im trying to update and got this message

Waiting for data... (^X or interrupt to abort)

what's /var/db/mergemaster.mtree

Title: JRG Systems - 802.11ac on FreeBSD with wifibox

Oops. Replying to the wrong channel. Sorry for the non-sequitur.

nmz, I guess I could answer your question though. mtree(8) is a command that scans a directory hierarchy and stores the result in a file. Then on a later scan mtree can compare the current files to the saved db of them and report if any have changed.

mergemaster(8) uses this to know which files need to be prompted for merging.

should I erase it?

more like can you?

i just ran a 14.1 unattended bsdinstall. i got 2 nvme drives i wanna mirror and i want a mirrored swap. install finished but there's no swap? termbin.com/t6j0. system boots too. just swapinfo returns nothing, top shows no swap

in /etc/fstab there's 2 lines. 1 for efiboot0 and 1 for /dev/mirror/swap.eli none swap sw 0 0

is it a bsdinstall bug?

zfs problem in 14.1 unattended bsdinstall. i got 2 nvme drives in mirror and i want a mirrored swap too. install finished and it boots fine. but there's no swap? termbin.com/t6j0 swapinfo returns nothing, top shows no swap

why have a zfs dataset mounted in /zroot/ instead of directly off /?

polyex, this depends how you use bsd

if you use beadm - it creates additional mounting weirdness just to keep you safe in cases of bad upgrades

this lets you switch between snapshots during a boot

i dont like this way of solving the system integrity, couple of times i had an issue with zfs, and this type of mounting becoming a big pain in the ass during recovery

what's safer about mounting zroot/bhyve at /zroot/bhyve/ instead of /bhyve/?

in case to change that you have to prior using partition the zfs in a different way, (manually)

and then you have to control the snapshots by hand

if you new to zfs and bhyve - then is is better to say in the "recommended" zfs partitions

read about beadm

modern openzfs become very stable and you probably do not need to recover anything..

polyex, man bectl

zfs problem in 14.1 unattended bsdinstall. i got 2 nvme drives in mirror and i want a mirrored swap too. install finished and it boots fine. but there's no swap? termbin.com/t6j0 swapinfo returns nothing, top shows no swap. in /etc/fstab there's 2 lines. 1 for efiboot0 and 1 for /dev/mirror/swap.eli none swap sw 0 0

how do i get ISA x86-64-level on FreeBSD? en.wikipedia.org/wiki/X86-64#Microarchitecture_levels. on linux one car run /lib64/ld-linux-x86-64.so.2 --help

Title: x86-64 - Wikipedia

jesus christ you and this unattended install

just install it and start using it instead of fucking around

unless you're going to do the same install on 1000 other systems

relax i think i found a bug in 14.1 bsdinstall. or atleast a hole in the man page

does the non-interactive install create the swap properly?

and how does it do it

eli?

what im saying is from man bsdinstall, termbin.com/t6j0 should be enough to create a zfs mirror with a mirrored swap. it does everything except making a swap

does mdconfig -l show anything?

no

rtprio: euhm. one (1!) system is an appropriate minimal number of systems to install with automation, because it gives documentation and reproducibility.

... disagree. you use config management for that, not automated install

polyex: so yeah, i suspect it gets confused when you do both encrypted and mirrored

you could test this theory trying only one and then the other

rtprio k commented out the swap encryption line. reinstalled. on startup, says swapon: /dev/mirror/swap: no such file or dir

ok, then looks like you found a bug

can anyone who knows more than me confirm it's a bug pls?

how good are you at reading shell scripts

not

Is there a reason that D35694 was never merged?

