is there any way to get any sort of status from freebsd-update? It's been sitting there saying installing updated (from an upgrade -R 14.0-RELEASE) for a while now, with no signs of progress. (h)top shows it's using like 0.2% CPU and doesn't seem to have many child processes, with the kernel using about 10% CPU

updates* -r*

are you upgrading from a directly previous release, 13.2; and did you run freebsd-update fetch and install commands before starting the upgrade to 14.0-RELEASE?

yep and yep

this is the post restart freebsd-update call

what does 'freebsd-version' say when you run it in a different terminal?

14.0-RELEASE-p4

that is the most recent version of freebsd on 14.0-RELEASE, iirc, so that is good

yeah, I just don't remember it taking this long after the reboot to complete the second freebsd-update

I just it's been a while since I did a major upgrade

understandable, it took a bit for me on my older laptop that runs most of my freebsd services; I'd say give it some more time. you can always run 'freebsd-update IDS' to confirm the system against a known good configuration after it completes to check

I would run "ps auxd | less" and inspect to see what is running. I would run top/htop and see if anything is spinning using cpu.

yeah I have htop running and not much is happening other than the kernel and a few services. I'm guessing there's a lot of calls to install happening which aren't showing up in htop since they're so quick

which would probably explain the kernel activity, since it's all file system changes

ok, running opensnoop shows a lot of callsa to install and gunzip, so looks like things are happening

how good is dtrace

is there a way to find out what zfs features are supported by the current kernel? I have a zpool that I exported from macOS, and I couldn't import it on 13.2 because it had features that weren't supported (I've seen comments online that 14.0 has them, but I don't know how to check)

@Axman6 "zfs version" will give you the version of zfs. My 14.0 box says 2.2.0.

You can probably do the same on the mac. Then, compare differences in version numbers.

is darwin running FreeBSD?

yes :)

Did you just do an install?

no; I did that the day 14 was released

oh okay

that was not so long ago

actually that was just an upgrade

my last fresh installation was maybe FreeBSD 11

you upgraded from 11 to 14?

big jump?

from 11 to 12 to 13 to 14

yeah

I see

but I don't always boot into it; recent years it's more that I boot into Slackware GNU/Linux for more hardware drivers *BSD doesn't have, but if I could learn to run Slackware in BHyVe I'd boot into FreeBSD maybe all the time

i did the same on my server (11 to 14 in the right order) but noticed it'd filled up the entire GBs of the server in some temporary directory not documented in installation instructions

does darwin use enlightenment on FreeBSD as his desktop?

no

i used KDE3 then XFCE

sometimes KDE4 & 5 but they're so damn slow

in the past or for short usage also TWM or something

use e

did you switch back to E from XFCE?

yes

it just looks cleaner, more stylish and more options to configure

the thing is, E doesn't handle multiple monitors as well as XFCE. In XFCE it's automatic but in E it was a major pain to get setup even partly right

just get one big monitor

i have a big one but I use three

dang, much head turning?

not really

my classic 1600x1200 (on KVM switch for current PC, classic P3/686, and servers) is above my 4K, and my drawing pad LCD is to the side but mostly used for occasional command-line stuff

so I have necessary uses for all these monitors

if E could return win3-style program management I'd consider trying it again

last time they said you used to be able to setup that manually but that that process was broken

yeah I forgot what that even is

manually would involve putting hundreds icons into program groups so not worth it

I just use a mac style dock

and tiling

well you could view a picture or video about win3. It was nicer than start menu or Apple equivalent

it's so easy to slip to a wrong menu or out of it entirely in that style

my .bashrc isnt being executed on login. However when I run bash it loads the file

program manager was windows containing program groups that all got moved to submenus later

so you can't slip out of it easily or go to the wrong one easily

additionally you can always see all programs of certain types you have available, reminding you if you want to do something. It was great!

XFCE and maybe some IBM GUI used to have this stlye also. MS allowed still using it in win95/98 but later removed it entirely. Companies always remove features with newer worse stuff they falsely claim is better then force everyone to use that

the program groups looked like a file manager in icon mode and the icons launched programs

well except had no menus or stuff to the sides

are you sure you have your shell set to bash, then mnathani ?

