Is there a way to get brasero working on freebsd or any GUI burner?

okay, i think i am going crazy - when i execute this command: find . -name "@eaDir" -type d -prune -exec rm -rf {} \; i get an error find:17:command not found: ;

for the life of me, i see no issue wit this command..

RhodiumToad: You use ipfw, no?

voy4g3r2, If using -prune I think it never makes any sense to follow it with a logical AND -exec, instead when using -prune I think it should always be followed by a logical OR with -o

But that sidesteps your question about command not found. That makes no sense either because rm would certainly exist. Was it a strange character next to "find" that is causing find not to be found?

voy4g3r2, I retract my statement now that I look more closely and see what you are trying to do. Logical AND is what you wanted. Sorry. But then is there a bad character next to "find" causing find not to be found?

not that i can see, but i retyped the command and got this error: find: rm: unknown primary or operator

this synology device crash still haunts me

i got tons of @eaDir throughout everywhere

Which shell are you using?

zsh

Let me suggest using echo in front of the rm to see what commands it would run without actualling running it. find . -name @eaDir -type d -prune -exec echo rm -rf {} \;

ahhh

i had a damn typo

errr

Every time you make a typo, the errorists win.

haha, that is quite jarring as i have roger waters amused to death playing in background

just the theme.. went dark

but much cleaner now

I thought you were talking about John Waters for a second.

i was never a fan of his movies.. hairspray was okay

that is if you are referring to the writer/director, this guy: en.wikipedia.org/wiki/John_Waters

Title: John Waters - Wikipedia

i need an openvpn mirracle man

mane have u tried transport UDP protocol ?

what you mean?

if it is set to tcp

i tried tcp, works much worse

proto tcp -> proto udp

yes with udp it works better

x))

out of options

XDDD

Title: Speed up OpenVPN and get faster speed over its channel

have u red this article ?

bookmarking

<33

also follow the comments

Hi all, what's the best way to move to EFI on a system that only has a 4 drive raidz1 ATM? (and free USB slots) gpart(8): x0.at/-jBt.txt

mane you will likely get slightly better perf with wireguards if thats an option

moviuro: 90% of the disk space needed for UEFI boot is /boot/loader.efi, a single file thats written to a FAT12 or FAT32 partition, of "EFI" type.

hi! I'm having troubles on my 14.0-RELEASE-p3 system,

moviuro: so in principle you could add an EFI partition into /adaX/**** the 492K bit

moviuro: format it appropriately, and copy loader.efi in as .../EFI/BOOT/BOOT64.EFI and this will work

moviuro: you will also need to check & adjust partition info in /etc/fstab like swap if anything else changes

617K /boot/loader.efi --- 7814035456 1672 - free - (836K) talk about "fits like a glove"

doesn't the kernel also need to be on the ESP?

the other option is to remove the swap partition, re-add a larger 492K+headroom partition, and then re-create the swap partition, aligned and a bit smaller

woops I looked on the wrong system!

mine here is 644K. I have an additional custom boot loader in the efi partition...

I tried today to fetch and install updates to p4 for the openssh advisory but I get errors regarding the stat command not being found... any suggestion?

moviuro: so yes you need option 2, carve up swap, and fix it

how much trouble would "no more swap" be?

moviuro: you'll still have swap, it will be 1.98G for example.

that sounds *really* tight for an ESP

CDiMa: first I'd use `freebsd-update IDS` mode to see what is missing, and then deal with that.

moviuro: 200M is fine. how many EFI boot loaders and tools do you plan to install? mine is 1G here and I'm using 14M total, and I have 21 additional EFI versions and boot loader tools.

the kernel lives as usual in zroot/ROOT/default or whatever boot env you have

moviuro: alternatively, add it at the *end* of the disk, that 836K spot is just enough. Not sure if firmware is smart enough to use that, probably but its not my thing.

good to know, thanks dch . On Linux I have 248MB used on /boot (which includes kernel, and efi loaders)

aah yes linux has extra stuff

CDiMa: basically you can unpack base.txz from source, which will have all components in it, but it may also overwrite some customised files.

CDiMa: some of people.freebsd.org/~dch/posts/2021-02-23-sideloading-freebsd.html may be useful for you.

