i'm getting errors about zfs when i boot up. its before the acer screen comes on something about the zpool. yet when i boot into freebsd everything is fine. zpool list and zpool status show everything like it should

jb1277976: if you don't have time to read the error, trying taking a video on a mobile phone (if you have one)

ok let me do it now.. where do i upload it ?

well, i thought you could just watch it on your phone and type the message out :-)

oh yea that makes sense. i will brb

although in general, the first thing i'd do with loader/zfs issues is to update to the latest loader, if you didn't already (assuming you're on EFI)

Hey, anyone using freebsd collectd logparser plugin?

I need help understanding the pattern regex for collectd logparser plugin. Does anyone can point me to a config sample that reads journalctl messages (or any kind of unit / log)

I have found this example in the man page, using var/log/syslog, but cannot make it work. Any clue/doc indication?

i don't know about collectd specifically, but freebsd doesn't have journald (or systemd) so you won't get anywhere with that. /var/log/syslog doesn't exist by default, but you could create it in /etc/syslog.conf

(look at the commented example for 'all.log' - that's basically what /var/log/syslog.log usually is)

unixwitch: BTW, I dont care about what log file it will read. I am trying to make it work - even if it is a ascii file to parse.

well, try /var/log/messages to start since that exists by default

it is in the usual syslog format which should be supported by any log collector

unixwitch: I did. It captures the portions of the logs I want and form a collectd UDP packet and send to influx-db server. I capture the packet using wireshark and the collectd-message is somehow 'malformed'

zfs: i/o error - all blocks cpioes unavaialble

zfs: can't read MOS of pool zroot

BIOS 634kB/185143kb avaiable memory

unixwitch: saddly, the collectd documentation about log_parser_function is very limited. And I dont know where else to look/search.

ah, i have no idea about that part of it then, sorry

copies*

jb1277976: that's weird. have you run a scrub on the pool just in case? it's weird it produced that error but still booted fine though

let me do it right now

unixwitch its scrubbing right now

 

oops

so i am having a problem upgrading a jail from 13.2-RELEASE to 13.2-RELEASE-p9. When I run the upgrade from the jailhost (running -p9) it runs and then the jail craters.

ld-elf.so.1: /lib/libc.so.7: Unsupported relocation type 37 in non-PLT relocations

if i run the fetch install from within the jail it fails with

Installing updates...chflags: ///lib/libc.so.7: Operation not permitted

chflags: ///sbin/init: Operation not permitted

and 3 other chflags errors

any thoughts as to what is going on?

there's a sysctl controlling chflags in jail

man jail, /allow.chflags

hmm, sctp_recvmsg(3) says the 'from' argument is restrict, but it's not in the actual prototype. i wonder if that's a bug in the manpage or the header

yuripv: set allow.chflags=1 in /etc/jail.conf and restarted the jail. Do I need to reboot the host?

also thanks.

I will put my head in the man page next year. for now I am gonna leave it at 13.2.

happy new year all.

kevans: so it is possible i am downloading images from the current directory on freebsd.org that are NOT aligned to checksums?

are the checksums not the same as the files you downloaded

?

i was NOT checking before i did the install

but now the installs of CURRENT are blowing up

i am trying ta virtualbox image now..

i can't say i've checked maybe but once or twice in the 20 years i've been using freebsd

it's possible it could be broken; verify the checksum. or download the previous week's image

did that.. i did 23DEC2023 and 16DEC2023 same result :)O

blow up how?

i install the whole thing, says checksum are not correct.. restarts install

is that from the image or from the network

both install local to my machine and over the network.. same result

but looks like the virtualbox image works.. i am logged in

but i am now going to be stubborn and try to do an install WITHOUT hte virtualbox image

well, which exact image did you download; i can try it now

FreeBSD-15.0-CURRENT-amd64-20231216-ca39f23347e1-266973-bootonly.iso

FreeBSD-15.0-CURRENT-amd64-20231223-dac33a65b965-267058-bootonly.iso

FreeBSD-15.0-CURRENT-amd64-20231228-fb03f7f8e30d-267242-disc1.iso

Title: Index of /snapshots/amd64/amd64/ISO-IMAGES/15.0/

that is where i got them from

hrm, hard to install with this broken down arrow key

i got it

okay

Did anyone see my error? It’s an error before FreeBSD ever loads

Aw unixwitch I forgot you told me to scrub. Let me check on it

voy4g3r2: ok, ufs auto, main server, ...

fetching distribution

ufs auto and zfs

tried both

voy4g3r2: could you just install 14.0 and upgrade to 15.0 from source? that might be easier

easier, but might take longer

depends how long it takes to fix this :-)

voy4g3r2: does it die on base.txz or what

after a full install, yes it dies on base.txz

just says.. checksum bad.. wanna restart?

