I have all of my wifi setup effectively the same.

ISPs put such weird constraints on things

but there's not enough ipv6 address space!

Think of those poor neutrinos coming from the sun who will be forced to use NAT!

edenist: hahaha

hmm, bit concerned that the linuxulator-steam-utils readme just mentions to forget about anything working properly on an AMD GPU

so iwm doesn't seem to reliably associate with my wireless network, not sure how to troubleshoot this

is this on the new laptop?

yeah, the T495

I'm gonna see if the scan_ssid=1 bit that bsdinstall put in the network block has anything to do with it having trouble, because I've never had that flag set on anything else

also, update from yesterday, I have just decided to go the whole fresh installation route instead of just trying to clone the install

I'll just end up cloning over my home directory, I think

if you've got customizations based on hardware sounds like a sane way to do it. will thank yourself in the future, heh

well if I cloned the install, I think the main things I'd have to change is loading amdgpu instead of i915, and disabling wifibox to use native wifi instead

also disabling coretemp in favor of amdtemp

but doing a fresh install means this hasn't been touched by pkgbase which gives me the opportunity to rethink whether or not I want to continue using that

compiling kernel + world on this Ryzen 5 PRO 3500U should be a lot less painful than it would be on the i3-10110U of the other ThinkPad

my two desktop/laptop installs are using sandybridge/ivybridge, so I'm definitely not doing world fast on those, lol

I remember not having much trouble with either world or kernel on my ivybridge i3 back in the 10-STABLE and 11-STABLE days, but it sounds like LLVM/clang has bloated considerably

i cant get ethernet to work

or wifi for that matter

its loading the drivers as far as i know but it cant find em0

only igb0

and when i try to use igb0 it says something to the effect of "no link"

thegman: you should use supported hardware

i dont know what the name is of this specific ethernet hardware

il check

if you want specific hardware support, contribute it

is there no support for the "Intel Corperation I211 (rev 03)"

i thought that was in the em driver

freebsd.org/releases/14.0R/hardware according to this the i211 gigabit ethernet controller is supported

Title: FreeBSD 14.0 Hardware Notes | The FreeBSD Project

em(4) does support i211 and anecdotally it works on my apu2

"ifconfig em0" shows nothing?

im pretty sure i tried that

if memory serves ifconfig without any arguments shows igb0 and lo

do you have something weird in your rc.conf?

its a fresh install

wait, my bad,

but i think i did set the boot delay in there

boot delay is a loader.conf, iirc. "dmesg | grep em0"?

it was also doing this when i went into a live shell from a usb

i guess it was in loader.conf

im still new to all the configuration files

i saved the output of dmesg and put it in /

those are basically the main ones I think for this kind of problem.

but i cant get zfs to show me the file so i might have to reboot to re acquire it

root can use dmesg command at any time

i know but im not booted into freebsd right now

its on a dual boot

nevermind i got it to work

what was wrong with it?

grep -i em0 dmesg.log shows nothing

i mean i got zfs to import the pool correctly so i could get the log file

wow, that's amazing. from my experience, if supported hardware doesn't show up in dmesg, the controller is physically dead.

i dont know how it could be dead

im using it right now

and ive only had this motherboard for around 3 years

that's why I'm amazed, obviously not dead

"grep 211"?

igb0: <Intel(R) I211 (Copper)> port 0xd000-0xd01f mem 0xfc600000-0xfc61ffff,0xfc620000-0xfc623fff at device 0.0 on pci6

thats the only line with that

okay, so, igb(4) is using your I211, and igb0 is what you want to be configuring

when i was using the installer dhcp wouldnt work with igb0

i tried to do it manually but whenever i put in the router as 192.168.1.1 it just resets the value i put in there

what is the output of "dhclient igb0"? that should get you where you want to go. if your router expects dhcp then you need to use that.

ok il boot into freebsd to check

output of ifconfig igb0 is on bpa.st/OA5Q

Title: View paste OA5Q

perfect, hit it with a "dhclient igb0"

i think the last time i did that it said "no link........."

how do the lights on the nic look?

isnt that just the led next to the spot the ethernet cable goes in

yes, there are two of them

theres a orange and a green one

one of them is orange

the other one is flashing green every now and then

sounds about right

so, when everything sounds about right, but it still won't work, I use "service netif restart".

