Hello, I would like some help finding a wired pcie Nic from here umart.com.au/pc-parts/networking/ne…rk-adapters/wired-pcie-adapters-972

The FreeBSD hardware list for Ethernet interfaces is about 10 years old and doesn’t help freebsd.org/releases/13.0R/hardware/#support

I bought a tp-link tg3468 but freebsd didn’t see the Nic.

linixmint OS worked fine.

Title: Wired PCIE Adapters | Network Adapters | Networking - Umart.com.au

Title: FreeBSD 13.0-RELEASE Hardware Notes | The FreeBSD Project

Test

when I do ldconfig -rv | grep ssl, I see several libssl's

how do I know which one is invoked when using -lssl ?

kipukun:

last1: -lssl as in when you link your program?

yes

so if I tell it: -L /usr/local/lib -lssl but in there I have like 4-5 libssl

i only have one; commonly there are many symlinks for those libraries

/usr/local/lib/libssl looks like it's nss; /usr/lib/libssl would be openssl in base

what does ldconfig -rv | grep ssl show for you

Title: dpaste/7f3g (Plain Text)

ah, so you only have the system libssl

you didn't install the one from ports

i have /usr/local/lib/libssl3.so but it's not in the output of ldconfig

and, it doesn't need to be if i recall

ok, I figured out what the issue was

great

hi

elgrande: sup?

coachella :)

and @u?

Are there any PCIe wired NICS compatible with FreeBSD? the FreeBSD hardware Ethernet Interfaces list is about 10 years outdated.

I'm looking for a non-FreeBSD router firmware compatible with NICs...any suggestions? I used DD-WRT ages ago.

OpenWRT?

allendale: thx, I'll check it out.

what hardware is it?

allendale: I had a TP-Link TG 3468.

FreeBSD can't handle Realtek chipsets and Intel chipsets are $999!

rly? It does seem both FreeBSD and OpenBSD has functioning realtek driver.

allendale: apparently not one. I tried one and FreeBSD doesn't see it, but Debian does. umart.com.au/pc-parts/networking/ne…rk-adapters/wired-pcie-adapters-972

Title: Wired PCIE Adapters | Network Adapters | Networking - Umart.com.au

there is realtek-re-kmod

Hallo Snetry

has the audio input been verified working after the audio discussion I took a glance at some days ago?

kevans: I was doing esp8266/esp32 dev on Windows before, but now that my main workstation is FreeBSD, I need to figure out how best to setup my devenv for it. I was looking at the ports and saw various one, but it was a little confusing as to exactly which ports do what (by a 30 second glance anyway, haven't dug into them yet).

kevans: I bascially just need to be able to plug in one of my devboards, read or write an image, and uh...that's about what's required. I'm probably going to use ESPHome, so I'll prebuild the image with the initial config for connecting to Wi-Fi (and possibly BLE, but would rather avoid that).

kevans: I saw arduinoOTA, but haven't dug into it yet. Also saw a meta-port for arduino tools? That's in addition to py-esptool. Might be moot if py-esptool will let me flash ESPHome and then just use Wi-Fi for the rest.

why I get full ram bar ? b) why its full despite 16/128gb c) what is 16"z" imgur.com/BdSSPm3.png

is is possible to create completely isolated BE, where i can perform downgrade ( 14_built_from_source -> 14_binary ), and completely new set of packages installed (default ones from freebsd repo ) ? 14_from_source box has ports built from source. so idea is to have completely new system

well, not quite a downgrade

Letiute: because empty RAM is wasted money

Letiute: iirc, 16z is a bug in htop that was already fixed?

meena so fbsd uses all the available ram? if so, good

so all is buffered?

yuripv I see. Doesn't loks like. but what does it mean now?

Title: Mem: 16.0Z on FreeBSD · Issue #1039 · htop-dev/htop · GitHub

1039 – two different spellings of "timezone" bugs.freebsd.org/bugzilla/show_bug.cgi?id=1039

angry_vincent: I think so, yes: see bectl (8) create -r

yuripv: I'm pretty, and sure, I've seen an update to fix that

Letiute: are you installing from latest or quarterly?

meena: i remember there was something about -r in bectl

nor did i really notice that it's htop…

htop supposedly fixed in ports by daf1fd1e557dfb3d1ca4f1a18494cda77d966644

so 16 z means 16 gb?