I do

try ~/.bash_profile

darwin: I'm running centos on bhyve, slackware shouldn't be too hard

i like ksh like on NetBSD UNIX

great

or symlink ~/.bash_profile to ~/.bashrc ?

well those files have different purposes

what is this bhyve?

is it pronounched "behave"?

yer a wizard AumShivaya

it worked

darwin, what about using e's "everything launcher"

it may be better to load one of those bash files in the other. They run in a certain order

bhyve ia pronounced like a bee hive the place were bees live

its a virtual machine

great mnathani :)

not like linux compat?

but more like virtualbox?

the order is: /etc/profile , ~/.profile , ~/.bash_profile , ~/.bash_login but I don't have a note on when ~/.bashrc runs

linux comp runs things under freeBSD, the program run but uses a sys call emulation later that translates linux sys calls, bhyve is more like virtual box or vmware

I guess, but I am not sure, that bash_login runs when you log in to a tty

in ~/.bash_profile at the end I source ~/.bashrc

but rc does indeed run for me when I open a terminal, say, terminology etc. in X

that is probably a good way to do it darwin

rather than a symlink

"bhyve" is pronounced "bee hive", and is the hypervisor of choice for FreeBSD hosts.

bee have

No, "hive". Like where bees live.

johnm: finally finished, using execsnoop was helpful for tracking that something was happening

becasue i basically have the worse luck in the world.. a machine that had 3 usb drives (setup to zfs) died.. my question is.. is it possible to "move" a zfs pool to another machine if you do not have backups?

i am currently looking at new machine.. to replace the dead one and now i have 3 drives setup with a zfs pool .. it is my synology dying all over again.. minus the linux filesystem

whats the difference between software found in the form of a package vs a port?

never mind looks like with a zpool import, that hsould allow me to do all this

mnathani, packages are precompiled, but you have to compile ports.

voy4g3r2: yes spool import is what you want to do and it should work fine, I've move rust between machines and just spool imported it. That's part of what make ZFS great NO hardware raid dependancy

crb: thank god, because i have a large disco biscuits, ryan adams, and everyone orchestra live setup

now time to see if this lenevo i have my eye on will work with freebsd

hmm, if i cherry-pick a commit from main to stable/14, then the commit is MFC'd, that shouldn't cause a merge conflict, right? because the MFC should be done via cherry-pick as well?

unixwitch: right, git should just drop your local commit when you rebase over stable/14

i cant get oss to pick up my subwoofer sound output. anyone know how to get that working?

i think it shoudl be under the surround sound channel but it doesn't show up in the sound options, only the two speakers work

kevans: thanks

spine-o-saurus: doesn't your sub have a crossover? i wouldn't expect a sub to be on its own channel

uhm, im using freebsd on my sdcard to rn opnsense, and i just noticed it isnt utilising the whole capacity of the sdcard.. how do i fix that?

pulled ports and run out of space heheheh

sams: use 'gpart resize' to resize the appropriate partition, then growfs to expand the filesystem (assuming you're using UFS). preferably take a copy of the card first just in case...

is that cli?

yes

ok great, im not worried about data there its just the opnsense router and have backed up the current config

oh you said opnsense, i don't know anything about that specifically, but i assume it offers a shell

yeah i only have web/shell access, cannot plug it into a monitor though so yeah headless

hmm... you might also try doing this with the /etc/rc.d/growfs script, which is intended to do this automatically specifically for the case of SD-card installations... but perhaps opnsense already uses that and it didn't work for some reason

otoh it's not hard to do manually and growfs script looks quite complicated, so perhaps don't do that

ok thanks

kid playing VR online atm will do it when she idle, if i get stuck i'll be back!

jwp, spine-o-saurus: depends what sort of sub it is... a surround sub would usually be on the dedicated lfe channel (assuming the hardware has one) but an audio sub nearly always uses a crossover which means the OS can't detect its presence...

ok that was painless

i had a quick window and did the growfs one and it worked

Avail 51G yay

i dont use the LFE port on the sub

it just goes straight to the mainboard sub plugin

if i install freebsd from source or install the system first then update from source and build a custom kernel would i be able to get sound working on this chromebook ? i've installed freebsd on it and had a working system for about a month already but sound never worked unless i used a dongle with headphones

jb1277976: the GENERIC kernel should include all available audio drivers, so i don't think building from source would make a difference

jb1277976: does it detect the built-in audio device at all? (cat /dev/sndstat)

also, is this amd64 or arm64?

unixwitch no I’ve taken FreeBSD off since I was upset. I only have hdmi/8 something I forgot. Nothing I’ve done in the past months have worked verbose boot has no other pcm0 devices I just thought building from source would give me some hope

amd64

there's a slight possibility that some weird audio device isn't included in GENERIC for some reason but it doesn't seem likely. hard to say more without more info, like a 'pciconf -lv'

