but you'd need to get it onto the server somehow

hetzner doesnt run the freebsd instasller. they dump a generic install onto the boot drive they also rent a lot of different machines...

isnt "No interface specified for IPv4 configuration" a message from bsdinstall?

they barely support BSD. in the past I have needed to boot a linux rescue disk to do any fixes. got pretty creative a few times using a hex editor to unmess a very unfortunate pf config. :)

typo

ivy: Yes within the installer

tercal: so if you have a usb stick, i'd suggest grabbing the realtek-re-kmod package for whatever release you're using, extract it, then copy the new module (i think it's called re_updated.ko) onto the usb stick and try to load it in the installer. might work, might not

by 'load it in the installer' i mean drop to a shell and use kldload

are you connected to it via their IP KVM?

also grab the pciconf -lv output at the same time

and a copy of dmesg wouldn't hurt while you're there

will let you know ivy and mrelcee as soon as I get the KVM access again, yes mrelcee I was connected to it via their IP KVM. I created a iso file containing only the if_re.ko from the package: realtek-re-kmod-1100.00_1.pkg for FreeBSD 14, so I will mount it on the KVM interface as virtual cdrom, and try to mount the iso file within the FreeBSD OS after I install it without networking. But is it possible to mount that iso, and get the .ko module from there

and embed it into the installer, ivy?

i don't know about 'embed it into the installer', i'd just copy the .ko file and kldload it to see if it works, if it does, you may want to put the .pkg on the usb stick or iso as well so you can install it after the installer is finished

i.e. boot installer, load .ko, install, reboot, install the package into the new system

if_re_load="YES" in /boot/loader.conf if its not finding it.

there's also realtek-re-kmod198 package, i'm not sure which is newer off hand

ivy: freshports.org/net/realtek-re-kmod seems 1100.00 is the latest one.

that is the official realtek offering and does support more devices

Okay got the KVM now, starting up

ivy: That's the output of your suggested command

pciconf -lv in the installer shell; i.ibb.co/60T1Zrd/ssh.png

did loading the module work? should show up as re0 if so

here it is: i.ibb.co/4Ngfr0Y/ssh2.png

"if_re.ko module already loaded or in kernel"

Perhaps it's good now - somehow? I will give it a try.

"ifconfig" reports only lo0, though

tercal: you need to specify the full path to if_re.ko

as the message says it won't load it from the current directory

the current working directory had already if_re.ko? I was in the extracted folder

yes, but you need to specify the full path

e.g. kldload $PWD/if_re.ko

kldload doesn't look in the current directory by default

kldload /tmp/test/if_re.ko reports:

if_re.ko unsupported file type

please check dmesg(8) for more details

that usually means a kernel linker error, is there anything new in dmesg?

Get out and vote, Americans. . . vote early, vote often.

ivy: Here it is i.ibb.co/5KX4yrT/dmesg.png

toss if_re_load="YES" in /boot/loader.conf and put: ifconfig_re0="DHCP" in rc.conf.

tercal: hmm. try copying it to /boot/modules/if_re.ko then do 'kldload if_re.ko', that's what the real package does

manually loading it puts it there this time...

it may be you need to load it from loader.conf, but that's going to be difficult in the installer

ivy: read-only file system

that's the shell from USB stick's installer image

maybe it's worth installing the system from the usb stick, then reboot into the new system and install the package there?

yep better

will proceed then

you don't need the network to install as long as you're not using the minimal image

or bootonly or whatever it's called

back to my commentary on hetzner not really being bsd friendly. :). if you can manage it we accept your money!

i don't guarantee this will work btw, it might be even the realtek driver didn't support this card

ivy: my image was FreeBSD-14.1-RELEASE-amd64-memstick.img - proceeding

tercal: i also notice you're using Raptor Lake, be aware the FreeBSD scheduler does not support hybrid architectures currently (as far as i know) so you might want to play with cpusets to get best performance

assuming you can get it working at all, of course :-)

ivy: That's the latest situation when adding if_ko module in /boot/modules and adding appropriate lines in loader.conf and rc.conf, then rebooting the os;

tercal: that's not great, if no one else has an idea you might need to ask on net@