16% ... 50%

i get through the WHOLE install of all packages, ports, src, base

it does the checksum check thingie and that is when it dies

and even when you install from th cdrom using disc1 iso

it still chceksum errors

unixwitch: that is possible.. but never did that before

right?

yes it does

another question for the 13.2 to 14.0 upgrade - in the process of upgrading my jails I was no longer able to get a console on the jails after upgrading to 14.0. This tells why, and also how I fixed it pastebin.ubuntu.com/p/5dhDcfFXrj

Title: Ubuntu Pastebin

i just picked the default sets base, kern, kern-dbg, lib32

and it probably worked..

because that is just my luck

has something changed in the way login is done? I am thinking maybe some config was made to the host os, that was not done in the jail configuration

BarnabasDK: yes, opie was removed from freebsd, so you'd need to update pam configs to match

voy4g3r2: 60% overall progress

yup par the course for me also

BarnabasDK: that is, remove opie* lines

77%

BarnabasDK: how did you upgrade the jails? either freebsd-update or etcupdate should prompt you to update /etc/pam.d/* which contain pam_opie.so (which was removed)

voy4g3r2: The checksum for base.txz does not match.

someone might have broken something-- anything on the -current mailing list?

bingo

i did not get that far

i was hosting some people, the wife preferred i speak with guest

that was the last thing i did but now i am being asked to help clean up.. ill bbl

rtprio: i take it the mailing list reference is osmewher eon freebsd.org? i can look

Title: freebsd-current⊙Fo

thank you, i will dive in a little later, at least i am not crazy

i was getting my b*** busted for using netatalk but it was not that..

:)

unixwitch, "sudo ezjail-admin update -U -s 13.2-RELEASE"

ah, no idea about ezjail

seems to use the freebsd-update utility - at least thats how the output looks

anyways - copied in the needed libs in order to be able to get a console and update the packagtes in the jails - but I would like to configure the PAM modules to behave in the jails like the host OS

i wonder if freebsd-update doesn't catch that one for some reason... although i'd expect more complaints if so since it breaks login for everyone

so - I guess there must be a set of config files in the host os - that can be copied to the jail config ditto?

BarnabasDK: were you using opie, by chance? (it was removed)

just edit /etc/pam.d/* and remove any reference to pam_opie.so

heh

or, if you really want to use opie, install it from ports (but you might need to remove it first so you can log in to do that...)

i suppose you could copy /etc/pam.d/* from the host, but that would overwrite any local modifications in the jail

check - I will look over the config files - thx for the help

and no - I would rather see the jails behave in the same manner as the host os

the most optimistic thing i've done in years: + 2 Dec 28 Qualcomm Support ( 25K) Thank you for contacting us

rtprio: i am not seeing anything on mailing list

i do notice that MOST people are just upgrading from 14.0 -> 15.0 CURRENT and rarely run the installers...

there's no real point using the installer since you'll be compiling from source to update current anyway

although, there *is* a point in reporting bugs in the installer, since that's partly why the snapshots exist, so i would drop a mail to -current about this

unixwitch: true and that is exactly right

lets see how i can subscribe to this without having to download all the dang emails :)

if you don't want to subscribe to the list you could file a PR at bugs.freebsd.org instead

Title: FreeBSD Bugzilla Main Page

"bad checksum" is usually a bad download/image transfer

yuripv: but two different people reproduced it in this case, so probably not

yuripv: i thought that, downlaoded to a file server, locally and rtprio confirmed same issue

unless it's exceptionally bad luck

any links?

to the isos used?

no, to reports

scroll up to rtprio's comment

no reports from installer.. not sure how to get them.. this is all new teroritory for me

i could rerun it and try to pull

same iso?

yup

i mean you and rtprio, same issue using same iso (which one?)?

rtprio: could you confirm which one you used? i shared 3 :)

i am retrying with this one: download.freebsd.org/snapshots/amd6…16-ca39f23347e1-266973-bootonly.iso

bootonly doesn't have any installation distributions

okay.. i can try with disc1 then: download.freebsd.org/snapshots/amd6…31216-ca39f23347e1-266973-disc1.iso

what is meant by installation distributions?

i figure.. if result is same for 3 different iso images.. does it matter?

meaning it doesn't include base.txz, it has to download it

ah okay.. let me try with disc1 then which DOES have it, thanks for clarification

voy4g3r2: fwiw, if you plan to run current, it's a good idea to be subscribed to the list, even if it means getting so many mails :-)

unixwitch: i know.. i am just being stubborn :)

and likely to commits-main to know what broke your setup :D

i just want ot fix manual pages :)

which i got a categorized list about 2 hours ago, so new post while this install happens

you don't need to run current to do that, just use the main checkout

the disc1 install is verifying checksums of selected distributions (base, kernel, kernel-dbg, lbi32, ports, src)

checksum fail on base, dang nab it

does it succeed on the image? compare with provided hashes

yuripv: true, it would be interesting to try installing freebsd, as i only installed it a handful of times and mainly raspberry pi images.. which does NOT do an install

checking now..

Title: dpaste/j41D (Plain Text)

looks like we have a mismatch :)

I could see no changes to the PAM in the jails - so I simply copied in the new set of config files from the host - and removed all opie libs from the basejail - seems to work

BarnabasDK: hmm, you shouldn't need to remove the opie libs since the upgrade should take care of that. seems like something went wrong somewhere

(no idea what, though)

ezjail is not really supported well any longer, I should find an alternative

if i'm not a committer, is it okay to open a PR and assign it to myself? (it's for the alc(4) msi-x fix which is stalled, but i want something on the bug list so i don't forget it... plus it is an actual bug)

Not sure you can self-assign if not a committer. I'd ask in #freebsd-bugs.

you don't need to assign it to yourself, you'll be mailed on any changes on that PR

(as submitter)

unixwitch: also, if you have the patch, opening a PR on github or better yet review on reviews.freebsd.org would help

i don't have a patch yet, i'm trying to find some documentation on the card

ohhh there is a way to subscribe to freebsd-current without getting emails.. not sure how this will work but it will allow me to send an emaiL :)

well here goes my first CURRENT mailing list entry

time to see if it is clear enough..

well i guess we can give a 14.0 install a try and upgrade to 15.0...

well egg on face.. there are multiple checksum lines in the checksum256 files that do appear to match.. but why would the installer say they do not ..

errr

voy4g3r2: which images specifically have you tried?