il try that

you know, I just realized he can't have the same inet and broadcast

restarting netif didnt work

my fault, I realized after a moment that you can't have the same inet and broadcast.

so, I am now thinking you should "ifconfig igb0 delete" to remove the addresses it thinks it has

and if rc.conf is putting those addresses there, you should remove that

once you don't have those crazy addresses in there, "dhclient igb0" is how you ask for a dhcp lease on igb0

right now rc.conf has ifconfig_igb0="inet 192.168.1.17 netmask 0xffffffff"

and defaultrouter="192.168.1.1"

that is what is causing your problem

you cannot have that and have dhcp, they're trying to do the same thing

i set it manually since dhcp wasnt working in the installer

it said it could not acquire a dhcp lease

is that configuration working?

the one in rc.conf you mean

thats just what ive had since installing it

yes

did it work when you installed it?

nope

internet hasent worked in the installer or after installing it or in a live shell

that makes sense. I am under the impression that you have to use dhcp if that's what the router expects.

i have noticed spectrum doesnt play nice without dhcp

come to think of it it never plays nice

honestly man, I use usb tethering. I've lived alone for nine years and never wanted to pay them for their garbage

never paid them a penny

i dont pay them either its just the internet im stuck with

if i had a choice in this matter it would be literally anything except for spectrum

they dont even let you do port forwarding from a browser

you have to install their gosh dang phone app

well, when I go elsewhere, dhclient works if I'm not trying to do something else

having addresses specified in rc.conf is trying to do something else

should i just not do anything related to ifconfig in rc.conf

I don't, except wifi, because I'm always doing different stuff wherever I am

il see what happens after i comment out all the ifconfig lines

you can say "ifconfig_em0="DHCP"" in rc.conf if you want

ok il try that then

but I would get it working with nothing in there first and then add that once you're sure it works

ok

i tried it with and without ifconfig_em0="DHCP"

and i tried doing ifconfig igb0 delete

didnt work

what did "ifconfig igb0" say?

the same thing im pretty sure

actually

when i removed those lines and when i was trying the DHCP thing i think it just didnt have any information about the ip in it

also, you can use your phone browser to get on this channel at web.libera.chat

so you don't have to keep rebooting

il see if theres a phone available

it needs to not have the addresses in it to get a dhcp lease

the phone thing doesnt want to comply

it's also possible that the daemons don't like you

that is a possibility

sometimes nothing works, and sometimes it's like, did benevolent hackers upgrade my csh configuration?

apparently i cant join #freebsd on web.libera.chat without registration

if it says no carrier it's a physical problem

the only thing i can think of is the ethernet cable being in the wrong slot

or some other problem, not having to do with addresses

and i dont see how that could happen since im talking on irc right now

he's saying it works when he boots to $otheros

dunno

I think I can figure out out. I think that igb0 is his interface, and as long as nothing is trying to specify an alternative and conflicting configuration, dhclient will work.

and we've established that rc.conf is, in fact, specifying an alternative and conflicting configuration

but i cleared all network stuff from rc.conf and it still didnt work (after a reboot)

well, we've made some progress and I still have hope

as far as i know, no carrier always indicates the lack of any link, so at a layer below any addressing

that's very good information

il try dhcp again

but i think it doesnt know about the router

is there something i should set in /etc/rc.conf to make it actually use the router

the green light means it is talking to the router

i know but it doesnt appear to actually talk to the router when im on freebsd

I think dhclient will work if nothing else is going on

wow

i was testing in the livecd and dhclient worked when i switched the slot the ethernet cable was using

the strange thing is that slot doesnt work here on gentoo linux

the other slot is igb1?

if you're judging "works/doesn't work" based on internet access, your internet access is separate from whether the interface has a link

you can have an interface with a link but with wrong config and you won't have access

i dont see how i would have misconfigured it in this case

i just booted into the livecd

im assuming theres something i was supposed to do that i didnt know about

im pretty sure the other slot is for lan or something like that

i dont know all that much about networking

but theres a picture of a box with a line going out to 2 more boxes on that one

the motherboard manual identifies that as "LAN2"

savages, numbers start at 0

apparently the fools at asus dont know that

actually i suspect those guys are smarter than me, just being silly

i would respect them more if they didnt have that "armory crate" download button in their bios

i really believe seabios is a thousand years ahead of any efi implementation in simplicity, maintainability, reliability and security

i grit my teeth when i see "legacy bios" written, because they have a 16-bit bios implementation that can start massive amounts of cores, nvme storage, gpus, everything.