It just seems lost in the aether

Title: ⚙ D35694 Add ts(1) command

polyex: wonder if there is possibility its get mounsted later than something try access it ?

try what drathir_tor?

it doesn't take 15 min to start swap up heh

polyex: yes true, but more thought about hierarchical steps of encryption script maybe its happens before swap location get mounted... obviously its only just idea...

anyone have unattended bsdinstall end at rm: /mnt/tmp/installscript: no such file or dir?

i commented out any of the swap env vars and it seems a swap partition was made for me

how do i check if it's a mirror swap?

polyex: normally df -h or lsblk -fs but need admit them from linux not sure if present under freebsd...

didn't show me if swap is mirrored

hi, I want to set up a small "homelab". I have been looking for potential servers and found some recommendations for Dell 320-420-620-720 , or the subsequent generation Dell 330-430-630-730.

I would like to know if someone has practical experience on FreeBSD support for this one or can recommend alternatives

these ones*

polyex: fast search " zpool status " maybe from docs.freebsd.org/en/books/handbook/zfs

Title: Chapter 22. The Z File System (ZFS) | FreeBSD Documentation Portal

that shows the main mirror but nothing on the swap

polyex, why dont you partiton disk as you want and then extract bsd files there

it is very simple to do

because i'm bug hunting in bsdinstall

ah ok then

nerozero do you know how to check if my swap is mirrored or not?

if your swap is placed on a mirror device - it is there

if you have 2 swaps on 2 different drives - it is not

what command output would i look at?

cat /etc/fstab ?

swapinfo -h

or that one

looks like it's not mirrored. the swaps are 2G in size, with 4G total

so can't be a mirror

but this is depending on how you mounting your swap

since new ppl are here now:

zfs problem in 14.1 unattended bsdinstall. i got 2 nvme drives in mirror and i want a mirrored swap too. install finished and it boots fine. but there's no swap? termbin.com/t6j0 swapinfo returns nothing, top shows no swap. in /etc/fstab there's 2 lines. 1 for efiboot0 and 1 for /dev/mirror/swap.eli none swap sw 0 0

you can do that by using gmirror against yor drives then put the swap on the gm device

but why ?

look

you do that wrong

how am i doing it wrong? i just followed man bsdinstall

why you need your swap data in a mirror ?

first

so i don't lose swap with a disk failure and lead to a crash

second swap partitions is separate of zfs once

what?

then make a volume partition on zfs and add is as a swap

how about you just let me find bugs in bsdinstall and get them fixed? why the pushiness

zfs create -sV2T -o volmode=dev zyourpool/zyourvolume

ok i forget that you are looking for a bugs .,.. sorry

I'm not using bsd installer for ages .... (>15y or so ... its too restrictive for me)

my turn to ask question. what is the best way to check if drive is mounted or not ( in scripting )

also better approach to find mount_dir is used already by parsing `mount` output

polyex: but isnt You loose ram anyway and swap is incomplete on crash ? to have full dump the system need to migrate all ram into swap file/device to being able resume from it... ?

sorry i dont understand what you say

polyex: like in meaning when system crash before Your all ram is stored into swap file/device You will be unable to resume the system bc of incomplete swap data... at least if im not wrong...

dunno bro i just want mirrored swap

polyex: oh i see...

nerozero: not sure if best (probably not), but grep "df -h" for mountpoint handle or "lsblk -fs" for partition uuid could works i guess ?

meena gone?

what's better, swap on zfs, or no swap?

polyex: no swap. Swap on ZFS can lead to kernel problems.

vortexx ok then very best option is swap the way bsdinstall sets it up, where it's separate from zfs?

polyex: if you really need mirrored swap, you do create a raid 1 device and tell the swap config that that is the partition to swap to

you don't involve ZFS in this

ya

that's how bsdinstall sets it up right? parallel to the zfs config

not mirrored, i gave up on that

never used bsdinstall

but docs.freebsd.org/en/books/handbook/geom/#geom-mirror-two-new-disks <-- how to set up raid1 with two partitions, shouldn't be hard to add to bsdinstall

Title: Chapter 21. GEOM: Modular Disk Transformation Framework | FreeBSD Documentation Portal

polyex: get swap even if unused always safe approach... You at least notice slowdown and start investigate when swap kick in... (assume no monitoring present to mail scream first)

yeah processes will get randomly killed if no swap and you run out of ram

vortexx: Not fully randomly, and see protect(1)

 

ok, i wanted to buy a DVD of freebsd 14, and a few physical copies of the freebsd mastery books (ensuring the author gets paid), but i want to do that in EU

where could i do that?