(not just the name, the hash/date as well)

((that is, not just bootonly, the full filename of the iso is sufficient)

)

* FreeBSD-15.0-CURRENT-amd64-20231216-ca39f23347e1-266973-bootonly.iso

* FreeBSD-15.0-CURRENT-amd64-20231228-fb03f7f8e30d-267242-disc1.iso

* FreeBSD-15.0-CURRENT-amd64-20231223-dac33a65b965-267058-bootonly.iso

and now that first current mailing list entry is just garbage.. i am annoyed.. i figured a sha256 <checksum file> <iso> would do the magic

Title: Checksum Error on installer (3 iso images in CURRENT)

this has more details, if required

i am now trying the install one more time with teh bootonly 20231223 one and going to see if i can get more than just the dialog

like a message giving more context..

voy4g3r2: if you roll bootonly, go with 20231228

bootonly is going to download whichever sets you select from here: download.freebsd.org/snapshots/amd64/15.0-CURRENT

Title: Index of /snapshots/amd64/15.0-CURRENT/

iirc it doesn't trust the MANIFEST there, it has its own

actually, that can't be right

hello

either way, with bootonly go with the absolute latest to be safe

i found this chatroom from the freebsd.org website pleased to meet you :D

I have no idea how the mirror syncing works, if there's a chance that you were just getting really lucky and hitting it in some discrepancy between MANIFEST and the dist files

aeronot: o/

if I want to run a GNU/Linux virtual machine (VM) guest also to get some drivers *BSD doesn't have, should I run Xen or is BHyVe good enough or better?

kevans: i will give the latest version a try..

for some reason.. the installer does not generate logs of steps anywhere in /var/log

still downloading..

woohoo, a different error this time

Error while extracting base.txz: Lzma library error: Corrupt input data

and run it a second time, back to the checksum thing.. what a visicous cycle, that can wait till tomorrow

darwin: bhyve works pretty well. what driver do you need that freebsd doesn't have?

Alcor Micro Corp.: USB hub & AU9540 smartcard reader & multi flash reader

they're in a 5.25" 'multi-function hub' I got

unless it's a pci device i don't think you'll be able to passthru it to the guest

it's just plugged into USB & SATA headers

headers/sockets/whatever

again, unless it's pci, you won't be pass it to the guest

and freebsd can't be a dom0

alright. Is it possible any this might change in future?

i would be surprised

voy4g3r2: first error is incredibly concerning

when did we stop being able to run freebsd dom0?

thanks

er, domu. "the host"

it's been a while since i've used xen

rtprio: dom0 is the host, and we reportedly run as both for years now

(on x86)

aarch64 work was in progress, not sure if that was domU or dom0

kevans: that's not whan `man xen` says

BUGS FreeBSD is only able to run as a Xen guest (DomU) and not as a Xen host (Dom0).

sounds like it needs an update then

alright, fired off an e-mail to roger about updating that

i bet there's other missing bits

i'd like to know how to domu freebsd

i just assumed it wasn't a thing

the first couple of results in google for "freebsd xen" are pretty useful there; wiki.freebsd.org/Xen is maybe the authority here

Title: Xen - FreeBSD Wiki

rtprio: Bhyve might well be more flexible. Type 1 hypervisors are a little less flexible than Type 2.

i'm getting along fine with bhyve now, sure

I ran Xen for years, and then moved over to KVM and it was good. Now everything's in a container, more or less.

yep

often they're in a container even when they don't need to be

Not here generally, no.

I'm having a problem with my freebsd 14.0 installation

no services are running, pkg is not able to run the error message is as follows : ld-elf.so.1: Shared object "libcrypto.so.111" not found, required by "pkg"

sixpiece, was just though the same - try using pkg-static

and start with a pkg-static bootstrap -f

will install a new pkg built with a new version of libcrypto / libssl

ld-elf.so.1: Shared object "libcrypto.so.111" not found, required by "pkg"

yes use pkg-static

to fix this particular problem

I did

obviously not :-)

I did though

Title: dpaste/tu6P (Plain Text)

try a pkg-static update first then

everything is up to date

Anyone here switched from isc dhcpd to isc kea? Saw this morning that isc dhcpd was EOL end of 2022. I must be living under a rock.

sixpiece, what is the output of $> ldd /usr/local/sbin/pkg-static

ldd: /usr/local/sbin/pkg-static: not a dynamic ELF executable

as it should be

what if you run that with a full path

"/usr/local/sbin/pkg-static bootstrap -f"

what the linker is telling you is, this is a statically linked executable - not using libraries (shared objects)

same thing

have no explanation for that

root@videotron:/var/run # /usr/local/sbin/pkg-static bootstrap -f

pkg(8) is already installed. Forcing reinstallation through pkg(7).

ld-elf.so.1: Shared object "libcrypto.so.111" not found, required by "pkg"

is it possible to make it work?

