aic: thanks, I'll try that in the morning.

if i'm only adding 1 ip to a system do i need the "alias" in "ifconfig_em0_alias0" or can i just use "ifconfig_em0"?

kerneldove_: the primary IP address on an interface is not an alias, so you should not use ifconfig_em0_alias0

That's not what rc.conf(5) says - ..._alias<n> starts at n=0.

so who's right? :/

the primary IP address goes in ifconfig_em0. you only use the alias entries for the second and subsequent IP addresses, which are called aliases

a system with only a single IP address does not have any aliases

ok that's what i thought. would be helpful if rc.conf included an example of that no?

there is an example in rc.conf(5), but for whatever reason it's documented under the entry for network_interfaces. search for ifconfig_

kerneldove_, I think both are right. I misunderstood what ivy meant, I believe.

(The first alias, which is the second IP address, is _alias0. This is admittedly confusing.)

if this was being designed from scratch i think there are several much less confusing ways to do it, but unfortunately it dates from when aliases were a new and fancy thing

iirc someone told me once to always use the _alias# syntax because that's easier to change than if you have an ip directly on the interface itself is that true?

I have no idea how that would be easier.

aliases *are* directly on the interface, they're just configured with the 'alias' keyword, so i don't see how that's true

on some systems i assign addresses to lo0 rather than the interface, in which case they have to be aliases, but this only works for ipv6

ok ty

what would happen if i used the igb driver for an intel nic capable of using the ixl driver? less performance or?

i would expect the card to not work at all, unless the hardware has some sort of backward compatibility system

if you mean em instead of igb, they are the same driver, the interface name is just different for historical reasons

no ive been using igb0 on this rented server but i just noticed it has ixl0 too

that is a different network port

V_PauAmma_V: This computer has an externally visible IP that I want to assign statically, and an IP on my internal network that I want it to get via DHCP

On the same interface? You like living dangerously.

V_PauAmma_V, why is that so dangerous

Because if that server (or anything else on that network segment, really) is compromised, the attacker may get to observe traffic on your internal subnet as well. Unless I'm missing something of your setupn that's something you definitely don't want to happen.

s/setupn/setup,/

i make a lagg0 with 2 physical nic ports in it. the lagg0 is then a member of bridge0, and bridge0 has the ip assigned to it. but i see in pf it's seeing traffic come in on lagg0, not bridge0. is that normal and expected?

folks, I'm moving to terminal-based email client, because that's where I spend most of my time. Any suggestions on a good email client?

mutt

In the days before e-mail was filled with MIME crap I used to like elm

mutt

by mime crap do you mean attachments

anything other than plain text e-mail

e-mail wasn't designed for file transfer, that's what FTP is for

i make a lagg0 with 2 physical nic ports in it. the lagg0 is then a member of bridge0, and bridge0 has the ip assigned to it. but i see in pf it's seeing traffic come in on lagg0, not bridge0. is that normal and expected?

and is it right to put the ip on the bridge and not the lagg?

just wanna make sure

Sounds fine, lagg0 is a bridge port, so of course you can see traffic there. You can filter on the bridge, but also on and between the bridge ports

Think there was some caveats for rules on bridge ifaces, check the man pages

kerneldove_: plase do it

mzar do what?

move address to the bridge

ya it's on there

i think i read in freebsd 15 it's not even allowed any more to put ip on member of bridge

lessless: I heard mmh is good, but I have yet to test it myself

mmh looks unmaintained, the last release was in 2019

although nmh (which it was forked from) is also barely maintained nowadays...

lessless: aerc, alpine, mutt, neomutt... I use aerc for a couple of years now.

First time I hear abou aerc, thanks! Been using mutt for hm forever, but this one seems worth a try

Seconding aerc. I think it's also common to pair it with mbsync/isync to download the emails as local files ("maildir" format), and then read the email from those files.

When sysctling to use kernel gdb in 16-current I get

sysctl: debug.kdb.current=gdb: Invalid argument

I checked, option GDB is in. That is the GENERIC kernel.

hey! what hardware do you expect to back this?

bhyve

I think there is an issue with port detection for gdb and it might be that no module is loaded early in boot to register itself

so a uart?

yes

can you create a bug I'd like to add this to my todo list

will do

as a work around you can use gdb on the bhyve process (bhyve(8) -G flag)

let me know the bug and I'll take it, unless you can assign it to me

Need a moment, my bugzilla password seems to have changed.

@[tj] bugs.freebsd.org/bugzilla/show_bug.cgi?id=289434

to be clear, using the -G flag you just attach gdb to the tcp remote on the vm host

you won't get gdb listed in the available debugger backends

okay, dch & divlamir got my podman setup completely unblocked. Anyone got a preference among docs.linuxserver.io/images-by-category of which to port first? I'm starting with smokeping since I've used it a bunch and it's one of their oldest images.

what's the easiest way to test boot a custom built kernel in qemu? can i use -kernel somehow?

yeah

-kernel file -initrd file -append 'console=ttyS0 root=...' ...

oh. thought this was #qemu

actually unsure of -kernel works with kernels other than linux

don't quote me on this, but i think for some platforms (arm64, riscv64?) we have a build option to generate a kernel that Linux boot loaders can load

Hi guys

I had some trouble connecting to my IRC

I am mosaid

I successfully install Ubuntu jammy in jail, but can't run wine on it..