Let me oull up a Freebsd fourm I posted

pull*

forums.freebsd.org/threads/trying-t…eadphones-working-on-a-laptop.91214 If your went to look over yet everything

Title: Solved - Trying to get headphones working on a laptop | The FreeBSD Forums

Actually unixwitch that was the wrontnthread it's buiried deep

I will install FreeBSD again and try

booting the installer may be enough to debug this if you don't want to wipe whatever OS is currently installed

i don't remember if the installer includes all the kernel modules though...

ok let me download the iso boot live cd and go from there brb

unixwitch it only has <intel kaby lake hdmi/dp 8ch> (play) default and that is pcm0 on normal machines like this old racer I would have more pcm0 devices

jb1277976: output from 'pciconf -lv' would be useful here, but if you're on the console i guess it's hard to paste the whole thing... do you see any other audio-related devices in the output though?

(especially if they're "none⊙p." which means no driver is attached)

those shoule have class=multimedia

s/shoule/should

yea let me post it

Did that post ?

one sec

i didn't see anything

did you paste it at bsd.to? you need to link the url here too, it won't post automatically

hdac0@pci0:0:31:3 but its comet lake PCH-LP cAVS

Normally on Linux I need sof-firmware

jb1277976: is hdac0@pci0:0:31:3 different from the pcm0 device? i think dmesg should show which pcm is attached to which hdac...

(your issue is that pcm0 is the HDMI audio output and there's no pcm1, right?)

Nope no pcm1

Let me install FreeBSD will take like 5 mins I have nothing important on this laptop. Need to setup WiFi also brb

i guess "sof" just needs some new sort of driver that no one has ported yet...

Yep

that was my only conclusion I spent a week trying to get it to work lol

I’ll give it another shot I got nothing to lose 😁

Looks like GPU-specific driver would be needed for sound-over-HDMI: forums.freebsd.org/threads/amd-gpu-sound-via-hdmi.74863 ; forums.freebsd.org/threads/sound-over-hdmi.36130

Title: Solved - AMD GPU sound via HDMI | The FreeBSD Forums

parv: i think the issue here is with the non-HDMI output

unixwitch, Oh right!🤦🏽‍♂️

so i've used zsh since forever, but i'm trying fish... it has no completions for poudriere though :(

That could be "Just a matter of programming" surely

yes, i had about 47 other things to fix first though :-)

(got my nfs bug fixed, ported sublime music, now i need to port mediaelch... i might have a look after that if i'm still using fish)

unixwitch, What does "mediaelch" do?

parv: it creates XML ".nfo" metadata files for Kodi/Jellyfin media servers

unixwitch, Thank you

hopefully won't be hard to port since it's just a Qt app and alreadys runs on linux and macos

actually i should also look at why Jellyfin is broken on freebsd, i had to convert my media server to linux temporarily... and also update/fix the navidrome port...

oh, and fix that kadmind bug too

think the next item on my todo list might be 'find a todo list manager for X'

hmm, if a port supports both Qt5 and Qt6, should that be done via options or flavours?

unixwitch: "it depends" is the correct answer here (imo), what is your use case?

TommyC: no particular use-case, it's just a normal desktop application. i checked an existing port (audio/pavucontrol-qt) and it doesn't seem to specify a Qt version...

oh, it does: qt:5

unixwitch: it's possible that that's the only version of Qt it can be applied to (either the developers of the original project use something specific to that version of Qt or the maintainer hasn't had a chance to try and build against Qt6 for the project yet)

feels like only yesterday everything was on qt4 :(

am I getting old? no...it is the developers who are moving quickly

perhaps i'll just use Qt 6 for now (as it's the latest version) and see if anyone objects when i submit it

unixwitch: best of luck :)

ports-mgmt/submodules2tuple is handy

ok i'm treying to boot the intaler on a thinkpad x250 but i get error 19

it fails to mount the disk image]

i tried this forums.freebsd.org/threads/installi…ting-dvd-failed-with-error-19.36579

Title: Solved - Installing 9.0-RELEASE: Mounting DVD failed with error 19 | The FreeBSD Forums

didn't help

i need a faster CPU for building these huge GUI applications... i keep thinking i should rewrite all the GUI stuff i use in C/Motif to reduce bloat

I'm a total noob with build systems. I have a little program using the header <netinet/sctp.h>. On Linux you have to build that with $(CC) -lsctp, on Freebsd you don't need / must not have this flag. Why is that?

