nullfs seems to be fine

(other than the issue with procstat -v, but that's mostly harmless)

poudriere for example makes very heavy use of nullfs for jails (in the absence of zfs)

that sounds too complicated at the moment, but i will figure out how to trim down the world

for ufs, there's also the option of mount -o union

which unions only on the mountpoint dir, unlike unionfs

this is a sas hardware raid, so it's ufs

"unfortunatly"

can we install listxattr on freebsd somehow? it's a dep of a rust thing i wanna compile

No

Is "listxattr" available as package|port?

doesn't seem like it

Get the source to compile (& install) then

it's a linux syscall?

Oh

what's weird is i found this solidiquis/erdtree #160 to make building it work on *bsd, yet it changes a cfg! from (target_os = "linux") to not(target_os = "macos")

Title: Use 3 arguments listxattr() on non-macOS Unix by iquiw · Pull Request #160 · solidiquis/erdtree · GitHub

so it runs on freebsd... yet we don't have access to listxattr? crazy netbsd guys

not sure what to do now

who said it runs on freebsd?

well the comment in that PR "In order to support *BSD especially."

is it ok to edit /etc/ssh/sshd_config or should i edit something in /usr/local/etc?

edit /etc/ssh/sshd_config

ok ty

Bummer. I lost sound upgrading from 13.1 to 13.2.

just finished upgrading a bunch of boxes to 13.2 from 13.1 and all's well, ty very much freebsd project!!

Is there a tool in ports to monitor keyboard state and emit events on changes?

I basically just want to monitor the state of caps lock, num lock, and scroll lock in a blocking way from a script so that it emits a line for each state change.

misc/kbdscan may be a starting point for this?

Ah, that would probably work...but not ideal since it appears to automatically exit after 5 seconds of inactivity.

Oh, I'm an idiot. I totally forgot the --query flag to evtest.

Powercycling brought sound back, apparently. Not sure what happened.

anywhere to get commit stats for fbsd going back 5-10 years?

wanna look at trends and stuff

Github?

polyex: what kind of trends are you looking for, exactly?

meena commits a day, committer quantity, ftp bw from ppl downloading installers

I think first two can indeed be extracted from the github API. note that committer and author are distinct concepts. but the third one has nothing to do with the version control

freshbsd.org/freebsd/src/branch/main aggregates some info, but I'm not sure you can easily extract stats

Title: FreeBSD / src - FreshBSD

the split from committer / author happened only with Git, although it was present from the start. but I'm fairly certain we don't have stats of that from the before time. it would've been a lot more effort to convert the subversion tree with that in mind

that means we can only really (easily) quantify outside contributions since git

if you don't want to mess up /etc/profile but want some system-wide variables/aliases, would you put those in scripts in /etc/profile.d ?

yes. with .sh extension

# Load each .sh file in /etc/profile.d/, then /usr/local/etc/profile,

# then each .sh file in /usr/local/etc/profile.d/.

so that's the order it does it in?

wish i could figure out why my server was crashing

rtprio: Anything in the loggs?

nothing. it's at a family members house. there's nothing notable on the screen when it hangs. nothing in logs after it's rebooted

i hope it just needs vacuumed out, but i doubt that's it

I assume it has non ECC memmory?

If the machine has more then one memmory stick try removing one of them at a time

Got all kind of confusing behaviour.

it's a dell r710; it's ECC. i actually ordered ram there which i will be putting in next week

going from 48gb to 288gb. it's runs bhyve vms. and will be running a lot more of them.

Have you run iDRAC diagnostics? The built in tests from DELL are quite good.

Just to rule out some (not all) hardware faults.

i'll have to do that when i'm there. and get the drac on the network. it's drac-light so it's missing a dedicated drac lan

Make sure that you upgrade all the firmware, there may be some bugs that DELL has patched.

it's pretty old now, but i'll check

good morning everyone

hello there

I was wondering if anyone could speak to the maintainers of the transmission-daemon port and ask them to update it? its been 8 months since the last update and I think the daemon is a relatively simple package. Thanks

Title: FreshPorts -- net-p2p/transmission-daemon: Meta-port for Transmission BitTorrent client

there has been some updates here

Title: 269504 – net-p2p/transmission-cli - New version available 4.0.3

is it waiting approval or something?

I was considering doing an old school make and configure but then I saw that bugzilla page :)

I would only make a mess been 20 years since I built from source, I remembe suse 5.2

OK i see it is still under review

Title: ⚙ D40910 net-p2p/transmission: Update 3.00 -> 4.0.3 & modernize port.

thanks!

anyone alive here?