Title: Side-loading FreeBSD versions using Boot Environments - Random Musings

it makes sense that the fbsd efi loader can read ZFS :)

it would be very awkward if it didn't ;-)

dch: what puzzles me i s what may have removed stat from the system

CDiMa: is this system quite a small one?

not really

60Gb of disk space

hmm during BETAs I saw a few small systems run out of memory, lose some of the update files, and get borked

CDiMa: more on the cpu/ram side I meant

but zfs rollback and re-run they had no issues

didnt see it in release upgrades

dch: I'm using stub loading on my everyday linux machine, with the EFI loading the linux kernel directly. 27M /boot/EFI/Linux/archlinux-linux-lts.efi -- and systemd-boot can't read encrypted btrfs 100K /boot/EFI/systemd/systemd-bootx64.efi. I'll dig into carving up swap space for the ESP, though that will probably have to wait until January

4Gb ram and 2 CPU

CDiMa: the ones I saw this on are 1G RAM + 1 shared vCPU.

anyway, `freebsd-update IDS` will tell you what's missing, and you can add it back from base.txz as required. Just make sure you run freebsd-update after.

dch: well, 2 vCPU for that matter, but of a largish server

TBH I am not sure if freebsd-update will put stuff back if it's missing, but I don't expect it will run well without stat(1)

dch freebsd-update IDS has the same behaviour: Inspecting system... /usr/sbin/freebsd-update: stat: not found

dch: and than lots of stat not found lines

CDiMa: its amazing it even boots... welp, what else is missing?? then keep copying missing bits from a working system until at least IDS works

CDiMa: for info did this box go from 13.x -> 14.x directly, or did you try one of the betas or release candidates in between?

directly

huh thats not good

I had 14-RELEASE working

a couple of days ago I fetched and installed p3

uptime tells a week ago

today I did a freebsd-update fetch and started getting that error

dch: btw this system started from 10 or 11 and I upgraded it regularly

dch: I always used fetch and install without any troubles

dch: what was the correct way supposed to be?

dch: anyway I can think of rollbacking it

CDiMa: your way sounds correct. I'd not rollback, just copy stat(1) and whatever else is missing from another system, and use `freebsd-update IDS` to clean up.

that seems like less work

ok, I just have to catch a 14-RELEASE system

I see from the snapshots that I jumped from 13.1-RELEASE-p9 to 14-RELEASE

dch: 4 GB RAM is plenty btw

my apu4d4 has 4 GB RAM and a quad core Jaguar and it has never had issues with updating/upgrading

yes, I run many 1G systems without issue, with zfs.

zfs can be quite a bit hungry if there's lots of disk I/O

also I think `freebsd-update fetch install` might theoretically recover missing files

zfs ARC is disabled on systems with 4 GB or less RAM anyway

Remilia: trouble is that the fetch already fails

huh

maybe you could just get the matching base.txz and extract that

the one from 14-RELEASE

Remilia: yes, but there is a risk that you end up with some important lib that is from p0 and some other tool from p3 and stuff doesn't work.

hmmm

dch: /usr/bin/stat is reported 0bytes in the txz I just downloades

d

dch: that sounds impossible, tbh

plus you can always freebsd-update right after

if you extract base.txz you are replacing the entire base system tree

and old base will always work with new kernel

(well, within a single release)

[10:18:43] <CDiMa> dch: /usr/bin/stat is reported 0bytes in the txz I just downloades <- check the size of readlink

because stat is a hard link of readlink

$ ls -i /usr/bin/readlink /usr/bin/stat \\ 104951 /usr/bin/readlink 104951 /usr/bin/stat

thanks Remilia

Remilia: I simply made a hard link in place to readlink that was present on the system, this should take care of dch's worries (if nothing else is missing of course)

`freebsd-update IDS` will tell you about missing/altered files

running now and taking ages in the "Inspecting system..." state... looks like stat il present and working

is

\o/

it may complain about passwd and some other similar files

hrm, so i keep getting an error about dbus unable to bind socket to /tmp/XXXX, i've lowered the security level, allowed raw sockets, etc. what else could keep a jail from starting a dbus socket?