jemius: it's just an implementation difference. this has historically been fairly commonly for sockets... for example, Solaris used to require -lsocket to use the sockets API (rather than the XTI API), but BSD/Linux didn't. i guess Linux decided SCTP isn't important enough to include in libc.

other examples of this include the historical -lm (libm, for mathematical operations) which is often included in libc nowadays so it's not required, but oftena dummy libm is shipped to avoid breaking software that uses -lm

jemius: if you're using something like cmake or autoconf, you should just add a build-time check for whether libsctp exists, and link it if it does, otherwise don't

wait, sctp should always be a separate lib, shouldn't it? what is in libc and what is not is standardized

although, look at netinet/sctp.h, i don't see anything that would require a library... i wonder what's in linux's libsctp?

jemius: no, that's not standardised. POSIX defines things that *must* be in libc, but it doesn't say you can't include other things.

all it says if that if you're using a POSIX conformance macro (like _XOPEN_SOURCE) then certain things should be excluded from the headers on build, but they can still exist in libc

if you couldn't include non-standard things in libc, then functions like sysctl(3) couldn't be included, but they are

It seems to me that the sctp-helper-library is somehow separate in Linux: "client.c:(.text+0x4cd): undefined reference to `sctp_sendv'"

it seems like you're using a Linux-specific library called LKSCTP: github.com/sctp/lksctp-tools - i don't see it in ports... you might need to rewrite to use the APIs that FreeBSD providers, or port the library

Title: GitHub - sctp/lksctp-tools: The Linux SCTP helper library

FreeBSD does provide the functions sctp_sendv() and partners. They're specified by the RFC for the Sockets API, so it can be expected that they are provided

FreeBSD seems to provide them differently, though

hm does it? there aren't any manpages

Well, I can perfectly use my Linux program on FreeBSD if I build without -lsctp

So the function is there. Somewhere.

*oh* i see, i thought the linker error you showed was on freebsd, not on linux

yes, on freebsd it's in libc: lib/libc/net/sctp_sys_calls.c

Title: sctp_sys_calls.c « net « libc « lib - src - FreeBSD source tree

It should still have a manpage, though

yes, it should

it's not mentioned in sctp(4) either

there is a manpage for sctp_send(3), so perhaps sctp_sendv just needs to be added there

similar to how write() and writev() are both documented in write(2) since they're basically the same thing

Title: RFC 6458 - Sockets API Extensions for the Stream Control Transmission Protocol (SCTP)

Ah, anyways. I better start using Meson or sth

is "USE_GL= gl opengl" redundant? make stage-qa told me i need both but it seems like it should be one or the other

ok i got that working

14.1 in april?

polyex: sounds realistic

good stuff

anything bad about a pid file in /var/run having -rw------- perms?

owner is server's user:group

i saw other pid files are world and group readable

group being who?

is there not a periodic script to run zpool trim, or can i just not find it?

chrony is root:chronyd with -rw-r--r--

polyex: PIDs are generally not secret unless you use see_other_uids=0 and are *really* paranoid about it

ok ty

oh, i guess you asked about the version that's 0600, that should be fine too. although it's a bit weird

pid files are generally only used by root when starting/stopping services so they don't need to be world readable

xrdp.pid is 600

root:wheel

patch to add daily_trim_zfs_enable to periodic.conf; does this seem reasonable? le-fay.org/tmp/zfstrim.diff

periodic is like a built in cron system?

Yes.

does periodic work with cron or they separate?

cron calls periodic. look at /etc/crontab

Is someone with amdgpu experience here and willing to help me?

I'm unable to get the thing to work... I'm stuck at 768p resolution.

isley cool!!

I'm trying to use 7800x3d integrated gpu.

unixwitch: why not just set autotrim?

rtprio: some SSDs dislike that sort of trim-behind behaviour and perform badly

oh

(this is mentioned in zpoolprops(7) under the description of autotrim)

went looking for a better source and apparently TrueNAS don't use autotrim because "for many SSDs it actually makes ZFS performance worse", which i guess is based on real-life feedback: truenas.com/community/threads/auto-trim-for-nvme-pool.99162

Title: Auto Trim for NVMe Pool | TrueNAS Community

does it really need done daily?

maybe autotrim fights some caching?

