how do you update zfs list ?

i tested out poudriere and now deleted it but zfs list still shows it

oxbar: uh, you don't?

oh

zfs list shows the current system configuration

even though the folder dosen't exist anymore ?

can you demonstrate what you mean by 'the folder doesn't exist anymore'?

it fixed itself thanks..

oxbar: If Poudriere created a jail (using ZFS) it'll remain even as a dataset even if you remove Poudriere.

ok

You need to delete the jail via Poudriere first and then delete Poudriere.

Or, you can simply destroy the dataset.

yea i destroyed it

-r

That'll do it.

ivy: I committed a fix for one of the IPv6-only wireguard issues you found, sent a patch to wireguard-tools upstream for the other

kevans: 7121e9414f294d116caeadd07ebd969136d3a631?

yeah

that fixes the "I accidentally added an IPv4 address" one

DragonflyBSD takes yet a different approach and they don't actually teardown the peer if you misconfigured something like that

thinking about that one a bit, it would simplify things

their theory is that you're probably just going to try again anyways

that's admittedly pretty hard to argue against. in the off-chance that you aren't, having to remove that peer isn't that bad

ah, you committed the 'bad' version of the PR since no one bothered to make a better one? :-)

no I committed the better version

their patch mostly just applied, only had to tweak it slightly

ah ok

they took away a point of improvement from the exercise, too, so that feels like a win

i suppose i should actually test this as you bothered to commit it

yes, yes you should

god you're SO demanding

first i have to demand you fix all these bugs then you demand i test them, mutter mutter

this world is so unfair gahhh

let's see if LINT-NOINET still builds

the song of freebsd committers youtube.com/watch?v=1sp33WgVWMA. that thumping sound is all the commits going past

hmm, this is confusing, i did build LINT-NOINET and installed it when i reboot uname -a says: FreeBSD freebsd15.eden.le-fay.org 15.0-CURRENT FreeBSD 15.0-CURRENT #0 lf/dev/traceroute-tests-n276342-a1b43b7bf0fa-dirty: Thu Apr 10 16:50:37 BST 2025 ivy⊙felo:/home/ivy/obj/home/ivy/src/bsd/traceroute-tests/amd64.amd64/sys/GENERIC amd64

okay i see, NOTES has a makeoptions DESTDIR=/tmp

tricksy

kevans: somehow this has completely broken my test host's ipv6 connectivity

Are all patch levels of a given supported release supported thru the whole release cycle or does older patch levels go EOL when the new patch is released?

ex: 14.2-p3. p2 and p1 are considered out of support?

skered: the patches *are* the support for their associated revision. if you're running 14.2, you install 14.2-p1, 14.2-p2 etc. to get the support updates for 14.2

it's not that 14.2-p1 is "out of supported", it's that if you're running 14.2-p1, you're running a supported release but you chose not to install the fixes that were provided as part of that support

I guess it's more of ports question and I'd guess it's the same for releases that just end of life. When can you start to use a given pX fix that doesn't work in <pX?

I think the recommend time frame is a month or so.

this should not affect ports at all, a release patch should never introduce functional changes

except maybe for things like tzdata updates, but even so

should? maybe do? yes.

if you're maintaining a port, you can't rely on anything except the oldest supported 14.x release because that's what the package builders use

For instance the latest daemon update fixes something that would otherwise be broken.

ivy: I understand that but we're talking patch levels now.

ah, so your question is, you have a port and it requires a daemon fix that was only introduced in an errata?

So if I stop hacking around the fix from p3. Do you just just tell p1/p2 users upgrade

Correct.

i would suggest asking ports@ at this because it's a fairly unusual situation and i don't think there is any official guidance

i suspect the answer might be that you still have to support 14.2-RELEASE though

My gut would say it would follow the same rules as a release going EOL. But we'll see.

(i don't agree with that, i just think a lot of people will say that)

ivy: what

kevans: wait, i am building GENERIC to test

okay it's not your patch that broke it, GENERIC works fine, GENERIC-NOINET (i just made this up) does not have ipv6 working at all

funky

ugh i am going to have to debug this aren't i

yeah, who knows when bz will update his systems next :-p

bz bikeshedded my diff so i hope all his networking stops working

(this is a joke don't @ me)

hah

Other than editing /etc/fstab entries, what else would I need to do to move a drive from ada3 to ada0 (when I installed, I didn't realize that only ada0 was SATA3, and the other 3 are only SATA2.

break19, I suggest using GPT labels if at all possible as that avoids the kernel selected numbering on the ada# driver.

To answer your question there is nothing more needed to be done.

Humorously I have an ASUS motherboard with HUGE marketing letters for the 6.0Gb/s SATA. But only 2 of the 8 ports are 6 and the others are 3Gbps. Things like that catch everyone. I installed an LSI2008 SAS/SATA HBA card.

rwp: Thanks. And this is a 2013-era Optiplex that's been upgraded to 16G ram and had a handful of drives added. My HBA card is enroute, and I have 3 SAS drives ready to replace three of the 4 drives currently in it. :)

For my home use I have been buying datacenter used SAS drives for cheap. Have yet to have one of those fail on me. Have gotten some good deals. Have gotten some just okay deals. But with zfs+raidz2 plus backup I am feeling pretty good about things.