Good morning! Why does the RPI img use UFS instead of ZFS? Is there an inherent overhead in using ZFS that makes it unsuitable to use on a Pi?

thank you guys anyway

back to Ubuntu on this box

would you really want to run ZFS on a CF/MMC card? (that's what the pre-installed rpi images are intended for)

not to mention would you not rather have the additional free memory for your applications

I'm a freebsd newb and was hoping to get some experience with zfs on a bunch of pi's i had sitting around.

i figured there was a reason for ufs on them instead of zfs and what better way to find that reason than to ask?

I assume you have a fairly reasonable pc, why not just create a virtualbox, bhyve or kvm vm; install freebsd in it and test away to your hearts content

I can. I can also do VM's on my proxmox cluster. I was just trying to find something to do with 12 raspberry pi 4 8gb's. lol

if you do not need them donate them to a local school or college? :)

stdout: if you have a spare SSD, a useful learning exercise would be to install a ZFS-root system on the SSD and use that instead...

you REALLY dont want zfs on a sd card...

lol, ok

why is zfs on a sd card bad practice? does zfs have high IO or something?

it can end up doing more writing than UFS, yes. in high-load situations there's not much difference but small random writes can end up writing more data

it also relies on the disk supporting cache flush command and i would certainly not rely on any mmc card to support that

woulda been so nice if they had managed to include a sata port on pis, going back...

mrelcee, yeah, i would suspect that they didn't because of everything sharing the pci bus.

i just retired my pi4, my microsd managed to survive just over 3 years running pihole on it and later i had tossed a couple small dockers on it.. samsung evos do well..

I went for a 256G usb pen for mine over the sdcard, its still playing retropi happily in the kitchen

wait mine might be a 5 ... I have not seen it for ages ~whistles~

lol

5s havent exactly been out ages. i mean a while now..

time flies eh

I got fed up with raspis not being able to use freebsd-update to do os updates. dont remember if arm became tier 1 for 12 or 13.. i was used to it running debian by that point and it woulda been work. :)\

last one i had on bsd was my pine64 board

i didn't know that. 14 installed fine and updated to 14.1p6 without any immediate signs of an issue.

used to get a message if you ran fbsd-update that said it was not supported

majopr PITA copying your data off, imaging and installig apps and returning your data

i ended up having them runing very old versions

at that point they had enough power to be a really nice local DNS server

bind9/isc-dhcpd.

then i went slacker and did it all with pihole on a 4.

now ive dumped my ubiquiti router and stuck a proxmox/opnsense box on so ive got freebsd running my network again. little i3 minipc with 2 cores set for the opnsense VM is just above idling handling the traffic with all the features turned on.

now ive dumped my ubiquiti router and stuck a proxmox/opnsense box on so ive got freebsd running my network again. little i3 minipc with 2 cores set for the opnsense VM is just above idling handling the traffic with all the features turned on.

whoop

Am I correctly understanding 271673 (bugs.freebsd.org/bugzilla/show_bug.cgi?id=271673) in that once its blockers are resolved FBSD gets python 3.12 and can much more easily port 3.13 and future versions?

tercal: ah, you are Özgür i assume?