(i don't have many SSDs)

i don't know, it seems like something that could perhaps be configurable, except i don't think periodic(8) supports that. if you're doing a lot of writing to the device (oltp or something) you might want it trimmed daily...

unixwitch: re bugs.freebsd.org/bugzilla/show_bug.cgi?id=275965 thanks for the bug report, but patches are way easier to review on Github or Phabricator

Title: 275965 – periodic: add a daily zfs trim

meena: i thought phabricator was only for developers. is there a guide on how to submit patches like this somewhere?

unixwitch: pick a target (Github or Phabricator, whichever you find easier to work with) and follow procedures there.

meena: alright - i suppose i should move #275912 and #275915 at the same time? (those also have patches)

wiki.freebsd.org/Phabricator / wiki.freebsd.org/GitHub / docs.freebsd.org/en/articles/contributing

Title: Phabricator - FreeBSD Wiki

unixwitch: I haven't looked at those. but, yeah, also, you can just use the bugs URL field to point at phab / Github

and the commit messages should point at the bugs, see github.com/freebsd/freebsd-src/blob…ls/git/hooks/prepare-commit-msg#L43

Title: freebsd-src/tools/tools/git/hooks/prepare-commit-msg at main · freebsd/freebsd-src · GitHub

unixwitch: I hope that's not too much (red tape)

meena: no that's fine :-) although the phabricator docs end with "use git push as usual", so i might try with github. is this only for src/ or should i be submitting ports like this too?

no ports have a whole sophisticated framework built around bugzilla

src, docs, etc do not

many of ports committers do complex stuff in Phab, but "update port" is usually handled via Ports

Github does not scale well to ports

do you guys have console none unknown off secure or insecure in /etc/ttys?

not sure which to go with

polyex: insecure is only useful if a) you have dialup modem lines, or b) you have a configuration where someone can get access to the console but not to the bootloader, which is fairly unusual nowadays (because usually those are on the same device)

well if someone can get access to the console i guess that means they are next to my physical computer?

unixwitch ^

what situation is there where someone has access to the console but not bootloader?

what i wanna avoid is someone being able to bypass security by booting into single user mode and not needing to enter a pw

polyex: either they have physical access or, if it's a VM, they have access to the VM console or something like that. as to when you could have console access but not bootloader access, i can't think of any situation that would usually occur in. perhaps if you set a password in loader.conf...

if someone can boot single user, it usually means they have physical access and could just remove your disks to read the data. or boot from a USB stick. or any number of other ways

well if i have full disk encryption in the zfs, then they can't just take the disks out

i still don't totally understand this ttys feature. i feel like it's a way to bypass the user account security

if the disks are encrypted, you probably don't care that much about someone being able to boot single user, since they presumably can't actually boot without the encryption key

anything bad if i just set them all to insecure? all that means is sometimes i'll have to enter password more often right?

DNSSEC errors for freebsd see dnssec-analyzer.verisignlabs.com/geo.freebsd.org

Title: DNSSEC Analyzer - geo.freebsd.org

"No DS records found for geo.freebsd.org in the freebsd.org zone"

"No DNSKEY records found" "No NSEC records in response" "No RRSIGs found"

whoa

we're so fucked

meena: before i submit a PR to freebsd-src... does this look right? github.com/freebsd/freebsd-src/comp…xiwinter:freebsd-src:daily-zfs-trim

Title: Comparing freebsd:main...lexiwinter:daily-zfs-trim · freebsd/freebsd-src · GitHub

just needs a fixup. can't run package updates on my system.

unixwitch: periodic.conf.5 needs a .Dd bump, but otherwise this looks perfect

unixwitch: also, no harm running shellcheck over that script

oh, haskell, this will take a while to install :-)

unixwitch: pkg install, or shellcheck.net

Title: ShellCheck – shell script analysis tool

this Script is short enough to just do it on the Web

(Also, no need to follow blindly everything it says)

meena: i cleaned up that ugly $newline thing, which comes from 800.scrub-zfs, so i suppose i should submit another PR to fix it there too. or maybe not, if it works...

unixwitch: it's ugly, but it works is a great slogan. I personally don't subscribe to it, because if you have trouble understanding something, chances are someone else does too

So, yeah, just add a second commit to that PR to clean up the script you copied from

n.b.: I'm tidying and making dinner. my capacity for a proper review is slightly diminished, hence such very general advice

I expect better after kiddo is asleep or at least fed

does anyone know of a good site that could help someone understand what can work with a freebsd install, hardware wise?