Demosthenex: probably boring permissions?

su to user expected to create socket, cd /tmp/; mkdir XXXX; see if that works?

dch: i've tried changing the dir and other things, it doesn't appear to be anything simple, and it looks like a jail issue

i moved tmp off and on ramdisk (tmpfs) and checked perms already

i was checking the jail.conf for allow.raw_sockets and allow.socket_af, but those didn't help either

Demosthenex: are you trying e.g. to run firefox in a jail, or similar?

I assume dbus etc are all inside the jail

i'm trying to run signal-cli (java wrapper to cli signal messenger) in an ubuntu jail

headless, no X

aah

see vermaden.wordpress.com/2021/12/15/secure-containerized-browser its for firefox but has a couple hints for dbus related stuff in there

Title: Secure Containerized Browser | 𝚟𝚎𝚛𝚖𝚊𝚍𝚎𝚗

and for some reason, it uses dbus. but even with dbus-run-session i can't get it going

i also allowed sysvipc

and antranigv did a few extra dbus-related steps too weblog.antranigv.am/posts/2020/11/freebsd-signal-cli-scli

Title: Signal-cli with scli on FreeBSD | Antranig Vartanian

I bookmark all the things all the time

unfortunatly the scli and signal-cli ports are so far behind, they no longer function at all

which is why i tried an ubuntu jail

omfg, dbus may be trying to update linux capabilities?

dch: freebsd-update IDS reported lots of file permission mismatches

dch: how came it got all messed up?

Demosthenex, Could try "truss" to check on "dbus"

parv: i was stracing earlier, it provided no insight

Demosthenex, Ak

i was trying to run dbus as a user session, now i'm trying to start the system one, it failed silent

so i run the command from the init script manually, and get an failure to drop capability, operation not permitted

CDiMa: no idea, I only saw missing files in beta, and that was because I ran out of ram during upgrade. I trawled bugzilla but didn't see anything relevant. Would be worth logging a bug if you're in a position to compare (e.g. zfs snapshots pre & post to compare)

dch: and how can I fix that? a fetch and install cycle would fix that?

CDiMa: see my sideloading page above if you have zfs. For ufs, you can just do the untar of base to fix it, as root this will also (re)set permissions.

but you don't have the nice snapshot/rollback support, so its a bit more fiddly

I surely have a few snapshots both of 13.1 and 14.0

CDiMa: bugs.freebsd.org/bugzilla/show_bug.cgi?id=261329 may be relevant, also if you do upgrades with a custom umask, maybe that changes things?

Title: 261329 – freebsd-update IDS has nonsensical output when link permissions are wrong

There is "mtree" to fix the issue of permissions (& create missing files). On forums?.freebsd.org, one posted a mini guide

aah yeah thats a good way to fix permissions etc, but will it fetch missing files if they're not just hardlinks?

Don't know

I don't think so

with mtree I should create a specification from a fresh installation and then apply it to my system?

Yes. Also see /etc/mtree/

s/Yes/Perhaps/

there should be a law against starting more messaging platforms :P

;-)

so scli/signal-cli is out of date and doesn't work, linux jail sucks. bitlbee with libpurple doesn't have the signal libpurple lib...

CDiMa: hey can you share your output of `freebsd-update IDS` somewhere please, ideally as PR in bugzilla?

how's a civilized human supposed to use this stupid messenger

only on a proprietary closed-source smartphone ?

dch: can't type there

even if i didn't care about the cesspool of cell phones, they lack keyboards

ditto

i despise the progressive war against power users in the software world

more than the proprietary crap

webapps don't even have hotkeys, and browsers won't let you set your hotkeys anymore. "what's good enough for crayon eating icon clickers is good enough for you citizen"

dch: do you mind if I give you a pastebin link?

CDiMa: a good start, yes!