ZedHedTed: compat.linuxkpi.skb.mem_limit=1 in /etc/sysctl.conf

This seems to be the easiest way to run a command from a script, in a jail, which has multiple steps: sudo jexec freshports /bin/sh -c "cd /usr/ports && make fetchindex"

I suggest make -C /usr/ports fetchindex instead of sh -c "cd … && …", but that of course only works for commands that have an option to chdir themself

dvl: seems reasonable

now it gets more complex to do this one:cut -d \| -f 2 ${PORTSDIR}/INDEX* | while read d; do cd $d && make -V ALLFILES | tr " " "\n" | grep -v '^$' || : ; done

nimaje: why `make -C`

mrelcee: thanks

dvl: jexec freshports sh - <<EOF [newline] blah blah blah [newline] EOF

ivy: I'll try that format, thanks.

you do need to escape $ and \ for that syntax

dvl: ah, no, use <<'EOF' to avoid the need to escape

just checked and freebsd /bin/sh does support that

thanks

I'll try right after my current attemp

kevans: i didn't know about it either but it works!

every time i boot with a debug kernel i find a new LOR, so i may as well just go all in

tw/20

sorry

dvl: you don't need to spawn a sh to cd (if you only need make there)

background -> I'm attempting to implement this plan: bsd.network/web/@piero/113424940736028713

That command pasted above has been running for 35+ minutes so far.

Personally I prefer <<\EOF syntax over <<'EOF' but they are equivalent.

This is documented in the man page: "If the delimiter as specified on the initial line is quoted, then the here-doc-text is treated literally, otherwise the text is subjected to parameter expansion, command substitution, and arithmetic expansion (as described in the section on Word Expansions)."

My current block is:[18:20 r720-02-ingress01 dvl /jails/freshports/usr/ports/distfiles] % find . -type f | head

./clamav-1.4.0.tar.gz

That leading ./ is the issue. I have a list of distfiles.

I was looking for a way to ask find not to include the ./

I could use `find *` but that might hit shell expansion limits.

how can i setup wireless networking with bsdconfig in fbsd 14.1 there is no one wifi network in list and nothing found

also there is no ethernet card found 2.5gb

wpa supplicant custom setup not works

with commands and rc.conf

i don't know

i don't know how to setup the wifi i have two wifi adapters and no one wifi network is found

dvl: maybe `find . -type f -execdir echo '{}' \;`

ridcully: Yes, that helps. thanks.

Also just literally removing the first ./ like this: find . | sed 's/\.\///'

brutex: maybe pastebin what you put in rc.conf

ober: i'm doing this on a live system where the filesystem is read-only. i guess i should just bite the bullet and install it first.

lts: That might be better, because -print seems to remove the entire path.

ProTip for troubleshooting SSL/TLS: openssl s_client -connect example.com:443 -showcerts

Hi!, there's a server (a TrueNAS Core) that suddenly started resilvering. All disks are ONLINE without errors. What can be causing this resilvering?

As mentioned this is a TrueNAS Core, based on FreeBSD 13.1

Might just be a periodic scrub run which is doing its job detecting and silvering as needed. But check /var/log/messages and look for any errors logged there.

rwp: but if scrubbing shouldn't be written in zpool status somewhere?

I look at my notes and see, yes, it should say something similar to "scan: scrub in progress since Sun Jun 4 18:43:22 2023" there.

Then I would definitely look in /var/log/messages and see what is logged there.

rwp: I don't see anything related to zfs there.

zpool status from the command line say anything about scrub?

dvl, martinrame shared termbin.com/kkjt

martinrame, Nothing logged? I would have guessed that the raw drive device would have indicated something like "kernel: GEOM_MIRROR: Device swap: provider gpt/sw0K1G42J7B disconnected." (example from here) or something.

Or similar to "kernel: da5: <ATA HGST HDN726060AL T517> s/n K1G42J7B detached"

there's also "zpool history"

rwp: yes, on oct 31 there was a: GEOM_MIRROR: Device swap0: provoder ada0p1 disconnected.