I get back Segmentation fault,  I this related to needed gpu drivers or something?

ivy: well i'd need amd64...

leah2: you might find it easier to use bhyve with bhyveload(8), if that's an option, it's specifically designed to load a freebsd kernel directly

i feel like this should be possible with qemu too, but i don't know how off hand

hm, problem is my kvm host is linux and i'd need nested virtualization then

huh that url from josephholsten with a bazillion linux containers is interesting docs.linuxserver.io/images/docker-hedgedoc/#environment-variables-e

it has all the containerfiles used to build them e.g. github.com/linuxserver/docker-hedgedoc.git

ivy: so i see a menu, but then it fails with can't find '/etc/hostid' can't find '/boot/entropy'

pretty sure those two files should get created at first boot and not be required for booting

i'm just trying the example in the manpage...

those aren't actually errors

but then it quits afaics

er, sorry, didn't read the next line quickly enough

could I install GPU drivers on debbootstrap jail?

Retrofan: drivers, no. you can install userland software that *uses* the FreeBSD drm drivers, which may work since they're ported from Linux and the interface is compatible. you may need to create /dev nodes

Already created nodes, x11 and sound is working well.. but wine can't really know why it's fail

i would be surprised if Linux WINE works in Linuxlator to be honest, it does some very low-level things with memory management

oh

never know thta

*that

but it's a full userland..

is there a particular reason you're using Linux WINE rather the native FreeBSD version?

Bec I am running an linux browser

and I need a wrapper called pipelight to run windows plugins there

and honestly that wrapper never works well on freebsd

frankly at this point i would probably just use Linux, or at least a Linux VM

No, I am using that as my main one

browser

until the dev make a freebsd port

he promised of one very near.. but I will get back to the same problem, piplight never works on freebsd

could I create somewhat connection between native wine and the warpper?

in the linux jail

or something like modified version of wine?

Retrofan: which one?

i don't see your previous msgs

I successfully install Ubuntu jammy in jail, but can't run wine on it..

I get back Segmentation fault, I this related to needed gpu drivers or something?

this my first msgs

why don't you try native wine?

ty

Bec I am running an linux browser

and I need a wrapper called pipelight to run windows plugins there

and honestly that wrapper never works well on freebsd

like scratching my left ear with my right hand

heh, why?

it's hard to say if it's related to gpu drivers without understanding what caused you to jump to gpu drivers as a suggestion

Bec. I read somewhere that opengl in wine needs drivers

and I thinked linux jail uses it's own drivers

i guess i'd expect a better failure mode than random segfault if that was your problem

but it even fail to run any exe

kevans: ah, that happen when I use wine 7

but wine 10

linux jails also don't really run their own drivers? it's a freebsd kernel running whatever the host is running, with these applications running through a linux abi layer

says could not exec wineserver

but wineserver is their

*there

kevans: could I combine native and linuxjail apps like what happens with centos?

I also can't use centos; bec. flash is not working there, also the browser

winedbg says: "sh: 0: getcwd() failed: No such file or directory"

i don't think there's much luck in running wine in a linux compatibility layer

what's the browser?

fork of seamonkey

but supports Npapi

and JS very well

but fails in building on freebsd, needs tons of patches

what's the actual name of it so i can search it

and the dev promised for a freebsd release very soon

nxjoseph:https://github.com/wicknix/brass-monkey

It's originally for macos (Darwin) so I think it will be easy mission for bsd port

the patches is needed is for building system, but I have no time for that

github.com/chzigotzky/Web-Browsers-and-Suites-for-Linux-PPC latest builds for Linux is here

But be away from the newest one is so buggy due to the new UXP palemoon update

distributed linux build may work in a linux compat

i need to build the meta-port linux-rl9 to test it

yeah, I know

but upgrading centos to rl scares me

no 32-bit = no XMMS and alot of other stuff

Gtk1 even works there

also I don't know if flash will work there or not

there any problem not having a search domain set in resolv.conf? just nameservers and options edns0

kerneldove_: no, 'search' is entirely optional

ok that's what i thought. tyvm!!

ivy: Will a portable version of wine make things better.. I will try

it's the same error :(

I think there's no luck other that trying to make pipelight use native one

or rl9

Could I run freebsd software in the Linuxjail?

nope

but why that workers for centos and full userland no?

hm, should work if you install a freebsd userland in that jail, but it would probably conflict with the linux userland, why do you want to have it jailed instead of only running it via linuxemu?

ivy: using "poudriere image" now, but it's not the fastest approach

I mean debbootstrab jail

nimaje: I am using the full ubuntu userland

and want to connect software inside it to outside

is it really a jail? or did you create a userland and configured linuxemu to use that for linux software?

yeah a real jail

you can't run any software in it from outside

"Well, thanks ;) Quick explanation: A jail is just a jail, no matter what's running inside, and the FreeBSD kernel can execute both native and Linux binaries, no matter what."

here I go

let's test thta

*that

So I could run native wine there but only If succeed in mounting it's files there..

it doesn't read like you want a jail in your use case, how about setting compat.linux.emul_path?