and it has no application interface to run malware on, no network interfaces to even attack in the first place

i was considering installing a non garbage bios on this motherboard but i dont know if this one is supported

and id probably end up bricking the motherboard

also seabios is instant and you cant argue with that

im not qualified to port something like that, so i only buy hardware where someone else is maintaining it

maybe this is why ethernet wasnt working

according to the asus website

it lists "Realtek RTL8125-CG 2.5G LAN" then "Intel I211-AT"

and it says 2.5G on the slot that wasnt working on freebsd

so i guess i shoulda been using re0

then again re0 wasnt showing up

wow thats what jmnbtslsQE was hinting at, there was no link because that was the wrong interface

hardware issue

well, link issue

either way il work on it tomorrow

so if I want /tmp to be tmpfs, but ZFS has zroot/tmp set to mount to /tmp, would ZFS override a tmpfs mount if I add it to fstab? like would I have to delete the ZFS dataset?

or you could just unset the property

Hello, I am trying to run a program from an account that I have logged into after logging into another account in Xfce. In other words I log into Xfce, then I open a terminal and log into another account. Now I am getting an issue that X cannot Open the Display. I get this "(zenity:69161): Gtk-WARNING **: 01:01:53.391: cannot open display: :0.0" The DISPLAY variable is exactly the same in both instances that I am logged into. What can

be the problem? Thanks.

still having issues with getting this wireless card to reliably connect. I'm sitting only a few feet from this router, and wpa_cli status reports that the wpa_state is "SCANNING"

finally just connected after several minutes

tm512: is this regarding the t495?

yeah, with its Intel AC 9260 wireless chipset

yeah, it barely even works on linux ^^; on linux i had pass an option or two to not have the card constantly crash

er, to the kernel module

kinda weird since I'm using the old, presumably more mature, iwm driver currently. I may give iwlwifi a spin

gimme two secs and i'll get the stuff i had to pass, might be somewhat applicable to freebsd if you're lucky

right, so on linux had to pass power_scheme=1 to the iwlmvm kernel module

*i had

I don't think that's even a driver here on FreeBSD, it's iwm and then iwlwifi is the next-gen one which I think is the only option for some of the newer chipsets

with the 9260 being kind of in that transitionary period where it has support from both

dunno if iwm even comes from Linux, but iwlwifi does, at least in-part

iwlmvm is part of iwlwifi, its just that it's broken off as a separate module

quick grep in the freebsd src dir does show that the mvm stuff is in there at least

worst-case scenario I'll just set up wifibox again though it kinda sucks to need that. had to use it with the E14 I'm replacing, because rtw88 on FreeBSD was completely busted, causing kernel panics when any meaningful amount of network traffic passed through the card (i.e. anything beyond, say, pinging)

usb tethering works really nicely

kinda unfortunate that I've gotta hustle on returning the E14, kept forgetting to send bug reports regarding rtw88

from what I recall it was pretty weird. the kernel panic showed that the process crashing was rtw88, but the backtrace showed that it was crashing in drm-related functions, as in, the GPU driver stuff

tm512: there's a power_scheme sysctl tunable for iwm. try setting that to 1 and see what happens

I'll look into that. any idea what it actually controls?

oh, power management, so I guess 1 might make it more responsive but at the cost of increased power consumption?

hrm

it disables power saving completely actually

unfortunately it's the only thing that worked for me

I wish you a good workflow for today, my FreeBSD 14 installations works excellent so far!

I'm trying to get some linux software running under the linux emulation. There is a whole chain of ugly scripts calling each other which finally ends in checking the name -m arch which we report as amd64 but which I'm guessing linux reports as x86_64, how do people handle this without modifiy the scripts?

The most complete solution for all your business needs is the UNIX enterprise software.

i was just checking dmesg and the realtek adapter is showing up as "none1"

thegman: have you checked out the realtek website to pul that driver? for my new machine i had to go that route to get my realtek 2.5gig onboard nic to show up

i was about to install realtek-re-kmod freshports.org/net/realtek-re-kmod

Title: FreshPorts -- net/realtek-re-kmod: Kernel driver for Realtek PCIe Ethernet Controllers

im pretty sure its the right one

installing that driver fixed it

is it normal for a zfs snapshot to take up a different amount of room on the backup box as the production box? production dataset shows 344G used. backup box shows 235G used. pretty big discrepancy

I wonder if this is an ashift thing, or something in that neighborhood. Digging down, as one example, the same mp4 file on the backup box is 115M, and on prod it's 182M.