I have never seen a statically linked executable (non ELF) bork about a shared object before

So - I really don't know

ok thank you going to go to bed and try tomorrow

pkg(7) would be /usr/sbin/pkg ?

Hi all. Has anyone ever discussed a freeBSD GUI installer like you have in linux?

Rather than an ncurses one?

For those you wondering, we did a Discord Hackathon yesterday, working on superv, a supervised bhyve manager. The meeting was recorded and uploaded to YouTube: youtube.com/watch?v=Dmap39245b0

Title: 2023-12-28 bhyve Discord Hackathon - YouTube

as root, why do i get the error "sudo: you are not permitted to use the -D option with /usr/local/bin/npm" from command 'sudo -u polyex -D /home/polyex/foo'?

thorre: I use Kea

since 2019 I think because ISC dhcpd is not very good at DHCP6

Remilia: Any pitfalls?

thorre: if I recall right, if you configure it to use a database backend it will terminate if it cannot reach the database server

otherwise dunno

Remilia: Mine will be serving a home environment :-)

it is perfectly fine otherwise

i installed konsole on a laptop, but the configuration windows are longer than 1080 pixels and not resizeable :(

Ok, thank you. I will give it a spin

thorre: also it works just fine with powerdns for ad-hoc

(or I think it does? I forgot if I actually tested that thoroughly enough)

does 'su' have any way to chdir?

polyex: What do you mean?

when you do `su -`, it will do a login, changing ot the user's home directory

su seems great why even use sudo?

because all it does is giving full permissions of one user to another which has the permissions to run su. No more fine grains

also thorre indeed I run it in a home environment; was experimenting with using postgresql as the backend for static leases

gnrp so when use sudo and when su?

su requires the knowledge of the superuser password

sudo and doas do not

su also does not limit its invoker in any way, much unlike sudo and doas

nice

what's a good way to make one FreeBSD PC a duplicate of another, like get a list of all user-installed packages and maybe in future run pkg in parallel over ssh?

NFS

maybe I'd better not make them exact duplicates because the hardware differs slightly

mount /usr on both from the same NFS source

when i run pkg upgrade it says updating freebsd repository catalogue... then a few lines of pkg: an error occured while fetching packages. then repository freebsd has no meta file, using default settings. and eventually error updating repositories

but i can ping pkg.freebsd.org so i know network is good. pf is off

same when i run pkg-static bootstrap -f, fails with error fetching, host doesn't resolve, address resolution failed for pkg.freebsd.org... but i can immediately ping pkg.freebsd.org

nvm got it

is it possible to get "global" EQing in FreeBSD that is more granular than just the bass/treble knobs hint.pcm.*.eq enables?

maybe by piping global audio unmodified into a sound fx daemon before output?

polyex: so… jack probably? i have no idea how i would plug that together, tho…

well i could be wrong too. haven't done audio yet

also no idea if jack itself is enough, or if i need something else to act as an lv2 host or somesuch thing.

when you figure it all out pls add to freebsd wiki or write blogs or smth

will do, but i might actually get to doing custom DSPs before, which would remove my need to have this in the OS.^^

ya sounds awesome

earliest i might manage to mess around with that will probably be summer, tho.

and till actual hardware exists, even longer. so an intermediate solution would be nice^^

maybe email the mailing list? like freebsd-audio or smth

unixwitch does this look like a good way to setup ldap server on frebsd? wiki.freebsd.org/LDAP/Setup

Title: LDAP/Setup - FreeBSD Wiki

actually, this might be a good place for me to start doing dsp… no need to build hardware and C + math is the same as on a uC…

mane: i haven't used ldapscripts before (i use ldapvi mostly) but that seems basically fine... i do strongly suggest reading the openldap manual though, it covers some important things not mentioned there like ACLs

thanks

antranigv: where's the source for superv?

thorre: i've been using Kea for a few months, it seems fine. configuration is annoying (JSON) but it works well enough. this is only for a home/soho network, nothing fancy (except dhcp relay maybe).

unixwitch: Thank you for the feedback

I am really glad that I spent to much time on my Neovim setup so json and other things are "taken care of" by LSP, formatter and linters.

But yeas, the configuration for a home environment becomes very big. But I see clera benefits for real world production configuration.

Nice that kea seems to be API dirven so that it can be configured without re-start.

in rc.conf are these two lines sufficient?

kea_enable="YES"

kea_config="/usr/local/etc/kea/kea-dhcp4.conf"

thorre: kea_config shouldn't be needed; it autostarts the appropriate daemons based on whether the file exists, so just don't create the configs you don't want

JSON: slightly better than YAML…

unixwitch: Thank you.

oh no, devel/nspr broken on current

dpaste.com/2BXGKFMKY obvious patch for anyone else who hits this (i assume it'll be fixed soon anyway)

Title: dpaste: 2BXGKFMKY

unixwitch, you're not the first one to mention this. there's currently no PR open for that. Maybe because everybody thinks the same way - this is your chance ;)

jbo: it seems like -current is already on top of it, so i just fixed it locally

although maybe i'll check if there's an existing PR already and open one just in case... need more coffee first

polyex: "why use sudo" - the original purpose of sudo was to give users restricted access to root or other accounts; you can permit them to only execute certain commands, for example to restart a service, but without giving them full root access. Linux (i think Ubuntu specifically) decided to use sudo instead of su for all root access, and other people followed that trend.