Fri Feb 24 17:30:19 2023

angry_vincent: how badly did you screw up your src build?

so, much likely fixed htop as pkg should be available

Letiute: no, it means Zetta Byte

meena: no, i did not screwed it up, but i want to create be with completely new set of pkg there.

meena but actually its 16gb used? right?

so I should consider z to e bg?

so I should consider z to be gb?

meena: reason, is that i have so many issues with drm-kmod, so i want to get rid of it completely and disable all the compat_linux and switch nvidia driver and Xorg, instead wayland. This is hard choice for me as i cannot stand X anymore too. is why the completely new BE. i can just wipe out all the offending things that make me sad and unhappy but i don't want to lose custom changes i made in this current system

good, this laptop has 2 cards

Letiute: no you shouldn't. it's a bug.

angry_vincent: that sounds like a good plan

Letiute: sixteen Zettabytes are a lot of bytes

meena so how do I know the correct ram used?

in htop

have your tried top instead?

our upgrading to a less buggy htop?

meena how to upgrade htop? I just installed it and it must have been the lastest in ports?

maybe, you are on quaterly branch

anyway, 2023Q1 also has fixed htop

in b54763d2a73da3a5f4147eaae79a38d11770cfc9

just installed freebsd. i selected NO for the time setup and chose Local Time and it showed the correct local time and i hit OK, but once installation is done and i've rebooted, the time is in local time but it's off by 8 hours :(

zostj see forums.freebsd.org/threads/time-is-utc-but-time-zone-is-local.83068

Title: Time Is UTC, but Time Zone Is Local | The FreeBSD Forums

hello, we hit a situation with LIST_FOREACH_SAFE() - in our case, a loop iteration (via a callback) can potentially free not just the current entry, but the *next* one (tvar in the macro definition). any interest in a variant that's safe against that too (along with a LIST_REMOVE() variant for it) ?

tsoome: thank you. i'll check it out

jlevon: open a bug and if it's trivial enough a GitHub pull request, if it's more complex a phabricator review. further discussion can go to #freebsd-dev

how well or bad does Dell server hardware run FreeBSD?

in "general".. or should I be looking at something else.

meena: thank you

No worries

I hope someone will produce vrtio hardware soon… then my purpose in life will come full circle

dude vrtio? =D

seems like I have been a bubble for the last decade

yes, that's what i mostly work on. I am not allowed to touch real hardware, because real hardware can explode

but anyway, yes, there is the idea of producing vrtio NICs or other PCIe components because it would mean you don't also have to write drivers for it because everybody already has them

what that in effect means, i believe, will be that OS vendors have to patch their virtio drivers to accommodate various buggy hardware implementation as opposed to only various buggy software implementations

geom: Command 'load' not available; try 'load' first.

huh?

angry_vincent how can I get the latest htop?

via pkg?

pkg install htop

well I did that yesterday

I still see 16z as ram

Letiute: you need to configure your repo to point at latest instead of quarterly

meena how?

quaterly branch has htop fixe, though

angry_vincent ok but why I am not getting those?

How can I be sure or get latest branch code via pkg ports

/etc/pkg/FreeBSD.conf change quaterly to latest

angry_vincent but that will get latest pkgs for all pkgs.. not just htop

isn't that bad?

meena  yuripv is 'latest' ok instead of 'quaterly'?

i have no opinion on that and cannot tell whether it's bad or good, it's on you to decide. i do not ue htop. i looked in 2023Q1 which is a quaterly branch and it has fixed htop

why you have no such version- i do not know

is "latest" as stable as "quaterly"?

i can't tell too

i never used quaterly.

angry_vincent oh you use latesst?

i use ports

angry_vincent ports has 2 options, latest and quarterly. Did I understood correct?

don't know why those 2 exist. why not just do latest all the time

if stability is same

quarterly gets updates quarterly, and otherwise just bug fixes

(and security fixes)

Letiute: i use main branch, HEAD

stability is such a volatile terminology

perhaps it has it's merits

i prefer to update when i feel i need too, by looking at updates but i prefer to get latet software

it's a matter of trade offs, where you, in reality cannot be sure about stability.

i don't use crazy software which changes abi every week or month

i'm so sick of updates.