Back

it doesn't read like you want a jail in your use case, how about setting compat.linux.emul_path?

nimaje: do you understand how linuxjails work?

*understood

it's not like centos

you can'r run any binary with it outside the jail

look like it's only works with rescue stuff.. need to mount /libexec

is that safe? or will mess up my system...

that's why I suggesting to ignore the jail part and only use linux emu, as you don't want it to be jailed away from your host (ok, the wiki does just a chroot, but that doesn't make a diffrence here)

Centos can't run browser nor flash so it's useless for that

I need to upgrade to rl9

where did I suggest centos? you have a linux userland, point compat.linux.emul_path to it

and I don't know if the flash works there.. and can't use Ubuntu flash in same time with another Linux userland

compat.linux.emul_path="/compat/ubuntu"

makes the jail to work

*gets

after that sysctl try to run a binary from inside that linux userland directly (so without doing a chroot or jail)

nimaje:ELF interpreter /lib64/ld-linux-x86-64.so.2 not found, error 2

but /compat/ubuntu/lib64/ld-linux-x86-64.so.2 is there?

yeah

ln -s ?

symlink it?

why would you symlink it? and to where? it is exactly where linux emu should look for it

system need to mounted to run

can't find paths

has anyone passthru'd a usb adapter to a bhyve vm?

as sanity check, sysctl compat.linux.emul_path

brandelf(1), perhaps

man it's /compat/ubuntu

already

this never works only in jails (mounted)

It will try to search for paths without "/compat/ubuntu"

it thinks thats a full system

*itself

only freebsd binaries could be run in the jail, not the opposite

My only question now is there is any danger of mounting libexec?

to ubuntu jail.. will it mess it up?

you're not listening (reading)

read only?

so I will need to copy freebsd needed libs inside then...

to be writable.. heh it's going weirder and weirder

lets try copying..

oh. heh works

hm, if the binary isn't branded as linux binary (ok, or sysv binary) it shouldn't work even in the jail, as linux emu isn't used and diffrences in the syscall interface would likely trip it up, but no idea why the path translation feature of linux emu doesn't work for you or how to debug it

need now to copy all the libs to make big program like wine to work XD

nimaje: Please read more about it.. that's the normal it's an full separated userland

it's not mine problem

if there was a path translation for stuff like that, every thing will mess up

it's have own bin and libs and other stuff will mess up with freebsd ones

so making native wine runs there, installing partial FreeBSD system over the Ubuntu user land that runs over FreeBSD kernel..

nevemind.. that's crazy I will find fix for linux wine or just no windows plugins..

or rl9

Retrofan: path translation is how linux emu works if you run linux binaries directly on the host, see man linux as the linux userland is under compat.linux.emul_path (which is /compat/linux by default (because linux- pkgs get installed there))

 but not for debbootstrab full userlan

*userland

yeah, you debootstrapped to somewhere else (probably a good idea to not conflict with stuff installed via pkg), that's why you have to change the compat.linux.emul_path sysctl, so that the path translation looks there

Did you done that?

on your machine

nimaje: please try it first

back

ok, tested, for some reason linux emu seems not to like absolute symlinks for ELF interpreter, when I changed /lib64/ld-linux-x86-64.so.2 to be a relative symlink it works, not sure why that bug wasn't already found

O_o

crazy

so it worked?

after using file on the ld-linux-x86-64.so.2 and then changing that to be a relative symlink instead of an absolute one it did work

but will break the jail

why would it break the jail? it would still point at the same file inside a chroot and the goal was to get it to work without a jail or chroot

nimaje: it still not working :(

ELF interpreter /lib64/ld-linux-x86-64.so.2 not found, error 2

after symlink

apparently, the first thing I need to get running for myself is znc

what did file tell you where /lib64/ld-linux-x86-64.so.2 pointed to and what was the ln command you used and what does file tell you now?

"/compat/ubuntu/usr/lib/x86_64-linux-gnu"

josephholsten: use IRCnow one

that's another absolute symlink, you should have used ../usr/lib/x86_64-linux-gnu

(but I would expect that one to work directly on the host)

not working

ah fixed it

and this nice

and the jail is not working XD

ah fixed the jail

nimaje: Thank you, look like I am still so stupid XD

BYE

I notice the smokeping linuxserver.io has github.com/heistp/irtt, but we don't have a package for that. Anyone used it on *BSD? Is it worth volunteering to maintain in ports?

dch: do you remember what foo was saying about python in podman during the jails call-for-testing? I'm getting "pkg: cached package python311-3.11.13: missing or size mismatch, fetching from remote" that's surprisingly resistant to my normal fixes

I'm guessing your containerfile has a pkg cache / data from an older layer

try nuking your /var/db/container entirely and let it repopulate

this would happen if you're using latest package repos i guess