Is there some problem with efnet.org? irc/channels seem down too.

efnet's irc channels are working but that webpage seems to be down

quit

sorry for offtopic, news.ycombinator.com/item?id=37011324

Title: Bram Moolenaar has died | Hacker News

ngortheone: There's #freebsd-social too, which is often quite a reasonable place to be.

yes, sad news :'(

having trouble mounting nfsv4 shares as a user with vfs.usermount=1 enabled and user having enough rights to mountpoint (root mounts without problems)

what error?

mount_nfs: nmount: /mnt/temp: Operation not permitted

what exact mount command?

looks like filesystem permission error. but user can touch files inside mountpoints

no command but mount from fstab 10.10.10.10:/srv/temp /mnt/temp nfs sec=sys,vers=4,rw,soft,noauto,exec 0 0

mount -t nfs 192.168.10.10:/srv/temp /mnt/temp hangs

actually spits out ([tcp] 192.168.10.10:/srv/common/temp: RPCPROG_NFS: RPC: Port mapper failure - RPC: Timed out) after a min

if the server is nfsv4 only then mounting without the nfsv4 option won't work

(nfsv4 does not use portmapper)

-t nfv4 or nfsv4 in fstab do not work like on linux clients

and nfs_client_flags="-n 4" is enabled in rc.conf

ohh wait that is a -4 isnt it ?

what fbsd version?

(I see nothing that uses nfs_client_flags anywhere? where did you get it?)

if portmapper is not responding, make sure your server has port 111 open and rpcbind running;)

13.1

portmapper is not used for nfsv4

sure, but there is still the error message, you can not ignore that;)

server is not running portmap ( root mounting works fine) and linux clients are just fine without it on server

micdud: so, what _EXACTLY_ did you do that produced the "mount_nfs: nmount: /mnt/temp: Operation not permitted" error?

mount -t nfsv4 10.10.10.10:/srv/common/temp /mnt/temp

micdud: as what user?

mount -t nfs 10.10.10.10:/srv/common/temp /mnt/temp rather

as user yes

that command is the one that you said gave the portmapper error?

yes

it gives a portmapper error as root and "Operation not permitted" as a non-root user?

If you read mount_nfs, you should notice: nfsv2 Use the NFS Version 2 protocol (the default is to try version 3 first then version 2).

therefore, to mount v4, you need mount -t nfsv4

mount -t nfsv4 192.168.10.10:/srv/common/temp /mnt/temp

you mean -o nfsv4

yes, sorry

mount -t nfs -o nfsv4 ...

or -o vers=4 (same thing)

if you want to use the options from fstab then do mount /mnt/temp with no other options at all

192.168.10.10:/srv/common/temp /mnt/temp nfs sec=sys,vers=4,rw,soft,noauto,exec does not work for user

works for root

yes, because by default, users are not allowed to mount.

how EXACTLY are you doing the mount as a user?

tsoome: they already said they have vfs.usermount enabled

vfs.usermount=1 enabled

mount /mnt/temp

I assume this is not inside a jail?

no jails

is the nfs client module built into the kernel or already loaded?

i.e. does nfs show up in lsvfs

shows up

ls -lod /mnt /mnt/temp shows what?

drwxrwxr-x 2 root wheel - 512 Aug 6 14:26 /mnt/temp

weird ls -al shows different

and user can touch

that explains the error

ahh nevermind ls -al shows same

it's not enough for the user to have write access to the mountpoint dir, they are required to _own_ it

(or have PRIV_VFS_ADMIN granted by some mac module, but that presumably doesn't apply here)

hmm , chown and ug+w same prob though

ls shows what now?

drwxrwxr-x 2 dan dan - 512 Aug 6 14:26 /mnt/temp

are users in client and server in sync?

+ nfs4 domain

uid and guid same

and user names?

what is nfv4 domain ? same a krb5 domain ?

yes username and guid uid same

and the exact error from the mount command is the same?

mount_nfs: nmount: /mnt/temp: Operation not permitted

You checked perms on / and /mnt I presume?

those would likely give a different error, but check anyway

micdud see -domain description in NFSUSERD(8)

Usually not permitted means it is running asd non-root for some reason.

I did ask for ls -lod /mnt /mnt/temp and not just /mnt/temp

I saw that request, but never the answer

drwxr-xr-x 5 root wheel - 512 Aug 6 10:20 /mnt

that's fine

user is in wheel

k

ktrace it ?

never used that but mount ; ktrace -i mount ?

ktrace probably won't be informative here

got a binary out :)

but try it anyway: ktrace -i mount /mnt/temp and then kdump -E

look for the nmount call and paste that part of the output to dpaste.org or termbin.com

nmount not in there

oh? then look for an EPERM

(but there must have been an nmount for it to report that in the error?)

You checked the logs on the other end too, I take it?

1164 mount_nfs 0.005518 CALL nmount(0x801986000,0x16,0<><invalid>0)

1164 mount_nfs 0.047249 RET nmount -1 errno 1 Operation not permitted

Do you see the input params?

ktrace doesn't know how to log them, afaik

Ah, maybe that is the openbsd version

Or dtrace.

I can't recall

Might be NAMI's in there ?

Look around that call also for read()s from the filesystem., Sometimes you will see config files and salient bits

oh, there's no NAMI, that's significant

Might be after GIO fd [0-9] read

I don't have a remote volume handy to mount

Was just looking at trace of mount -a

micdud: you're sure that vfs.usermount is set?