Demosthenex: freebsd.org/where/#distribution

Title: Get FreeBSD | The FreeBSD Project

Demosthenex, see docs.freebsd.org/en/books/handbook/mirrors/#mirrors-disc for getting installation media.

Title: Appendix A. Obtaining FreeBSD | FreeBSD Documentation Portal

yeah, i found a french vendor of the discs, i'll order one there

but the cool books, physical prints seem to be a problem

There's a German one, too, there: hinner.de/linux/freebsd.html

Title: Hinner EDV: FreeBSD und Linux

I'm guessing that you're looking for EU-based vendor, not just shipping? If shipping is not a problem, you'll find the books on amazon.de

the german one says no eu sales on the site, weird

ugh, i have to give money to bezos?

Demosthenex: mwl.io/faq#buying

Title: Frequently Asked Questions – Michael W Lucas

regis: yeah, i was on that page.

tiltedwindmill, bookshop.org, etc were all US specific

really, i need a US credit card to order there.

looks like it may only be amazon in EU to buy a physical copy with local currency

*sigh* ideally he still gets paid if i buy it from that sleazy seller

donner.nl/freebsd-mastery/michael-w-lucas/9781642350005 HA Donner had it, screw you bezos.

Title: FreeBSD Mastery, Michael W Lucas Allan Jude | Donner

I am surprised tiledwindmill requires a US credit card. You might email him and suggest an alternative payment method allowing you to "tip his rats".

well, he'd have to do EU shipping too, which is a PITA. i do understand

i'm just trying to buy them from EU, and see that he's fairly paid too

and i just don't get the same value from ebooks

I agree that I don't get the same value out of ebooks. I am much better with the physical paper books.

Finally now I know how Daemon sounds like youtube.com/live/uW9iEWv_UIE

Title: Interview with Beastie, the BSD Daemon - YouTube

His Treat the Rats item though does not ship you anything. It's just a little extra cash to send to him directly.

Heh, I got a flashback from Blank Reg: youtu.be/dZ4qli_1C48?t=33

Title: Some Blank Reg moments from Max Headroom - WIlliam Morgan Sheppard - YouTube

DanDare, Excellent. Hahaha!

:)

Hi, what could be the reason for dhclient to say "No working leases in persistent database - sleeping." at startup, when there are valid ones? (it happens after an outage when the ISP's infrastructure is not fully up yet)

I'm revisiting a cascade of failures that happens in this scenario, that used to be fine in freebsd 12.0. I originally posted it in forums.freebsd.org/threads/slow-mod…ng-up-ignores-timeout-setting.91250

Title: slow modem results in dhclient 'no link, giving up', ignores timeout setting | The FreeBSD Forums

yay, i bought about 100eur of books today

I have found that 13.0 dhclient only waits 10 seconds for link (hardcoded). I have patched that. I have realized that I need to have dhclient 'timeout' at least 120s to wait for the ISP to prevent the abovementioned "No DHCPOFFERS received, Trying recorded lease. No working leases in persistent database" scenario. I still want to understand why it's doing this though (is it verifying the lease by

pinging a host?). Finally, I have identified an ancient mistake in the pf 'nat' rule where I learned to make the rule dynamic by adding parentheses, but I didn't put them into the 'from !($ext_if)' part. Freebsd 12 and 13 seem to have autocorrected this, but 14 takes it literally and rejects the entire rc.conf file if ext_if isn't ready.

ultramage: Hi, you could try if simply using dhcpcd instead of dhclient works for you if you have issues with dhclient. That worked for me many many times in the past with very different issues.

ultramage: If you receive no DHCP offer, it looks if there is some DHCP assignment that is locally cached, at least that's my understanding of what a recorded lease is in that context.

armin: yes it's supposed to use dhclient.leases - which it has, and there's a valid non-expired lease in there. Prior to freebsd 13 it had no problem binding to it. Now it gives up instead.