Demosthenex: what's the problem?

oh just "stable" vs "latest" and "no commit for a year! omfg it's dead!"

The Churn is the enemy.

so, the default rpi images for the rpi4 come with UFS as a root file system

I'd prefer ZFS I think

What is the alternative to "systemctl" and "journalctl" as in linux for in freebsd

`service` is the command you are looking for

Turned on my screens when I woke up, and FreeBSD with Firefox running was averaging a load of 0.00, 0.00, and 0.00.

yourfate ok, what about journaltcl?

afaik there is no service for that, it just logs

yourfate how to see logs for a specific appplicaiton?

in logfiles, as linux used to

in /var/log/messages?

Linux does binary log files, FreeBSD (and every other Unix-like) uses plaintext files in /var/log/

Is the "message" file is what I should look at?

debdrup yourfate ok

hm, it seems like most guides to change the rpi image from ufs to zfs use another bsd machine to do it

/var/log/messages is the system log, whereas /etc/syslog.conf defines where other logging goes.

the only other BSD machine I have is in a datacenter ;s

If you're used to one Unix-like and want to know how another oen does it, consult with the unix rosetta stone: bhami.com/rosetta.html

Title: Rosetta Stone for Unix

I don't know whether that's been updated now that some Linux distributions change how they work but others haven't.

cgit.freebsd.org/src/tree/sys/dev/vt/vt_core.c#n1483 huh, this is apparently the code that's responsible for displaying the orb if one sets boot_mute="yes" or use boot -m.

Title: vt_core.c « vt « dev « sys - src - FreeBSD source tree

I was never quite able to figure it out on my own (I was looking in completely the wrong place), but the hackers@ mailing list suddenly got me there when I was least expecting it.

How to know if tor is running and on what port?

ps / sockstat

netstat as well.

Just now is the first time I've ever run 'netstat' with no options.

Didn't even know that was a thing.

IT's always been -an or -rn or -m or -i

I always miss the Linux -p option, which shows the process listening or connected.

(I tend to use tools that work everywhere, and sockstat is only on those of my systems running FreeBSD.)

That's what sockstat is for.

Yes, but not everywhere, and for most tools I can use the same tool for the same job.

I'd rather linux's CLI networking tools didn't suck so much.

Unix-likes are the same, except when they aren't. Hence why the unix rosetta stone exists.

They just keep making them worse too.

Linux used to have a decent ifconfig. . .

then they removed about half its functionality, and broke those features off into ethtool.

That's.. not really what happened.

Then they started with this "ip" nonsense.

meena debdrup ok. works!

linux networking is a frustration at best compared to what it used to be.

changing just to change cause some bright idea

Yeah, the linux network stack has always been shit. .. even according to Alan Cox, who wrote the IP stack.

I've dealt a lot with multi-homed servers, mulit-port NICs, etc. . .

Always a wild ride with linux.

=)

*LOVE* it when your ethX numbers would be different on every reboot.

What happened is that nettools,, which was the name of the project that contained ifconfig, route, et cetera ad nauseum, stopped being maintained because the prior maintainer got a case of life getting in the way, and nobody stepped up to the plate. So when the ABI of the Linux kernel changed (which Linus always claims isn't supposed to happen, but does regularly), it stopped working for certain things.

crtxreavr, lol i remember that

think I have been dealing with 3 ways to it in the last 10 years.... to no apparent reason.

CmdLnKid, what's worse was some of the horrible kludges put in place to "fix" that.

(regarding linux and netconfig work)

was so frustrated with it i about threw a laptop across the room needless to say the server would have been injured

That meant that anyone, and I do mean _anyone_ was free to step in and create an alternative. That alternative could've taken the form of a fork of nettools that was maintained, but _someone_ saw the oppotunity to do something completely different.