polyex: i think the main advantage of this is you can give multiple people root access without needing a shared root password. also, many end users are used to having a single password, not a separate user and admin password, because e.g. Windows desktop works like this.

makes sense ty!

i'll try doas next

same advantages of sudo but way less code and openbsd ppl are good

ever try it?

i have not

just a little nitpick sudo is also from a openbsd guy

oh ty

cve.mitre.org/cgi-bin/cvekey.cgi?keyword=sudo cve.mitre.org/cgi-bin/cvekey.cgi?keyword=doas

Title: CVE - Search Results

polyex: many of those CVEs aren't in sudo itself, if you read the description.

oh shit

hrm.. so got my first email response for the iso issues in installer, for 15.0, a propogation issue..

what iso issues ?

polyex: install of 15.0, based off amd64 iso images, if you download the iso the checksums are not aligned.. which causes the installer to NOT allow you to install the system

ah ya

how's 15 otherwise?

i would not know, i am trying to do a 14.0 -> 15.0 upgrade right now

i updated to current yesterday and it hasn't crashed yet, so take that for whatever it's worth :-)

that is a good sign, at least this 50 millionth install may go right.. for once (joking)

i need to some help with nfs on freebsd

clnt_create: RPC: Program not registered is what i get when i try to mount

mane: are you using NFSv3 or NFSv4?

nfsv4_server_enable="YES" is what i got enabled

did you specify the 'nfsv4' mount option on the client (or vers=4, which is the same thing)?

no

try that, and if it works, you might consider disabling NFSv3 entirely unless you have a reason to use it

v4 is much easier to mount because all the RPCs are inline, it doesn't need rpcbind

ohhh i should disable nfs_server_enable myself then

i went through that hell a few days ago..

the rpcbind stuff

voy4g3r2: you want nfs_server_enable=YES, nfsv4_server_enable=YES, nfsv4_server_only=YES

hello I upgraded to freebsd 14 and having trouble to get it work. Can someone please help me here?

and probably nfsuserd_enable=YES, nfsuserd_flags="-domain my.domain.com" (not strictly required as it will pick up the default from the hostname, but i prefer to be explicit)

is there someone awake that can help me?

ld-elf.so.1: Shared object "libcrypto.so.111" not found, required by "pkg"

like what do I do about this for example someone tried last night I stayed up until 4am est I am now

there is no documented solutions that I can find with google I tried 2 or 3 one giving sim links another installing openssl via the ports both did not work

the ports created another issue I also tried pkg-static -f stuff and that also did not work but maybe there is another variation that I am not aware of

sixpiece: i would try fetching the current pkg package from pkg.freebsd.org for 14, and then install it with pkg-static

can you explain how to do that first part?

or both parts

I'll try to search that

to fetch the package (assuming quarterly on amd64): "fetch pkg.freebsd.org/FreeBSD:14:amd64/quarterly/All/pkg-1.20.9.pkg", then "pkg-static add pkg-1.20.9.pkg"

there's also an issue that uname -v is not the same as /etc/OS-release one is 14.0 release p3 the other is p4

(i haven't actually tried this though...)

I get Not Found

hm, out of date mirror? try this url instead, that definitely works pkg0.sjb.freebsd.org/FreeBSD:14:amd64/quarterly/All/pkg-1.20.9.pkg

any way to make my little bash code better? pastebin.com/VFxsff02

Title: #!/bin/bashd () { s=($EPOCHSECONDS) mkdir ~/.trash/"$s"; - Pastebin.com

polyex: use -p to avoid an error if the directly already exists, and either check for error from mkdir, or add 'set -x' at the start to automatically exit on errors, to avoid trying to move the file if mkdir fails. ('set -x' is a good idea for scripts like this in general)

the most recent version of pkg-1.20.9 is already installed

also /bin/bash doesn't usually work in freebsd... if you really need bash, '#! /usr/bin/env bash' is better

do i set -x at top of the file or at top of fn body?

sixpiece: add -f (pkg-static add -f pkg...)

same error

polyex: well, usually at the top... but is this something you source in your .bashrc? in that case don't, i would explicitly check the exit status of mkdir instead

ya i source it

ld-elf.so.1: Shared object "libcrypto.so.111" not found, required by "pkg"

sixpiece: can you show the whole command and output? -f should override that error... bsd.to/tWOZ

Title: dpaste/tWOZ (Plain Code)

Title: dpaste/OiQG (Plain Text)

sixpiece: show the output of "ldd /usr/local/sbin/pkg"?

how am i supposed to use a var? $s works but shellcheck complains. so i tried "$s" but that complains, same with "${s}"

I did this before btw

Title: dpaste/vfmF (Plain Text)

polyex: that depends on the actual complaint

sixpiece: does it work if you use the entire path, i.e. "/usr/local/sbin/pkg update"? if not, please run "truss /usr/local/sbin/pkg update" and the output (it will be quite long)

meena shellcheck.net/?id=cb64225 trying to make warnings go away

Title: ShellCheck – shell script analysis tool

not that I know of I tried that but will try it again just in case

yes it works unixwitch

on the first one /usr/local/sbin/pkg update

everything is up to date

okay, well you should be able to do an upgrade now - use "/usr/local/sbin/pkg upgrade -f", the -f is required when upgrading to a new freebsd release

i'm not sure why it doesn't work without the path; if it still doesn't after you did the upgrade, show "ldd /usr/sbin/pkg" and "type pkg"