I have the source code open. There have been no obvious changes to dhclient.c, but I do see that instead of just doing stuff, it crafts a short 'script' and then executes it, and only performs the binding if the script succeeds. So now I need to study the internal scripting engine to understand this change in behavior.

ultramage: oh cool, interesting!

how is if_wg_load="YES" in freebsd 14 used? i currently use wireguard-tools on freebsd 13.

it works in the same way

well no it doesn't because i kept all the same config as 13 and wireguard didn't magically work

man wg, I guess?

ya i read that. it doesn't say where to put the config file

with 13 wireguard-tools, i put wg0.conf in /usr/local/etc/wireguard/

yeaaaa. driver src doesn't seem to either. I'm trying to locate the userland 'wg' component atm

wg showconfig might reveal something

you still need net/wireguard-tools package

so why bother enabling if_wg_load if it doesn't add anything?

I don't know why you were bothered

that's interesting, man has a 'wg(8)' page, implying it comes built in

that's not what that means

ya it's weird

other BSDs have native tools for management, ours lacks the tool yet

polyex: my first guess is that having a kernel driver reduces the syscall / buffer i/o overhead.

but no worries

wg-tools will load it

is wg-tools the same as wireguard-tools?

yes

pkg install net/wireguard-tools

ok so keep everything the same with wg for 13, just enable the kernel module in boot loader for more efficient operation?

no need

it will delay boot, that's only benefit/drawback

no need to if_wg_load="YES"?

polyex: yes, please dd this line

polyex: how do you like FreeBSD ? are you considering migration to another BSD or other variant of Unix ?

no i like freebsd good enough. you?

same here, I got used to this OS

I am upgrading my desktop from 13 to 14.1 and it happens to have /usr/src/ populated and the upgrade has been runnning doing installs on /usr/src for the last FOUR hours.

I am about ready to kill it, snapshot things back to the previous state, and then do something different. Maybe remove all of /usr/src and associated and then try the upgrade again. It is just crawling slowly through /usr/src/ and this is taking forever and a day to get through it.

Any wisdom hints from the group?

rwp: what is it hanging on?

If you press ^T a bunch, what wait channel is it usually in?

The prompt is just "Installing updates..." but in another terminal I am running ps auxdww and looking at what it is doing in a loop and it is running "install -S -o 0 -m 0755 [long hex hashish] ///usr/src/tools/test/stress2/misc/mkdir.sh" and just crawling through every file in /usr/src/

I am quite confused. Are you doing a source or freebsd-update update?

I don't care about freebsd-update updating /usr/src/ and but it seems insistent.

I would just blast /usr/src and try again, as you suggested

It's running "freebsd-update install" after having run "freebsd-update -r 14.1-RELEASE upgrade"

I don't recall it taking so long to update /usr/src in past upgrades. And the other systems I have upgraded don't have /usr/src populated.

I am going to do it. Interrupt it. The worst that happens is that I have to revert to a previous Boot Environment and I am well practiced with that recovery.

freebsd-update often does weird shit for me. The machines I use it on usually don't have src present.

I had forgotten that /usr/src/ was populated on my desktop.

Another weird panic for y'all. bugs.freebsd.org/bugzilla/show_bug.cgi?id=279899

Title: 279899 – pf_unlink_state mutex unlock page fault panic

I started this at 10:43 and it is now 14:30 so it has actually been running almost 6 hours now. It's just taking way too long.

amigan, That panic does not look like fun.

Yeah. I went through commit log trying to brain-bisect it and can't see anything obvious.

rwp: do you happen to run the os off of an usb2 thumbdrive? it reminds me of how portsnap destroyed the one I was using way back when

ultramage, Nope. This one is a mirrored pair of 3TB SATA spinning drives.