unixwitch: cuz no ones put it there yet

meena: it's such an obvious thing to be there that i'm wondering if there's a specific reason it wasn't added :-)

maybe everyone uses buildkernel/installkernel nowadays

I dunno, i mostly just build specific modules I'm working on

hmm, speaking of kernel modules, maybe i should try to fix the alc(4) bug where it doesn't work with MSI-X enabled (PR#230807). i'm not really sure where to start there though, it might just be a hardware bug...

actually i see there's a suggested patch on the PR

unixwitch i've come to realize that most desktop computers don't have sound working out of the box right ? they need some type of headphones/speakers i will just use my headset or by a dongle that plugs into my usb audio jack that i have already and i should be good

jb1277976: i wouldn't say that's true for "most" desktop computers, at least not in my experience. FreeBSD detected both my built-in HDA and my USB audio device without issues (although i did a fair amount of extra configuration, that was just by personal preference)

i haven't bought a new computer in a while though, this motherboard is about 5 years old (AMD X470)... maybe they're all "sof" now?

check out bsd-hardware.info/?id=pci:8086-02c8…et+Lake+PCH-LP+cAVS&dev_ident=4414f looks like the sound is detected in Current ?

Title: Intel Comet Lake PCH-LP cAVS

or is that old news ?

jb1277976: if it needs firmware, just detecting the PCI id might not be enough. but you could try booting a current snapshot and see: download.freebsd.org/snapshots/ISO-IMAGES/15.0

Title: Index of /snapshots/ISO-IMAGES/15.0/

ok brb

that page says it's supported by "12.2 and newer" though so if it didn't work in 14.0, i doubt it'll work in 15. but it doesn't hurt to try

ok and current isn't like a release right i can't use freebsd-update and stuff ?

jb1277976: you might also try posting on one of the lists like freebsd-questions or freebsd-desktop if that doesn't help

ok

there are no freebsd-update binaries for -current as far as i'm aware. you'd need to build from source

also current can be a pretty unstable... although the last time i tried using it on a desktop was during the 5-CURRENT kernel locking changes which broke *everything*, maybe it's better now :-)

dam lol

ok rebooting

i moved a zfs pool to a new machine and when i try to do a zpool import -f, i get this message openzfs.github.io/openzfs-docs/msg/ZFS-8000-EY

Title: Message ID: ZFS-8000-EY — OpenZFS documentation

the previous machine it was attached too died and the zfs did not dismount properly and when i do the zpool import -f, it also tells me a hard drive in the pool is corrupt..

Title: dpaste/SfDq (Plain Text)

so it seems i can not import this zfs.. but without importing it i can not fix the drive.. is there a way to "fix" the fault without impacting the zfs pool

voy4g3r2: Are there bits missing? That's only talking about potentially two disks, and it thinks there's data corrupted, which could conceivably mean there are more disks.

i did notice that myself.. there are 3 drives.. not sure why only 2 are showing

freaking usb!!

voy4g3r2: Ah. USB is suboptimal. Make sure they're all there, and specify them explicitly in your import line. man zpool-import

haha that is what it was

mason: agree.

all of the drives are now online but it is still giving me that another system thing.. even after zpool import -f

oh yeah.. never mind.. gotta name the pool name

i need someone that will tell me how to tune my tcp connectivity on freebsd ;-;

voy4g3r2: And if things are marginal, think about scrubbing and/or backing up data

mason: i am looking at another hard drive to move to another location on my property

i first had to see i can get this zfs pool oeprational on a pi

voy4g3r2: Also, I'm not sure how far I'd want to trust raidz1. I'd be more inclined to do a three-way mirror.

yeah i am looking at this, i got 3 2tb drives

right now my consumption si: bsd.to/JhNT

Title: dpaste/JhNT (Plain Text)

i gotta read up on this more.. either way, i take it the data would have to be moved "off" this array

yes

what hosts does freebsd connect to to run freebsd-update and pkg commands? i thought only pkg.freebsd.org and update.freebsd.org but i allow those in my fw but the commands still fail

polyex: i forgot to answer your last question about insecure: no, there's no disadvantage to enabling it, other than having to type the password sometimes. it can be an issue if you have a server located somewhere and need remote hands to work on it (and don't want to give them the password).

yay, bhyve can boot openbsd again

voy4g3r2: did you try import -f

also, i see three drives, third is /DISK-AAAABBBBCCCC0003

ty

i gotta read up on this more.. either way, i take it the data would have to be moved "off" this array

rtprio: yeah i got it to work

yeah, if a drive is bad that would be the safe move

i git pull the freebsd repo, but then when i build an installer image and boot it, freebsd-update fetch then gives me p9 updates. but why don't i have those already since the git repo is up to date?

i'm on branch releng/13.2

unixwitch: freshbsd.org/freebsd/src/commit/30c…6cacfd626552c647c9327dd510e7f421d97 it's merged!