+1! I had not thought about "zpool history" but that's an excellent log record of events.

and ada0p1 is the one being resilvered

now, why the message "provider ... disconnected", maybe cabling issue?

in zpool history there are many zfs snapshot... and zfs destroy (I have a task to create snapshots once per day and once per hour depending on the dataset and destroy those older than a week).

fetccbhbrdfijlblftkhlhdflhtvvvhrtvvjhulfjnfi

^ sorry

fetccbhbrdfijgvbkvhctehjrrvvhchittvfvfkgveee

If an auto-snapshot process is in effect such as zfs-auto-snap then snapshots are always being created and destroyed. I believe syncoid+sanoid does this as well. That's all probably normal.

martinrame, It does seem a mystery. I know you are going to let it finish. After it finishes check the zpool history and zpool status and see what the ending information is and please report it back here because I will be curious and will look for it.

rwp: yes, what I'm trying to understand is why it takes so long for resilvering. I started on 2024-10-31 and it only did 34%

martinrame, Hold-up. Has that one resilver been running since Thu Oct 31 22:05:33 2024? What type of drive is it. Is it an SMR drive? If so then this is a pretty typical problem with SMR drives.

rwp: those are 2 12tb drives.

rwp: let me see

Let me just come out and say this. SMR drives are simply incompatible with RAID storage. Okay as single solo standalone drives for archive storage. Not for use in a RAID.

rwp: yes, I know that. I took special attention when buying those, there are Ironwolf Pros (ST14000NE0008)

rwp: I'm re-checking

seagate.com/products/cmr-smr-list says it is CMR so you should be good-to-go there.

Here says IronWolf Pro are not SMR: seagate.com/la/es/products/cmr-smr-list

:-)

hey folks, do jails work on arm7 hardware?

I've got an rpi 2b+ running 14.1-RELEASE

the manual on how to create jail says to download a `base.txz` from an arm64 directory – I'm assuming I have to adapt this for my own machine

however, there doesn't appear to be anything other than SD card images available for armv7

martinrame, Then I don't know why the resilver would be running for 5 days now. I would check the interface speed. Should be 6Gbps, right? smartctl -i will list it, among other commands.

zip, First, I know nothing about running on an RPi but you can create your own base.txz image if you can't find one already made for you.

It would be possible to copy your currently running system and use it to create a jail of it. That would be okay.

zip, alternatively, you could mount the image file and extract the base.txz

ah, neat

That would work too.

I'm not sure the image file would contain a base.txz

A jail needs to have a full operating system file system with programs and libraries and everything. Everything except /boot and the kernel because that is using the hosting system.

rwp: yes SATA 3.3 6.0 Gb/s

ah, so I'd just extract anything that's not boot?

martinrame, Totally unpleasant and unexplained then. No idea!

rwp: hehe, yes I know

zip, Correct. And possibly not /usr/local if it already has /usr/local installed stuff there.

rwp: also SMART results are ok.

ports/pkgs install to /usr/local

In a pure base only install /usr/local is empty. And then when you install ports/pkgs they install there.

If you are using zfs then basically the contents of the zroot/ROOT/default dataset minus /usr/local constitute base.

the SD image is a ufs one, unfortunately

The files on disk won't know the difference either way.

indeedy

I just miss out on the helpful zfs stuff

apparently the rpi can happily boot off a USB stick and will do so by default if it's got a boot directory that only contains `bootcode.bin`

I suspect I'd need a bit more experience here before I attemt to surgically transfer an entire BSD install over to an external usb drive though

What is the architecture? arm64-aarch64 I look and see this: download.freebsd.org/snapshots/arm64/aarch64/ISO-IMAGES/14.2

arm7

by far the easiest way to install/upgrade jails is using pkgbase, and there are official pkgbase builds for armv7

Images can be mounted as rafe suggested and files extracted from them too.

wiki.freebsd.org/PkgBase -- however it's still a bit rough to get the initial setup done so maybe an intermediate/advanced use case

I have not played with pkgbase yet and so know nothing about it.

well, let's do this the raw way so I've done it the hard way once

so, next question: loop mounts – same as linux?