I've seen etcupdate do something similar. In fact, is etcupdate in the process tree?

Might be too late now, of course

I have already interrupted it. I am cleaning out files now. Surprised by there being 120k files in /usr/obj for example. Quite a few.

I am purging out /usr/src and /usr/ports and /usr/obj as this is not a source upgraded machine.

But being as this is giving me trouble I might take this as motivation to simply copy over a good system from one of my other machines that I have already upgraded.

166k files in /usr/ports

In 42876 directories.

yea no, after that incident I use a sparse git checkout. a pain to manage since portmaster fails hard on missing subdirs, but it's much easier on the filesystem

and my usr/src is a md mount of a 4gb .img file with a git checkout inside, to keep all that crap boxed in.

Okay. Have all of that cleaned up. Will a "freebsd-update install" be idempotent enough to restart successfully from here? I guess there is only one way to find out. freebsd-version -kru says 14.1-RELEASE 14.1-RELEASE 14.1-RELEASE-p1 so it feels like it should be mostly done but cat /etc/os-release still shows 13.3 so it hasn't been updated yet.

At least it's installing files in /etc /usr/bin and so on now. Seemingly at no faster than a couple per second though which still seems like it is incredibly slow for some reason.

OMG! It uses install -S and the -S forces an fsync() after every action. No wonder it is crawling along so slowly! It's being forced to run at slow disk speeds.

That's got to be new damage with 14 as I do not remember it running that slowly in 13.

installworld j1 from 13 to 14 only took a few minutes on a single sata drive. that package-based os update must be weird

I edited /usr/sbin/freebsd-update and hacked out the -S from "install -S" and am running it again and now it is moving along much faster.

This is not the pkg-base that everyone has been talking about. Not yet anyway.

It seems from experimental evidence that interrupting and restarting freebsd-update is idempotent and it picks up where it left off okay. That's good anyway.

This is running on a fairly fast ten year old desktop with 16GB of RAM and two Toshiba 3TB spinning SATA disks. It's not a slow system normally.

freebsd 14 unattended bsdinstall throws up a "rebooting" gui when it's done. 13 didn't do that. how can disable? because now every bsdinstall takes 10 seconds longer out of nowhere. not documented in bsdinstall man page

why does bsdinstall show mouse cursor during unattended install? should be hidden or off to the side lol

polyex maybe moused is enabled?

well ya prolly but still

i type a key and mouse hides. no reason unattended bsdinstall can't hide mouse cursor too

Hiding the mouse pointer is a terminal feature, not a bsdinstall feature. It would need bsdinstall to add code to talk to the terminal. And then there is hte problem of which type of terminal.

ok then just unload moused?

Right. I would disable moused.

ya

moused_nondefault_enable="NO"

Hmm... Even though I purged /usr/src the freebsd-update is putting it back now. I must not have gotten the source freebsd-update is using to reference and it still thinks it needs to be there.

/etc/freebsd-update.conf Components lists only world and kernel and does not list src.

when will the wifi adapter be supported on the PI Zero W1.1 ? wiki.freebsd.org/arm/Raspberry%20Pi

Title: arm/Raspberry Pi - FreeBSD Wiki

WEN�

WEN��

WEN��

WEN�

why does the file i set for /etc/COPYRIGHT get overwritten now?

psionic: when someone who cares enough works on it

be the change you want to see in the world :-)

Because software patents suck.

testbox with xorg / fluxbox setup usercontent.irccloud-cdn.com/file/l…nce_testenv_22_06_2024_09_18_10.png

kevans know why unattended bsdinstaller sits there for 10 seconds at the end before doing automatic reboot? that's different from 13 behavior

polyex: I hope you're relaying to a bugs mailing list all your complaints about bsdinstall

just dumping here isn't going to cover it

hi circuitbone

hello