Title: FreeBSD / src / 30ce26c / nfsstat: update option strings in docs - FreshBSD

is there no way to bypass freebsd-update to get the patch updates?

why does git pull not get them?

what repo are you 'git pull'ing

it does, you're probably just on the wrong branch.

freebsd-src

on github

releng/13.2 is my branch

it gets the updates but without an `make installworld` you're not actually updating anything

oh whoa

releng/13.2 definitely has the patch updates.

is that all i need to type or is there more?

yikes, you should just use freebsd-update

no

rtprio so i git pull then type make installworld then i can build the installer iso?

polyex: read docs.freebsd.org/en/books/handbook/cutting-edge/#makeworld

Title: Chapter 26. Updating and Upgrading FreeBSD | FreeBSD Documentation Portal

you're asking questions without enough context; what are you trying to do?

update your 13.2 via source?

Will ipfw in-kernel NAT support syntax like?: bpa.st/QCVQ

Title: View paste QCVQ

I'm specifically referring to the port ranges.

It's supposed to be a (mostly) drop-in replacement for natd, which supported them - but I haven't found any examples with the in-kernel NAT.

why does the raidz setup need to be so difficult.. i got 3 drives

rtprio: what would you recommend for storage with 3 drives.. for a setup.. right now i have 3 2tb drives, thinking of upgrading them to 4tb..

i had a 3 drive raidz1 for a long time

and even ran 2 drives for 8 or 9 months (the RMA process took forever)

That would be fast. . . but you'd only get one drive's worth of space.

yeah, i would like to get 2 drives and one as "backup"

but i do not think this is possible.. the wife said, get it working and have a redundnacy in our pole barn.. i am sick of hearing you "fixing" your hard drive bays

i did not think another computer to do on me.

voy4g3r2: do you have the 3 drives for the mirror from different vendors/factories/batches? this reduces the risk, compared to two or three failures in batch where the drives fail in almost same time because of same defect

nedko: all bought different years

voy4g3r2, that sounds like raid1 plus a hotspare. . . still only one drive's worth of storage.

yup :)

You dould do raid5, which would give you two drive's worth of storage, and you'd be able to take a single drive failure, but. . .

Read speeds are fast, but write speeds are slow.

a 3x mirror setup can use it for a 3x read speedup

i basically want to stream music, movies, and some basic git stuff

Plus, if you're in a failure state, everything gets really slow. .

so i am more in the i want to have my catalogue available without lossing hundreds of gigs of my cds

Plus, with 2TB drives. . . if a drive fails, you're statistically more likely to take a second drive failure and lose all data, than you are to complete a rebuild with replacement drive.

yeah i have a large storage need but not htat big to go all out

Hey unixwitch been busy. Didn’t work on current. Gonna go back to a regular setup and post to the desktop list and the questions list. Thanks for the help

ls

meena is that page relevant if i want to build freebsd installers and NOT modify the freebsd of the current machine?

hello, anyone encountered "ERROR: Build failed for target 'images' in configuration 'bsd-x86_64-server-release'" while building openjdk17 on 14.0-RELEASE - amd64 ?

polyex: `man 7 release`

iirc- cd /usr/src; make cdrom

(Prolly less work to download an ISO.)

did git pull now doing sudo make buildworld buildkernel in the repo's dir. then i'll build an installer and see if it reflects the latest patchlevel

for some reason i thought the installer build was all that i needed but i guess i gotta build the repo's code, then build installer from THAT

i still don't follow why you're trying to do this

so i can make installer isos that have the latest patch level applied already

is that bad?

polyex: nope, sounds sensible

ty meena

(i wish i knew why we don't do that)

my idea is if i build my own installers with latest patchlevel, i can then deploy those so it's only 1 DL from the freebsd servers, not hundreds. so save the project money

anyone seen this one ?

```

# freebsd-update fetch

Looking up update.FreeBSD.org mirrors... 3 mirrors found.

Fetching metadata signature for 13.2-RELEASE from update1.freebsd.org... done.

Fetching metadata index... done.

Inspecting system... done.

Preparing to download files... done.

No updates needed to update system to 13.2-RELEASE-p9.

root@osaka:~ # freebsd-update install

No updates are available to install.

Run '/usr/sbin/freebsd-update fetch' first.

Now