I bought 3 cheap drives, not noticing they were SAS. But its an excuse to upgrade lol

reminds me I bought a SAS by mistake and do not have a adapter XD

You might find those SAS not faster than SATA but with file system buffer cache and RAM that difference is not signficant in the end. And so far the enterprise SAS drives have been rock solid for me.

thedaemon: Either get a cheap lsi, or send me the drive. :)

rwp: Maybe. But it's a media server for a 3 person household, so speed isn't really a huge deal.

Is pkg frozen? I've not seen any repo updates in several days, and I'm on "latest" ...

there were a/some(?) issue(s) which lead to many fallouts and skipped packages, afaik they should be fixed and only waiting for mirrors to sync the last successful build of the pkg repo

I feel the same way. For the house I don't need the performance level as the 300 person design lab needs. Reliability is what I am needing.

nimaje: hm, alrighty then.

nimaje: is there a list of mirrors that are in/out of sync somewhere?

How does one live in the terminal and finally things like word processing watching movies etc… ? Isn’t that kinda hard ?

break19: yeah it's well out of sync. last build it seems there were lots of failures

only real way around this is to either set up a poudriere instance or upgrade the old way with something like portmaster

break19: last quarterly is more recent than 'latest'

cool

oxbar: TeX and why would you watch a movie in a terminal? but you could use mpvs caca, tct, sixel or kitty video output

word processing part is easy

ivy: never try to run a LINT kernel

goodness gracious, what happened?

kevans: the plot thickens: this is an hn(4) bug. if INET isn't defined, it comes up with a random non-default mac address that doesn't work; resetting it to the hardware address fixes it

no, actually it's an ifconfig bug. it seems to auto-detect an iP address as a MAC address if INET isn't enabled

ivy: huh, that's wild

i guess no one noticed because usually you wouldn't try to do this, and also it only works with certain IP addresses and not others

hopefully easy to fix, but i got distracted and will have a proper look later (the basic problem is that ifconfig defaults to "inet", unless INET isn't in the kernel, then it defaults to "link")

big fan of being explicit

interesting story, thanks for sharing ivy

kevans: unfortunately there's a bit comment in the source about why we can't just force people to write "inet" all the time :-d

s/bit/big

big fan of deleting comments I don't agree with =-D

oh that's hilarious

well, part of is you'd now have to say 'ifconfig hn0 link up' which i could see being contentious

Can a pr be opened up again ?

woohoo! go based packages have returned to the amd64 repo

oxbar: yes, if there's a reason for it, but i think only triagers and admins can do that. what's the PR?

I'm trying to identify which slot my drive is in. I know it is "da1 at mrsas0 bus 1 scbus1 target 2 lun 0" - I'm going to guess that translates to "Physical Disk 0:1:2" In my Dell R730. I tried sesutil map, but that's useless for this (dpaste.org/tkjvd).

is choosing which disk to boot off of, something you set in hardware or in freebsd itself? on macos, it's entirely a software setting

ivy: bugs.freebsd.org/bugzilla/show_bug.cgi?id=283956

The issue wasn't resolved the guy just installed another os and that was it

GoSox: the system firmware (legacy BIOS or UEFI) chooses the disk used to start the boot processes. once it's loaded the boot loader, the boot loader decides what to do next - e.g. for FreeBSD it looks at a hardcoded root path, or for a bootable zfs pool

oxbar: do you also have the same problem? i don't want to reopen it unless someone else is still affected (if so, please put a comment on the bug)

ivy: yes i have the same problem.. so does this dev forums.freebsd.org/threads/audio-ca…tiger-lake.96345/page-2#post-685550 he said he just didn't try to fix it

GoSox: Is this a general question, or is there a problem you are trying to solve?

no i'm just wondering how it will work when i actually set up my bsd server. on my previous mac servers, i'd just make a primary boot drive, and a backup boot drive. Since its all set in software, i can screen share in, tell it to boot off the backup drive, restart, then reconnect to screen sharing on the backup drive and run repair utilities as needed on the primary drive

i'm wondering how a process like that would work on a non mac

GoSox: what macOS is doing is almost certainly just changing firmware settings. on freebsd you may be able to do this with efibootmgr(8), although PC firmware is of variable quality so it may or may not work

there is efibootmgr if you want to configure the uefi boot manager of your system from within freebsd

(it *should* work on most modern, average quality EFI firmware)

i'm eying an old surplus HP proliant server

but i haven't bought anything yet, i'm still working on setting up the basic server i need on a test machine

and i never have time to work on that so i'm making basically no progress :/

oxbar: did you want that bug reopened? if you can just put a comment on it like "i am also seeing this" it makes it more obvious what's going on rather than just reopening random bugs :-)

ivy: I will when I get home.. stopped at a red light lol

oh okay :-) no rush just ping me here (unless someone else gets to it first)

i'm not actually entirely certain you can't just reopen it yourself but i don't *think* normal users can do that

Ok

ivy: you'll be able to with cluster login

also killing spammer comments/accounts

kevans: i already could as i have magical bugmeister role, i was just wondering if oxbar could do it themselves

ahh

or triager role, whatever it's called

not that one that makes you a bugzilla admin, i don't have that

*nod*

Hi, I'm sending a zfs dataset to an ssh'd server and apparently is sending more data than the origin. zfs list shows 5.6gb, but the progress of the zfs send | zfs recv shows now 6.4gb and counting. Is there a way, on the destination system to see how much data is received?

ivy: I see you on discord

me? on discord? this is more likely than you think...

Aw

My bad