Title: freebsd-update IDS - Pastebin.com

dch if I skip the /usr/src stuff there shouldn't be that much to fix

CDiMa: other than stat(1) were any other files missing?

not that I'm aware of

yeah, the SHA256 stuff all looks like legit changes

[12:15:56] <Demosthenex> webapps don't even have hotkeys <- it is not even about hotkeys, even the basic UI interactions that have been there for 40 years are missing

for example the click-and-drag menu usage pattern

CDiMa: persimmons aside, everything seems fine

you can ignore etcupdate stuff, that is freebsd-update merging /etc stuff

mmmhhhh persimmon pudding

<Remilia> [12:15:56] <Demosthenex> webapps don't even have hotkeys <- it is not even about hotkeys, even the basic UI interactions that have been there for 40 years are missing

I agree 100%

because it's hard for browsers and web apps to emulate that experience and make it seem authentic

it is almost impossible for menus because drag over text is selection

why would you drag over text in a non-webapp?

like, at some point in my life I stopped using clicks to select menus/submenus

because it requires more presses than dragging

signalblue: menus are arguably text

so how do you select menus/submenus?

using the keyboard? that's what I do

ok, thanks to dch and Remilia, i fixed permissions and updatet do 14-RELEASE-p4, all seems fine...

cya

if you are using the mouse for that (for keyboard, it is faster to use associated keys) which I do relatively often, you can press the mouse button on the item in the menu bar and drag it, without releasing, to open submenus

releasing activates the selected item

dragging while holding the button activates submenus without requiring a click to switch

lemme maybe record it

oh I see

just slick and hold something

yes

then move the mouse to whatever

click not slick

and when you release the item activates

it is a *lot* faster than single clicking items

yes that's not an option with webapps

and you cannot do this with drop-down menus in web apps (because text selection prevails) AND in custom-drawn interfaces (hello Firefox)

I think the interaction I described works even in Motif

pretty sure I used it in CDE and IID

I wonder if anyone still uses CDE as a daily desktop these days

Solaris users maybe do?

oh it also worked in OS/2 Warp 3

and in Turbo/Borland Pascal/C++ 7 for DOS too iirc

yes but Solaris is dead and Oracle has no interest in reviving it

signalblue: essentially electron/react ignore and obliterate 40 years of UI/UX development and design

yep I agree

and newish Solaris installs do not have CDE

someone ported and maintains cde in the ports tree, so maybe someone uses it as daily driver on freebsd

what is especially annoying about this 'cross platform' menace is that not only it is bad for power users, it is also terrible for accessibility

also note how just about nearly every messenger app looks the same and almost none of them, including Matrix's Element, support multiple windows

everything runs with the smartphone/tablet screen paradigm, almost everything is spaced to hell 'for touch interfaces'

Is it possible to run a FreeBSD 13 kernel on 64MB RAM? Can't seem to load the kernel with less than 80MB.

how much tuning did you already try? if you just tried -GENERIC then probably by compiling your own kernel and removing all kernel modules you don't need

Quite a bit of tuning. I've had a system running on 11.4 for a long time. I have the kernel file down to 4.5MBytes and I think that's as low as I can go.

I'm using nanobsd and targeting an ancient Soekris Net4501 with 64MB but testing with a VM. At 80MB RAM I have a working system I can login to.

At 76MB RAM the boot loader fails with elf32_loadimage: read failed. At 64MB I get no messages at all.

Demosthenex that's pretty sad. I should have a look at the port.

Demosthenex where did you get at?

antranigv: sorry?

longword you will need some hacking.

antranigv: the scli port is 0.7.1, current is 0.7.3, and signal-cli port is 0.9.0 and the current is 0.12.7. versions of signal-cli before 0.11 don't work at all with scli

longword for generic use case, the loader heap is configured to use 64MB and you still need memory for actual kernel and modules.

antranigv: i should have considered just trying to update the port myself, even locally. duh. ;]

tsoome_: I was thinking it might be something like that. Any pointers as to where one might nudge that loader heap in the right direction?

longword bios version? boot/i386/loader/main.c

Much obliged

+ you most likely want to strip as much as you can.... maybe even to the point to load kernel directly not with loader but with second stage

I'm moderately sure I'll be fine once I can get the kernel loaded. Booting with 80MB there seems to be plenty of memory free for userspace.

stand/i386/loader/main.c seems to be the thing

Title: 2024 FreeBSD Community Survey is Here – FreeBSD Foundation

Victory! My enemy was in stand/i386/libi386/biosmem.c - #define HEAP_MIN (64 * 1024 * 1024)

With that slapped back down to 16MB I can boot the kernel and login to a super minimal system with just 40MBytes RAM.

Thank you to nimaje and tsoome_ for the assistance

Very nice.

it feels that should be somehow settable