SuSE changed from ethX to eth<MAC address>

thats linux for ya

That was tedious, to say the least.

i can carve up a slackware linux based server all day but never felt the need to admin a suse server or client for that matter

tedious in either way

It's apparently just a happy accident a company, which exists because of support contracts, is the one that stands to benefit from the support contracts people are forced to switch to, in order to get help for new solution.

anyway ... this is bsd ;-)

This is truue.

i could dog on linux all day and not get anywhere... this is why i use bsd's now

and usually only freebsd

corps make that choice most the time

debdrup: thanks for the refresher on Linux ifconfig

debdrup: net-tools has been revivified as of the last... dunno, five or seven years ago

mason: yep, someone stepped up to maintain it afterwards.

Title: net-tools download | SourceForge.net

Hm, it's seen activity as recently as a couple months ago. Didn't realize it was quite that vigorously alive.

debdrup: I just did, same problem

_xor: iirc the arduino toolchain in-tree doesn't quite integrate esp support

kevans: Looking at it right now in fact.

kevans: I'm pretty sure it doesn't either, at least not currently...though it looks like maybe not too hard to compile?

I think we do have an independent espressif toolchain in ports

devel/xtensas-esp32-elf, but it's a couple of years old

Just duplicated it and in the process of updating it.

I'm updating it to the latest version (1.22.0 -> 1.25.0), which I hope satisfies the requirement for this other app to compile the firmware. I hope it checks the version using >= and not ==.

I wonder why a 1+1=2tb raid0 that I have shows 1.6TB capacity in `df -h` ?

Letiute: That does seem low. Is it partitioned? I'd expect 1.9 or something but not 1.6.

smells like MBR

ek not partition. I just installed freebsd with stripe raid 0 on 1tb+1tb

 df -h  Filesystem Size Used Avail Capacity Mounted on

zroot/ROOT/default 1.6T 302G 1.3T 19% /

Letiute: what does `gpart show` say

=> 40 1953525088 nvd0 GPT (932G)

          40 532480 1 efi (260M)

for both

Don't use striping.

kevans: Heh, I'm trying to figure out why it appears this toolchain is built using its own toolchain that in turn uses autotools (Makefile).

Makefiles are for make(1) though?

Unless I'm misunderstanding what you're saying.

kevans: It's also using yacc to generate some files from its own grammer. Mind you, this isn't for the resulting compiler, it's for the *installer* to get build the actual compiler.

It's parsers all the way down.

I meant that it's generating a required Makefile using autotools.

debdrup ya, dangerous?

domlaut so efi takes 260m?

_xor: oh. That seems.. strange.

Letiute: RAID0 aka striping is a good way to lose data.

it's also very fast at it

debdrup yes it is..

What's the point of going fast if it leads to data loss?

debdrup: but, speed!

true

Demosthenex: Don't butt speed.

I thought "hey, I have to backup anyway"

That's not the point, though.

who cares about data. look at all the webapps using non-ACID databases. who cares if we lose a few likes on pictures of cats.

so raid0 for OS + raid6 for backups

Letiute: always mirror boot drives.

it's a server (nominally)

Look, if you don't care about your data, you don't care about your data. That's your decision. And it's my decision to never help anyone who deliberately does dumb stuff because they think they're being clever.

I agree with you both.

raid0 should be avoided

you can care more about speed than the data. there are valid use cases for striping. it's only a generally bad idea, not a universally bad one

true

domlaut: we're gonna have to agree to disagree on this one.

so if I loose something, i replace drive, reinstall freebsd manually, then zfs.recieve() backups?

done?

Who knows.

no , I mean, is that the normal way?

If you insist on doing dumb stuff, I refuse to help.

assuming good backups, yes. the backups themselves might be corrupted

I am not asking you specifically only :)

you can have a drive in raid0 that seemingly works but messes up your data. can take a while before you notice

you can care more about speed than the data. there are valid use cases for striping. it's only a generally bad idea, not a universally bad one

domlaut oh.

so you not only have to rebuild *everything*, but you also need to know exactly when any striped drive failed

and that's... very, very hard

domlaut but scrub won't catch it? despite whatever raid version

theoretically, there are failure modes you can't catch in any other way than comparing known-good state bit-by-bit with whatever is on the drive

Read a book on ZFS.