Oh, you are still doing this as non-root?

apparently it works as root

# sysctl vfs.usermount=1 leads to vfs.usermount: 1 -> 1 so i think so

As Operation not permitted is usually a check for uid 0

Client is allowed in server's /etc/exports if applicable?

it's not getting as far as the server as far as I can tell?

if root can mount , and yes

root CAN mount ?

micdud: there's no NAMI line in the ktrace between the CALL nmount and RET nmount?

micdud: or any other line for that matter?

Maybe termbin a grep -n30 around it

just 2 lines of 1164 mount_nfs 0.005545 NAMI "/mnt/temp"

ok, so there is a namei call

so maybe it is reaching the server

this is getting messy , ill post on forums i think

and ps: i did have it work before somehow , even with mit krb5 , just do not write down exact steps (might need to just try again from a fresh install)

Where does mount get the root perms to allow a user to mount? It's net suid, nor is mount_nfs ?

Or is that not required? Weird.

mountd perhaps.

or nfsclient

perms checks for mount are done in the kernel

including checking for root vs. user and in the user case, whether they own the mountpoint

Interesting

i suspect a large spike in freebsd usage if ms forces win 11

Did you happen to have that bit of the source up? (like know right where that is)

the only reason I use windows for home user is computer games

for work is a different story

how is that freebsd related?

I am pretty at the point that I would switch to bsd for desktop use unfortunately that is not a consideration these days

it doesn't matter anyway bsd community don't care

there is no desktop bsd any longer

no suprise there

*clwn emoji*

I remember a guy like RhdoiumToad from college

problem is there is no marketing there for custom X

and 1997 was the year I went to college

Yeah, I was married by then, lol

do you know that transmission is used by 1000s of people?

is freebsd obnoxious

probably under paid over worked

so long as it is not government psyop bs i am fine

8 months to develop your malware haha joking

Clown emoji face

freebsd went woke already

years ago

feel free to take your trolling elsewhere

RhodiumToad: thanx for trying to help, i will spin up a new vm with 13.2 later, and try again from scratch (if does not work with then ill take it to the forums)

if you are thinking that is government psyop ps

it takes 8 months to update the largest torrent application ?

clown emjoi face

busy ya

call me crzzy all you want no answer for why it takes 8 months to update the most popular torrent client on freebsd

if i have a vm, is it hard to set up traffic proxying on it to jails the vm hosts? i wanna work up a solid system for using jails like app containers

but that means translating network to the vm, into the right jail (container) internally

in and out

jails can either have their own private IPs, or they can share one of the host's IPs

basically, vnet jails have networking isolated from the host, but usually with a virtual network link to a bridge device on the host

whereas non-vnet jails use the host's networking but constrained to one IP address

on the host end, it can either just be a bridge, or you can have a full-on nat or transparent-proxying setup

a lot depends on how you want to organize the IPs

i was thinking to have host have single ipv4 address, then internally have the jails use ipv6

to help save addresses even tho i want logically separate systems

that'd be rather hard to do

so i can put 5 systems on a single vm in the form of a jail but only use 1 ipv4 address

oh dang

if the host is v4-only, then you might as well make the jails private ipv4 addresses

oh ya huh

anyway, to do that you'd use nat on the host, with port forwarding rules if the jails are providing services

can nat route based on http host param? like setting up vhosting but each host is in its own jail

same ip to host, then www.foo.com to jail A, www.bar.com to jail B

no, for that you'd want to put a proxy such as nginx in front

hmm

kinda like now where nginx reverse proxies to socket file paths for my different sites, but instead proxying to the internal ip of the right jail, right?

ya

hm ok

because i want to make immutable infra, so i need to use jails for the actual workload, so when i update my app i just build a new jail then deploy that to the host

i could use vms for the workload, but then i need n vms (which have a $ cost) vs n jails (which can be run on a single vm, or more, so more flexible cost)

i'm thinking about that right?

seems reasonable to me

ty

when i start a new vm from an install iso i have to run through the installer. but that's error prone and tedious and slow. can we modify installer isos so all of our configuration is already in place?

so just boot the iso and it boots a configured server ready to go?

I wouldn't use an install iso, I'd make my own vm image

how do you do that?

are you doing this on your own infrastructure or on a cloud provider?

(because I'm not an expert on clouds)

neither, rented vm from a hosting company. through web panel i upload an image to run when the vm starts

so far only given it installer isos but i guess it doesn't have to be

like a vultr vps or whatever

the simpler option for you might be to customize the install iso; you can script the whole install and add packages and so on

man bsdinstall might get you started

ya but you make your own vm images. how do you do that? like start a new vm in a hypervisor, configure it, shut it down, clone vm image and in the future just launch new vms from that image?

I make mine from scratch from source, using buildworld/installworld

but that's more work than most people would want to do

you can instead install a distributed version into an image file using bsdinstall, I think

I haven't played with the installer much to be honest.