(as in the output of the command "type pkg", don't just type pkg :-)

yes, thank you performing the upgrade now

will keep you posted after it's done

polyex: it's complaining because you assigned $s as an array then used it like a string. you might want s="$EPOCHSECONDS" instead since there seems to be need to use an array here

pastebin.com/6dQQTkTV doesn't make shellcheck warnings

Title: #!/bin/bashd () { s=${EPOCHSECONDS} mkdir ~/.trash/${s}; - Pastebin.com

(i'm not sure what $EPOCHSECONDS is though... is that a bashism?)

ill try your version too

ya

it gets seconds since unix epoch

yours is fine too, but i prefer it with quotes around the expansion even if it's not technically required, i just think it's a good habit

(as in s="${EPOCHSECONDS}", you can combine the {} with the quotes)

fwiw, a non-bash way to do that is s="$(date +%s)", but since this is specifically for bash there's no need to do that here

ok fixed pastebin.com/cuuSVS7T

Title: #!/bin/bashd () { s="$EPOCHSECONDS" mkdir ~/.trash/"$s"; - Pastebin.com

ooo

shell neutral is nice

date + is useful in general though, for example date +%Y%m%dT%H%M%S to get an ISO-style timestamp

it still needs the path

what?

sixpiece: ldd /usr/sbin/pkg?

still has the issue though

Title: dpaste/h4zS (Plain Text)

libcrypto.so.111 not found

okay, i don't know the answer to that, but maybe someone else does. i thought /usr/sbin/pkg comes from the base system, so it should have been updated with freebsd-update

sixpiece: can you try "freebsd-update IDS"? that should check all the installed files are correct for your current release

still having that issue

it did something though

it won't actually change anything, it just returns mismatches. can you show the output of it?

(it might list some files in /etc, which is fine, since you're meant to edit those)

Title: dpaste/L07Y (Plain Text)

a part of it is cut off

sixpiece: try "freebsd-update IDS 2>&1 | grep -v /var/db/etcupdate" to filter those out, so the rest is visible

or "freebsd-update IDS 2>&1 | grep /usr" might also work

Title: dpaste/qwGS (Plain Text)

oh you're using csh, uh, i don't remember the syntax for that. can you run it again in sh?

how to switch?

just type sh

then run the command

did the second one and no difference now doing the first one

aka the might also work now doing the first one

freebsd-update IDS doesn't change anything (so it won't fix the problem), just show the output

at this point, i honestly think virtualbox is doing something "wrong"

i download the 14.0 iso and i am STILL getting checksum issues, heck i am evening installing the iso from the same dang machine

Title: dpaste/8sci (Plain Text)

I don't get all the input some is chopped off

voy4g3r2: that is weird. is that with disc1?

sixpiece: i think something has gone wrong with your upgrade somehow. is it possible freebsd-update got interrupted while running? i would try another "freebsd-update fetch", "freebsd-update install" and see if that helps

Title: dpaste/GQsp (Plain Text)

basically fetch says no updates need install says run fetch first

sixpiece: you need to put the correct files in place somehow, but i'm not sure what is the right way to do this with freebsd-update (i usually install from source). i suggest asking on one of the mailing lists or the forum, include the IDS output to explain the problem

(unless anyone else here has an idea)

unixwitch: yup and bootonly

voy4g3r2: any chance you can try a different virtualisation system (bhyve, hyper-v, ...)? just to see if it's an issue with virtualbox

i was curious to see if there is a way i can get a raspberry pi to use bhyve

but then i introduce a new arch.. so was trying multiple other options before

i wondered about that the other day, i found a few references to porting bhyve (vmm.ko, really) to arm64, but it doesn't seem to have been merged

i think, to test out htis manual page thing.. i may just go with a virtualbox image already setup

which i found surprising given the recent interest in arm64 as a server platform

hopefully that will be the path of least resistance

Title: Index of /snapshots/VM-IMAGES/15.0-CURRENT/amd64/Latest/

lets see if these work..

Title: ⚙ D37428 Import the kernel parts of bhyve/arm64

looks like it's taking a while but in progress

haha t least the vmdk files work..

no checksum issues there..

unixwitch: pkg is not part of the base install.. i just had to install pkg to be able to get packages.. i think i saw you ask about that

well pkg install not working.. i am going to do yard work.. this is frustrating

voy4g3r2: /usr/sbin/pkg is part of the base system, isn't it? it's the thing that bootstraps /usr/local/sbin/pkg

so you can run pkg on a fresh install with no pkgs installed

thank you unixwitch

i wonder why SCTP isn't more popular, it seems kind of neat. was it just too complicated for most applications?

is there any help here?

or I need to revert back to the way it was before

unixwitch: I use it since 2001 for every single telco grade application I run. (if you meant stream-control at transport layer). I guess it not popular cause it is not suitable for regular applications. end-user applications does not require same level of reliability I suppose

cybercrypto: interesting, i have heard it's used for some specialised applications

what's sctp?