```

What in the holy frack????

please use a pastebin of some sort for longer outputs

never mind.

stupid thing wont upgrade past -p9.

isn't p9 the highest?

yeah, there is no -p10

yeah that is the highest

audit -F says there are a few security issues between p8 and p9. (server was running 12.x so I guess I should just thank my lucky stars except that we have 2 days to safely upgrade 10+ servers before everyone starts melting down)

so from 12.x we freebsd-upgraded -r 13.2-RELEASE and its stuck at -p8

according to one of our other admins "all the service boxes are stopping at p8 too"

audit -F just bitches about python libraries that would require someone to be on the box so I suppose we can wait.

our first attempts to run 14 were a trainwreck.

so I was wondering if anyone else had encountered this.

feurig: what's leading you to believe you're stuck at -p8?

it finishes the upgrade (13.2-release) and its at p8 and wont pass an audit.

so we run the above and it complains about the sshd config and then goes into the loop above. (sorry about the 10 lines)

sounds more like an audit bug than an update bug

The other admins upgrading our core service boxes found the same.

what does `freebsd-version -uk` say?

-p9 only touched ssh it looks like, so you'll probably find userland at -p9 and kernel at -p8

# freebsd-version -uk

13.2-RELEASE-p8

13.2-RELEASE-p9

yeah, that checks out

so you're not stuck at -p8 at all, everything is where it should be

except that unlike all of the rest of our p9s pkg audit gives us 4 problem(s) in 3 installed package(s) found.

kevans: thats really unintuitive.

I will have the other admins check.

packages aren't generally part of the base system

so why did the kernel not get the messege?

pkg audit might have some information about the base system, but its primary purpose are for things you've installed from ports

if I dont have to reboot this would not be an organizational nightmare

i don't really understand what you're asking anymore, but updating ports doesn't usually require a reboot, no

kevans: the two releases.

we have a short window to upgrade our infrastructure machines to the latest patch level, and to upgrade the packages to where they pass an audit.

And the machines dont appear to be getting to the latest patch level and wont pass an audit.

upgrading both on our internal servers worked fine. Just the machines that count aren't.

If thats normal. Ok as I said nevermind.

how large is the register space on a PCI device? 0x0000-0x1FFF? or does it vary?

feurig: your machines are at the latest patch level. your packages are failing audit. these are distinct facts

(i'm trying to work out what Linux is writing to when it enables MSI-X interrupts on this device by writing to 0x2000... is that device-specific or does the freebsd pci driver do that for you?)

the former we established above when you checked freebsd-version

Is there a way I can coerce freebsd-update to replace a locally-modified file?

so i clone the freebsd-src git repo, run sudo make buidlworld buildkernel, then i build an installer and use it, boot new vm, uname -a says 13.2 p9, BUT then freebsd-update fetch finds 52 files that it says will be updated as part of updating to p9!!! wtf?

is there any reason that you dont want to upgrade to 14 polyex

?

kevans: Yes. And you are right and I appologize for being cranky.

ya

feurig: the unfortunate reality is that vuxml can probably be updated before the package builders finish a build of the fixed version, so the packages may just not be available yet at all

wow

so i cloned freebsd-src, built, built installer, installed, uname -a said 13.2 p9, freebsd-update fetch got 52 files needed to update to p9 (???) so i went ahead with fetch, then install, then reboot, and uname -a now says p8!!!!

polyex: oooo, that sounds like a freebsd-update bug

ya really think so

did you keep a list of what 52 files it thought needed updates?

no but i could get it again and throw it up if you want it?

hmm... probably better to just try it and debug from there instead of guessing from the list

Hm, shame there isn't an include directive to accomodate this sort of thing: bpa.st/RTKA

Title: View paste RTKA

what?

I ended up just fetching it from github (no /usr/src on this particular box) and that was what was different.

aah og my god it works! alc(4) now functional with MSI-X... or at least sort of... the rx is working now but the tx seems a bit wonky

unixwitch: NICE!

for some reason ICMP pings are delayed by exactly 1000ms, i think you need to do something else to set up the tx queue interrupt but when i try that the rx stops working...

kevans so anything we can do about this freebsd-update bug?

polyex: the kernel and userland can be at different patch revisions and be wholly up to date

but i had the latest repo source and built and installed from that. so why would there be any updates at all?

That's a good point.

I'm a little curious about your build from source saying -p9, also. Was it cloned from releng/13.2 or from stable/13?

bbiab, but will check in later

releng/13.2