Is that a prerequisit to talk in irc?

domlaut ok but you said "raid0 that seemingly works but messes up your data. can take a while before you notice"

anyone here bought any of the "zfs mastery" books?

well, I don't know for sure what you can and can't catch. I wouldn't assume a backup of a raid0 array is actually good/reliable

you'll have to dig deeper into that if you'd like. the only case I'd use striping for is ephemeral data that requires massive speed

debdrup don't get disturbed if things do not go your way. You gave this advice multiple times before too. I am not asking help from you specifically. You don't have to respond. Specially no need for "If you insist on doing dumb stuff, I refuse to help." I still respect for help you gave already though.

domlaut I se..

and there are use cases for that - eg. with distributed live video streaming

domlaut interesting

where you don't care when/if it fails, you replace it with another system, and any data after 1 minute is useless to you anyway

for example

right..

there's very *little* valid use cases for striping, but there are *still* valid ones (even with NVME drives that are very fast by themselves, and have largely eliminated a lot of what I've used striping for previously)

and you can imagine yourself in some "spy" film where security footage suddenly disappears

ok

yuripv: hopefully a system you'd never need to run striped :D

when installing openssl from source, it places a lib file in /usr/local/lib named libssl.so.1.1

but ldconfig doesn't see that lib, only if I rename it to libssl.so.11

have you tried to stat the lib?

I didn't, what would that tell me ?

it queries the file system

I think the issue is that ldconfig doesn't read libraries that have multiple dots at the end ? so libssl.so.1.1 is not ok but libssl.so.11 is ok

it's just a theory

have you executed stat on the file?

yes

what does it say?

99 19021646 -rwxr-xr-x 1 root wheel 38042728 697892 "Mar 16 23:02:26 2023" "Mar 16 20:38:05 2023" "Mar 16 20:38:05 2023" "Mar 16 20:38:05 2023" 32768 1472 0 libssl.so.1.1

     If no options are specified, the default format is "%d %i %Sp %l %Su %Sg %r %z \"%Sa\" \"%Sm\" \"%Sc\" \"%SB\" %k %b %#Xf %N".

ok...so how is that relevant to the issue at hand ?

why would you install openssl "from source"?

I knew this question would come

and I'd be told not to do it :)

exactly, install somewhere else to not clobber what was installed from ports, check what the port's Makefile does :)

not sure if everyone saw this from MWL:

Title: “Run Your Own Mail Server” chapter 0 – Michael W Lucas

sometimes I wonder what the W stands for, but not enough to try to actually find out

I think mostly to keep people from getting confused with the porn star that has the same name

you mean Michael Big D. Lucas ?

i hoped you knew that story spork.

So I cloned a ZFS dataset from a snap, and now I can't delete that snap (of course). I promote-d the cloned dataset, but that just moves the snap over under the clone. I still can't delete it. How do I get myself out of this?

gman999: just saw that on postfix-users@. Will be buying it.

i was just looking at that guy's site. i was considering getting his whole 2022 stack

MWL is one of the biggest assets of bsd land

i was a bit leery about buying the lot, never having read any of his books

was looking for a preview somewhere

Demosthenex: Maybe listen to an interview or two with him on BSD Now.

There are at least a few in the archives.

mason: frankly, given his prolific writing, i'm not too worried. ;]

ugh, it looks like the only answer is to `zfs send` and then rename. screwed myself on that one.

any opinions on borg backup? so i have a large raidz now, and i'm going to setup zrepl for layered zfs snapshots. gotta backup off the raid array though ;] so i was going to use borg to jbod for critical files, and my borg to offsite.

i don't see the sense in doing borg of my raidz to the raidz when zrepl snapshots should cover that use case in a more convenient manner

calling michaeldexter...

gman999: should we cue the doom music?

i've used borg for years, just pissy that borgmatic wants f*&(*&king YAML inputs

i dont use.. but i know dexter enjoys "zfs plus" questions...

sup

sup dex

hej

If I were to respond in Danish, are you sure you'd understand?

Unlikely. :(

Aw

Takk! or Tack, or Tak?