polyex: sctp(4), it's sort of like a combination of UDP and TCP with some extra features (like multiple endpoint addresses)

sixpiece: possible

wow

how does it compare to quic?

thank you orbifx how do I get it to work?

sixpiece: get what to work? If there is context, I'm missing it

polyex: it has nothing to do in comparison with quic i guess. sctp is for reliability, security (built in encryption supported) and multipath routing resilience. quic has a different purposes.

polyex: sctp is designed to make possible to have phone calls over IP network (in dummy words). SS7 is a wireless protocol that runs voice circuit switches intra mili-second. It also detects failutes in less that 10ms. TCP would never be able to replace it. At that time, SIGTRAN was borned.

wow

you must know telco stuff pretty deep

orbifx for example I want to be able to use the pkg command without specifying the entire path in order to be able to do so

on the off chance anyone is interested in sctp on freebsd, i just open freebsd/freebsd-src #965 to add some missing manpages... someone here was complaining about this the other day but i can't remember who

let's start with that

Title: sctp: update manpages by lexiwinter · Pull Request #965 · freebsd/freebsd-src · GitHub

niiice

some people does not understand the differences between voice of ip and telephony of ip (voip/toip). And thats really key. (your mobile apps like whatsapp runs over VoIP, not ToIP)

sixpiece: where is pkg installed? is the location included in $PATH?

root@videotron:/lib # $PATH

Title: dpaste/0x7X (Plain Text)

sixpiece: used `echo` to print variable values

use*

sixpiece: anyway, those are the locations (separated with :) your terminal is looking for command. If pkg is not there, then you have to write the full path to execute

orbifx: sixpiece's problem from earlier is that something has gone horribly wrong with their upgrade to 14 and most of the base system has checksum mismatches reported by 'freebsd-update IDS'. so their /usr/sbin/pkg doesn't work because it's linking to the 13.x version of openssl

unixwitch: I am not a developer to understand the code and compare to RFC statemet of compliance... I could take a look anyway - feel free to ask any other aspect of the features and use cases of it.

orbifx: we managed to get /usr/local/sbin/pkg fixed, which is why they need to type the full path to use pkg, because /usr/sbin/pkg doesn't work (it's not PATH related)

unixwitch: thanks that helps

unixwitch: did upgrading the whole system work the second time?

orbifx: i suggested running another freebsd-update fetch + install but it didn't do anything

orbifx: my next suggestion would be to extract base.txz, i guess, but i don't know how this works with freebsd-update because i don't use it much... so i didn't want to suggest something that would cause more problems

unixwitch, sixpiece: fresh install?

i think that would be my choice in situation... or roll back to an earlier zfs snapshot / BE if there are any. i'm really curious how the system ended up in this state though

thank you orbifx everything is almost working I got help on the discord server

sixpiece: cool :)

[15:37:12] <sixpiece> there's also an issue that uname -v is not the same as /etc/OS-release one is 14.0 release p3 the other is p4 <- check the output of `freebsd-version -ku` and you will see that you have p3 kernel with p4 userland

unixwitch: that is odd, i am trying to do a pkg install on 15 and it is telling me i need to download pkg

unixwitch: "extract base.txz, i guess, but i don't know how this works with freebsd-update" as I already said here several times, extracting any of the distributions (base, kernel, src, *-dbg, lib*) and then running freebsd-update will update the files that require updating

ok, i'm sorry i didn't see you say that several times? feel free to let the person affected by the issue know though...

yeah it was a week or two ago with you present, but you can also test it yourself in a VM

not like it matters

so usb drives moved to one enclsoure instead of 3.. but now when i look at zpool status, i get this.. bsd.to/TYOk

Title: dpaste/TYOk (Plain Text)

why is one drive NOT like the rest.. it is setted right, i am doing a zpool scrub right now to make sure everything is good with the move

Other than locally manually updating the leap-seconds file is there a FreeBSD plan for updating the leap-seconds.list file?

I have been watching it tick down to expiration expecting it to be updated with a base update but that did not happen.

rwp: it's already fixed in current by changing the URL, but it hasn't been MFC'd yet because there's some discussion about whether it needs an EN, i think

Of course I can trivially update it myself locally. That's no problem. It just feels like something that affects everyone and so should be globally applied.

rwp: bugs.freebsd.org/bugzilla/show_bug.cgi?id=275419

Title: 275419 – ntpd: leap-seconds.list should not run inside jails, and IETF is no longer hosting it

Just as background noise I'll note that Debian has fixed this with a tzdata update but that Ubuntu has not yet applied an update and Ubuntu has also expired.

Thanks for the bug ticket number. Reading through it now. Interesting stuff in it.

actually it was MFC'd, but i guess the EN hasn't gone out yet cgit.freebsd.org/src/commit/libexec…342381530cfc5dc78a457cb1f9e355d687a

Title: src - FreeBSD source tree

rwp: fwiw, this doesn't actually cause a problem because no leap second is being added this year. but it should still be fixed, of course. (just in case you weren't aware)

Right. Nothing is happening. But it logs an error and I actually do look at logged errors. So all of my collective is nagging at me about it.

I hate to put in an ignore filter for it because nothing is as permanent as a temporary fix and almost certainly I will forget to remove it when it is finally updated and not needed anymore.