It so happens that this dataset is shared via ZFS over NFS, and all the writing is happening from an application on a linux box (immich, a self-hosted google photos alternative).

Comression?

+p

scoobybejesus: it depends on the other snapshots also

Greetings. I have a Lenovo V330-15IKB laptop. I have FreeBSD 14.0-RELEASE-p5 on it. Usually after a few minutes, the laptop's fan starts working and it would not stop. Why is it like this?

How can I control it? Is there even any solution?

powerd

i have compression in both. in prod it's lz4, and i'm not sure if it's lz4 or zstd on the backup box. they are both only in the 1.01 - 1.03x range.

i started deleting some older snapshots, even though they are barely referring to any unique data, hoping i'd find that space was freed up. that didn't work. files are straight up taking up different room on each device.

i thought i was doing something smart at the time (and maybe/hopefully it is for read speeds), but this seems to be a likely culprit... on prod, my recordsize is 1M for this dataset. on the backup box, the recordsize is 128K

the software automatically transcodes everything, creating thumbnails and whatnot. i'm guessing the 1M thumbnails are able to fit into 128k on the backup box. though this doesn't explain (in my mind) why a 115M video takes up 182M on the prod box

interesting problem/issue, one way or another

scoobybejesus, Video and Audio compressed won't be able to compress any further. So if the compression is lz4 or zstd and on for both main and backup I would expect that neither would be able to compress more.

I would expect that if compression is off that though the A-V files won't compress but other files would compress.

I always have compression on with either lz4 or zstd.

indeed, indeed. I was just pointing out there compression doesn't seem to reflect what is different between prod and backup

I often run this command to give me a showing of interesting attributes for datasets.

zfs list -o compression,compressratio,refquota,quota,used,usedds,usedsnap,refer,creation,name

scoobybejesus: how are you determining that the example mp4 file has different sizes?

if you're looking at the filesystem and it's different sizes it's just not the same data, unless i'm missing something

(the file different sizes i mean)

but either way, one thing that can cause a different in dataset size in this situation with a 1M recordsize is if you had compression disabled at one point and had many files written without compression, filling part of that 1M with zeros. then when you send/recv with compression that gets compressed away

i used dust, but i could use ls

i haven't changed the properties of the datasets since inception

OK

Boy, the dependency list for mtr seems rediculous.

Title: View paste NV2A

I guess there's an mtr-nox11 package.

Interesting. `ls` shows the same 115M in both prod and backup. `dust` on prod shows 182M for the same file and 115M on backup. So, yeah, this 115M file is the same in both places, but somehow `dust` (in addition to zfs, of course) knows it's taking up 182M on disk on prod

`dust` is a newer `du`. `du` shows 186393 on prod and 118209 on backup

so perhaps these media files in some cases are sparse files, and so the OS/FS thinks it is intelligently using space? or maybe something else?

but i would think sparse files would benefit from compression. hmmmmm

if they're sparse on the source they should be sparse on the destination i think

github.com/openzfs/zfs/discussions/11533#discussioncomment-315578 talks about some random items that could take up blocks in addition to the file itself. not sure if it's applicable here

Title: Why does stat(2) .st_blocks reflect ZFS compression? · openzfs/zfs · Discussion #11533 · GitHub

(st_blocks is what du and dust are using)

so, dunno honestly

interesting. when i copy the file. the resulting copy is only 115M. apparently cp is good at identifying a sparse file and only copying the actual data

i think you are right that cp does copy sparse files properly. but i don't see how it's related to sparse files. i think that would involve entire (zfs) records that are zero

and besides, if you have compression enabled on both sides and have always had, then sparseness should not make a large difference

before copying, the file takes up space of 182M, and after copying it takes 115M. maybe there is something going on that's related to sparseness or has the same effect

the only thing I can come up with is that writing to an NFS-mounted filesystem is causing sparseness to not being handled property

the guy who founded the project linked me to a stackoverflow answer on the difference between `ls` and `du` instead of telling me whether he experiences the same thing. frustrating