hi friends, I have a problem. I'm on freebsd 14.1 and I'd likt to upgrade to 14.2 but the kernel module of virtualbox is not compatible with 14.2 as you read in the install message here:

freshports.org/emulators/virtualbox…oad%20on%20FreeBSD%2014.1%20kernels.

how do I knwo when the new kernel module will be availaible in the package system ?

I don't want use the ports

mystic: The issue seems to be in relation to bugs.freebsd.org/bugzilla/show_bug.cgi?id=219913 , this bug is now patched (not fixed, but shouldn't be an issue if you have < 1024 threads). The package is indeed available in 14.2, if you are still uncertain about its reliability on 14.2 due to the message on the package description, then I would recommend submitting a bug ticket on the

bugzilla for clarification of 14.2 suitability (and potential update of package description).

Alternatively, there's other virtualisation options (e.g. bhyve)

varsis: thanks.

hi, how should i send this patch? pastebin.com/PSHNradV email? bugzilla?

bugzilla i guess

quickrr:

thanks

where would the mail go?

i guess that would work

hey there everyone i was transferring 22 gigs of data from a zpool of mine to an ufs bsd volume and when the transfer was about to finish(was at 17 gigs i noticed the transfer just froze up), then i checked zpool status and noticed an I/O error and did a zpool clear, but the command is taking too long and i notice that it keeps writing to a disk belonging to the zpool, what could be the issue and how should i go around resolving, also s

ince the action was a move action are the files i was transferring but wasn't transferred due to the weird I/O error gone now?

zpool status says 13 data errors, but i can't see them right now because the pool I/O is suspended

it's so weird though, i was just transferring files normally, and for the pool to get suspended like that isn't something called rock solid would do

pool will suspend the io when there is no redundancy for broken data block

this is done to avoid your data getting inconsistent (some blocks written and some not)

that depends on failmode setting as well

yep.

what's the more or less used tool nowadays to manage backups through snapshots/send/recv ?

I was an happy zrepl user but it has unresolved issues with FreeBSD 14

can't choose between sanoid/syncoid, zrep, zxfer, zfs-replicate, ...

tsoome: wdym by redundancy and how did broken data blocks got created while a normal file transfer, and why does it keep logging "pc ZFS[xxxxx]: pool I/O failure, zpool=MA error=6" continuosly to /var/log/messages and consuming so much CPU resources

ah they left

slowly introducing freebsd at work. only on vms thus far, but I'm sure I'll get them to go native at some point.

currently setting up a fancy little setup with geli and gmirror for backups.

I'm running FreeBSD at work for 15+ years .. :)

small infrastructure (~15 physical servers, 50+ jails), but FreeBSD is powering everything!

i started at this job less than a year ago. first time i'm not only doing development but also administration. one of my assigned tasks is making the company depend less on MS shit.

yeah, personally i've been running my infra on freebsd since something like '09 i think. and since a year or two later also on desktops/laptops.

Nice assigment

i got pretty fancy stuff going, at least for a single person. all disks (minus /boot) on all machines including the hetzner dedicated machine are geli-encrypted and on both servers, everything is mirrored with gmirror.

ludo_: yeah, but sadly i haven't found a proper replacement for teams yet. xmpp clients are slowly getting there, but last i tested with a coworker (~half a year back) it looked like it'll be at least another year before there's any chance of it actually working.

and i still have to see about deploying .net webshit on *nix; that part might actually be way harder on freebsd because the .net build system has a built-in docker target. tho my longer-term goal is getting them hooked on python for webstuff. :P

did you try thinkgs like matermost? matrix ?

we are running mattermost here with nextcloud

(although I would not recommend Nextcloud)

ludo_: hackerspace i helped found used mattermost for a while, but it never really kicked off. also doesn't offer video conferencing, at last last i saw it.

yeah, nextcloud can probably even do video conferencing, but i'm not advocating getting rid of MS shit and then deploying PHP ^^

my first two jobs were php, i made an active decision to cut it out of my life :P

The problem with Nextcloud is that it's slow as hell

matrix does messageing + video cofrecing

yeah, also sounds like one of the biggest attack surfaces of any webthing. our main "server" is currently some small vm. neither gonna use zfs nor nextcloud on that sorta setup. :P

ludo_: i don't trust matrix' security and don't want to give false impressions of security.

was just giving ideas, as I don't like matrix either

aye. didn't mean this as any sort of attack in case it came over like that.^^

mhhh, mattermost seems to have video conferencing now… it may not be decentralized or e2e, but this might be an actual replacement for teams by now. thanks for the reminder ludo_, wouldn't have thought of checking it again.

yw

hah mattermost is open

i never looked it

ketas: yeah, looks like a freemium model with FOSS base like grafana.

always a wee a bit suspect, but something something ideal world.

like nginx, asterisk, etc?

it could fuck up too

that model

yeah, it introduces a new balance: community vs. profit. many projects seem to strike the balance pretty well, but we regularly see companies abusing their position, too.

it gets particularly cursed when features and sometimes security patches become premium option

instead of just support

ketas: yeah, that's the sort of situation where a community fork is in order.

how much of a requirement is overwriting the entire .eli with dd actually? with data auth, i expect without it i'll see a bunch of "failed to authenticate", but things should still work and be rectifiable later by writing files on an existing fs, correct?

mage: fwiw, i like/use sanoid/syncoid. sanoid does regular snapshots, and i regularly pull them w/ syncoid via cronjobs. doesn't take long to get the hang of

how do i explicitly register an efi bootloader again? it was with one of the efi* tools IIRC…

oh, nevermind that. looks like this thing doesn't support efi… o_O

mhh, bios bootloader installed with gpart bootcode does nothing but spam "Loaded only 545k". i made the partition for it small (1M) so I retried and set it up as 128M, but still get the same behavior. any ideas?

(this is a vm running on windows+hyperv). trying the efi route just gave me a black screen, efibootmgr essentially claims efi isn't supported (tho gpart shows whatever "efimedia" is…

phryk: you have to troubleshoot and fix it

mzar: no, really? :P

error message isn't very specific tho and i never had similar problems before. looked into hyperv options if efi is like an option you can enable, but found no such thing.

oh, i have a very dumb idea.

SONOFA

it works after i made the gptboot partition even smaller than before.

my assumption is that the thing tries to load the entire partition into memory but has a "sane" limitation for that to 545k. made the partition 512k and i get beastie

phryk: thanks for solving, of course freebsd-boot partition should not exceed 512k

there must be a way to break hw.snd.default_unit up so you can have speakers on one device active, and a mic on another.

Ober: there is

Ober: audio/virtual_oss

Hey there! I'm trying to run jdupes on large datasets, but on occasion the process stops. Currently when I use PS, it shows it in status "D+"

I've found in most cases that I can resume the copy operation that I've suspended in the background for a short while, and the fdupes process will resume.

All I can figure is that the disks are going into a power save mode, but I've never set any modes that I can remember (base install goes back over a decade, but is up to 14.2R)

top shoes the process status as "range"

wipt: yeah, sounds like the processing is waiting on I/O - so as you mentioned, something to do with either the disk going to sleep for some reason, or I/O is maxed out

wipt: you can check the disk status (and more) with camcontrol --> man.freebsd.org/cgi/man.cgi?query=c…ion=8&manpath=freebsd-release-ports