I did not realize that in FreeBSD the leap-seconds.list file is fetched routinely. I had thought a static copy was simply part of base. TIL.

yeah. although iirc there was some discussion about that in the PR - it's actually in base *as well* because tzdata includes it, or something like that

It's such a slow changing file that I had expected it to be in the tzdata in base. And also because that is how it is done elsewhere.

it may end up being easier to just ship it in base and send out an EN whenever it changes... which was last, what, 2017? ... rather than having all this complex code to get it

I am down to comment#22 which describes the two files.

although i think part of the issue is that the signature expires more often than it actually changes

They are trying to avoid the problem of having a file that is still date valid but actually needs to be updated because a change has appeared in it.

So they try to expire it before there can be any actual physical possibility of needing it updated.

well, qualcomm replied to my request for AR81xx docs with "we don't provide any documentation for our ICs". which was pretty much what i expected, but still... i find this really bizarre. it's not like some 10-year-old Ethernet IC contains super secret technology.

on FreeBSD 13.2-p8 with apache24-2.4.58_1 , apachectl reload apachectl graceful always results in a core dump: kernel: pid 33365 (httpd), jid 0, uid 0: exited on signal 11 (core dumped)

is there a known bug on this ?

I have to restart manually from usr/local/etc/rc.d/apache24 restart

Yes: bugs.freebsd.org/bugzilla/show_bug.cgi?id=268318

Title: 268318 – www/apache24 with www/mod_php8{0,1,2,3?}: opcache + ASLR turned on crashes Apache

thanks - that bug is wild

elfctl -e +noaslr /usr/local/sbin/httpd does fix it for me, although it doesn't seem right

At the time I gave up and moved my remaining apache server to nginx.

an alternative to switching to nginx would be to use php-fpm, which is probably preferable to mod_php anyway, as it doesn't load all of php into every apache process

meena temporary here, until I polish some things and move it to GitHub

Title: antranigv/superv - superv - git.bsd.am: Git with a cup of BSD

antranigv: why not keep there? ;)

is this similar to got on openbsd?

last1: I second the above suggestion, use php-fpm, it is a better way and gives you more control

orbifx: I don't think so. that's just a git forge (gitea or forgejo) — or what do you mean?

meena: what the superv project is, I think I misunderstood it completely

Any freebsd-ers on nostr?

is nostr still a thing?

orbifx: superv seems to be a wrapper for bhyve, sort of like vm-bhyve in ports/packages

Title: GitHub - churchers/vm-bhyve: Shell based, minimal dependency bhyve manager

I am more interested in what is different about antranigv's superv when compared to that one

it is fair to say, if i am working with CURRENT, i should just default to the ports tree to install things?

voy4g3r2: why?

i am trying to get pkg(8) on this virtualbox and i keep getting A pre-built version of pkg could not be foudn for your system

pkg: Error fetching pkg.freebsd.org/freebsd:15:amd64/latest/latest/pkg.txz: Not found

voy4g3r2: it should be pkg.freebsd.org/FreeBSD:15:amd64/latest/Latest

Title: Index of /FreeBSD:15:amd64/latest/Latest/

oh crap, i typed that wrong

but even if i correct the ABI I can't get pkg.txz

anyway, I will need to put the kid to sleep, then I'll be able to actually get to the computer

[21:41:15] <meena> but even if i correct the ABI I can't get pkg.txz <- because it is pkg.pkg

not pkg.txz

voy4g3r2: change the extension to .pkg

in your URL

like pkg.freebsd.org/FreeBSD:15:amd64/latest/Latest/pkg.pkg

that is what freebsd is doing, i do not have that

what do you mean you do not have that?

the link I just posted does not work for you?

i do pkg

it then says, it is not installed.. waant to install it, i go y

pkg-static install pkg.pkg

after downloading

yueah pkg-static does not exist

you can get pkg-static from any 14.x or 15.x install

and run it on your system

hrm.. okay, time to figure that out.. this CURRENT work is a lot harder than just downloading an image and going with it

makes you wonder why pkg(8) is not base

pkg.txz should be the last resort, not first

if it's not finding pkg.pkg, that's a lot more important than the message you've presented

i am trying to figure out how to enable copy/paste in virtualbox

Remilia superv is... supervised!

voy4g3r2: that should not be possible

antranigv: what does this imply? monitoring of specific processes inside the VMs?

Remilia to be clear, superv uses runit underneath, but we're planning to move to s6 soon, as soon as crest makes managible... managable... manageable? manageable config files!

Remilia for example, if you're using bare bhyve or vm-bhyve and a VM crashes, there's no way to know that it needs to be restarted, but in this case, it's possible.

since runit scripts are monitoring it

Remilia so technically we are monitoring the bhyve process itself. and doing some tiny things as well, like clearing stale vmm interfaces, managing tap interfaces in a sane way, just basic things for now.

antranigv: i really wish we had a native supervision… thing

I have a problem I didn't encounter before.... make buildworld is failing on one of my older servers

Title: dpaste/vTxV (Plain Text)

jbo: can you do a make clean in /usr/src/contrib/bc (or whatever it's called) and remove the appropriate /usr/obj files?

antranigv: I see; it is somewhat weird to me since I think vm-bhyve restarted crashed VMs for me

meena, did that a while ago