I see download.freebsd.org/snapshots/arm/armv7/ISO-IMAGES/14.2 too

mdconfig?

well, I can't run a 14.2 jail in 14.1 if I've understood the manual correctly

That would be running a 14.2 userland on a 14.1 kernel, which is not advised

The next older one that I see is download.freebsd.org/snapshots/arm/armv7/ISO-IMAGES/13.4

Since there seems to be no 14.1 images avialable. I have no information other than what I see there.

zip: Regarding loop mounts; if you want to mount an image, look at mdconfig. If you want to mount a dir at another dir, look at nullfs

looks like i can't `-o compress` my way out of this one

time to wait for a machine that was low-power 8 years ago to decompress a 5gb image

to an SD card

say, is there anything in the rulebook that says I gotta copy the base system out of this img or can I just leave that sucker mounted and use it directly

likewise does a jail suffer for containing `/boot` or does it simply not matter

Am I correctly understanding 271673 (bugs.freebsd.org/bugzilla/show_bug.cgi?id=271673) in that once its blockers are resolved FBSD gets python 3.12 and can much more easily port 3.13 and future versions?

... oh, I'd probably have to fstab it, and I'd have to figure out how to make `mdconfig` run before mount to do that

it is odd that there are no armv7 txz sets on download.freebsd.org

I'm gonna race this little operation with the thinkpad

14 seconds to decompress lol

zip, It's somewhat funny but you can create a jail out of your own running root image. "jail -vc path=/ name=foo command=/bin/sh" and then you have a jail running on top of your base OS.

fair enough

well, mount fails, telling me it can't find `/dev/md0`

I'm assuming that's because `/dev/md0` contains two partitions

It actually has advantages. One can "capture" an installed process and run it in a jail trivially.

well, linux apparently gives us `kpartx` to automagically mount all partitions, but it is not doing that

meanwhile the pi has finished decompressing

and `fdisk` on FreeBSD informs me that the start of partition 2 is 104448

so let's try mounting with an offset

never, ever use fdisk on freebsd

the command you're looking for is gpart

fdisk is some legacy junk from the 386BSD days

ah, okay

well, `mdconfig` doesn't know about offsets

If you have done the mdconfig -f imagefile and there are partitions then /dev/md0* should show the md0s* slices of it.

so there are!

well, you certainly don't get that for free with linux

even if the documentation says otherwise. maybe it doesn't like ufs

alright, I assume `md0s2a` is a slice and that's what I should be mounting

md0sa would be the first BSD partition in the second slice on the disk, which is another weird artifact of 386BSD

md0s2a i mean

well, seems like I can only actually mount `md0s2a` anyway

I would assume that `md0s1` is the MSDOS partition

You can also run: for p in /dev/md0*; do fstyp $p; done

oh cool!

is this the rpi sd image? i think that's probably the only place you still see this old partition syntax nowadays, everyone else has moved to uefi

or gpt rather

sure is!

Linux will autodetect most file system types. In FreeBSD you pretty much have to specify them. So for example: mount -v -t msdosfs /dev/da0s1 /mnt

anyway, I've got my root system image

Also "file -s" will identify them too. Mostly. for p in /dev/md0*; do fstyp $p; done

I assume I want to copy everything except `boot` into my template directory

Oops. I meant: for p in /dev/md0*; do file -s $p; done

if you're making a jail, you want to find the sets, which i think (?) are in /usr/bsdinstall or /usr/dist or something like that

oh, /usr/freebsd-dist

not the whole thing?

well... i suppose you could just copy the image but they have some odd changes like adding the 'freebsd' user which you might not want. if the sets are there they should be more pristine

I got a `/usr/libexec/bsdinstall`

but `/usr/libexec` kinda seems like it's for system utilities that you would never run manually like `getty`

wait hang on

yes, if you don't have /usr/freebsd-dist maybe it doesn't include the sets

could I have just run `bsdinstall /usr/local/jails/my-jail` this whole time

oh, nope, couldn't find a MANIFEST

right let's grab rsync so I can copy this shit over without looking at a blank terminal for an hour

if you can scrounge up an rpi4 i promise this is much easier :-D

armv7 is kind of a weird half-supported release

I'm gonna assume I can exclude `/rescue` from my jail base as well

usually, yes, although you can also build tiny jails entirely out of rescue

honestly it kinda seems like you could fuck about and have a minimal jail that only has exactly what the jailed service needs

but if you're not doing that you can exclude rescue

good good I didn't want to fuck aroudn with getting rsync to hardlink it properly instead of copying the same damn thing twenty times

i have a jail here that does nat64 and it's just rescue and a 15-line /etc/rc, works great

possibly I'm putting myself through a lot of unnecessary pain here

maybe I need this 300mb of clang.debug

see, this is why pkgbase is so nice, you can just not install all this junk

I'm still mad that my damn hp elitedesk mini decided to shit the bed earlier in the week

maybe if I give it the time for its caps to lose power it'll start working again

I don't much want to be using an ancient shitty raspberry pi for this

in the end I went with

`sudo rsync --delete-excluded --exclude usr/tests --exclude rescue --exclude boot --exclude usr/lib/debug -Pa /mnt/* 14.1-RELEASE/`

probably could have yeeted a bunch of `.h` files too

you might find the first time you run freebsd-update it puts all of that back... tbh i'm not really sure how it handles that

mm, maybe

makes me wonder what a minimal jail to run `freebsd-update` would look like

I guess a point in the linux column is that it's nice and easy to build a container out of alpine, which is 8mb

oh, I expect I'll need to set a password and delete the freebsd user in this jail

pkgbase isn't quite that small but i think a very basic installation is < 100MB -- of course once you start adding actual applications it gets a bit larger

I bet

I'm certainly doing this the clunky way

and I'm not convinced that the documentation here makes sense

like I'm just making files all over hte place and they don't seem to link up

I can see we're making a base and a skeleton directory

and it kinda sounds to me like the skeleton is runtimey stuff and the base is the common, not-modified-at-runtime stuff

so we symlink all the runtime stuff in base to `/skeleton/etc` and so forth

and then it just doesn't seem to add up in the suggested fstab

oh wait, maybe it does

wait are you doing the "thin jail" stuff? don't do that, it's awful

sure seems like it

just make your jail a complete copy of freebsd, no special tricks

ugh but that's 1.2Gb

well, I suppose I only care after the second jail

and yeah this is clearly too much work to simply spin up a new service

hi, i was wondering if anyone tried git-prompt.sh, which, starting at git 2.47, supports most shells, including freebsd /bin/sh (previously it supported only bash and zsh). i wrote the compatibility fixes, and was wondering where people tried it out and maybe have some feedback. thanks.

zip: once you get zfs this won't matter as you can just enable dedup :-)

s/where/whether/

i believe it is available in freebsd together with the git packags, except git@lite (thanks to dch for looking it up), at /usr/local/share/git-core/contrib/completion/git-prompt.sh

ivy: aye. well, maybe I'll try moving all this to a USB thumb drive on zfs or something

if the installed version is before git 2.47, then the latest can be fetched from upstream, ans it's a stand-alone script - git.kernel.org/pub/scm/git/git.git/…in/contrib/completion/git-prompt.sh

as*

I suppose technically i could use `zfs send/receive` to dupe my laptop to elsewhere and then have a go at bsd on the laptop

actually I've got `/pool/ROOT/void` and `/pool/home` already

I wonder if I could just stick BSD on here next to it

at the very least I imagine that I could put BSD next to it in a way that might require me to juggle bootloaders to go back to linux and no further effort

ugh, spamcop dnsbl is ipv4-only

the wife pinched my spare thumb drive to dump the data off a laptop she was formatting so I guess I have to go get a new one for projects

thank you for the help, by the way!

Hi! Is there any BSD based distro to recover/fix Windows installations?

not sure bsd is the best tool for that

as bsd can barely mount the filesystem

doesn't windows have its own recovery tools?

zip, I think it does however I haven't use them.

I remember long time ago booting some linux distro

There's a number of very handy, Linux-based "utility" distros.

i would start with the windows recovery options