First is Norwegian, second is Swedish, and third is Danish, I think?

tussentack

Bless you!

Danish is a fun language; one of the best things about it is that there's a dialect where it's possible to say an entire sentence without using a single consonant.

"A æ u å a ø u i a å" meaning "I'm out on an island, out in a river"

Sweeet!

Right?

Pieci vliki vilka vilku!

the string "rw" repeated 67 times without newline is a valid MUMPS program

Wait, is that Danish or Hawaiian?

michaeldexter: i brought up backups, and someone mentioned you're passionate there ;]

haroldp: well, it might be Hawaiian too, I don't speak that - so I have to hope it's not something foul in Hawaiian. :P

Title: John Prine - Let's Talk Dirty in Hawaiian - German Afternoons - YouTube

Demosthenex: What's the question?

To back up or to not back up?

oh no, i was asking opinions about zrepl and borg

i'm a current borg and zrepl user. setting up new server is all

debating whether zrepl zfs snapshots is sufficient for local backup ON the raid. i will use borg to backup critical data off the raid and offsite

Demosthenex: I just do everything using snapshots. It's very efficient time-wise: github.com/ChibaPet/backup-zfs

Title: GitHub - ChibaPet/backup-zfs: Shell scripting to automate back-up via ZFS snapshots.

mason: yeah, i have zrepl already configured making snapshots ;] every 15 min for an hour, 72 hourly, 30 daily, 12 monthly

and if that zfs pool dies, i'm doing borg of critical files to another drive and offsite.

I quite like the notion of using old tape-retention strategies with ZFS snapshots.

i'm just... not used to making my raid also be my backup via snapshots

and yes, i know, i'm making backups off the raid

Demosthenex: That's where I differ - my back-ups are all themselves ZFS pools. Live backup is a ZFS pool, offsite is another ZFS pool.

well, i'm hoping to get closer to that soon

Demosthenex: Definitely *don't* have your raid "also" be back-ups. I'm saying, using what snapshots give you to ease the whole process.

my laptop is freebsd, new server is too. so i hope to do zrepl to send my laptop drive whole to the server in a sub fileset

They'll minimize how much you have to transfer to have coherent, reliable back-ups.

i've been using borg to send to local server. soon i can graduate that to zfs

I went from homebrew rsync-based back-ups to homebrew ZFS-based back-ups when I moved everything to ZFS.

It ended up being a fairly similar workflow.

(The rsync wasn't too wildly different from when everything I had fit onto tapes. I still miss tape.)

so i think borg's been quite good.

There's something to be said for that. If you understand and trust it, it's a good back-up system.

michaeldexter: that one's easy. :P

dedup, encrypt, compress. can offsite to an untrusted storage

The answer is, never not back up.

and while i like zfs snapshots, i'm just debating the backup history in snapshots vs borg

i know snapshots are completely easiest

Snapshots minimize what you back up, and they keep your back-ups on something you can scrub, more than making the process easier.

i'm also well aware that if i take a borg backup of the raidz TO the raidz, it's not adding anything

debdrup, michaeldexter: my answer is: eh 🤷🏻‍♀️

hmm, i put i386 iso on a usb stick with dd and nothing is recognized.. before i by mistake put the amd64 iso on the usb stick and it got recognized just later boot coplained it is not a i686 image..

so not sure what went wrong with the i386 iso/image..

should we use disk-by-id instead of /dev/sda etc type of names? I heard if you reboot and replace drives and the name changes, zfs can  cause problems?

Sorry I could not chat. Try to catch me later. :(

is it possible to build a release image without being root? Right now it looks like you need to be root to do the make install{world, kernel, distribution}

net.local.stream.(recv|send)space is a number of bytes ? And why can I seemingly use setsockopt to go over that amount ?

How can I identify which process is creating files in a directory?

My /tmp is filling up, files are owned by www, but there are both apache and svn processes running as that user

have you tried something like: top -m io -o total

Figured it out. Started auditd, then "auditreduce -z <jailname> < /var/audit/current | praudit > audit.log" for a little while, and looked at the result. It's the apache process, indeed.

Demosthenex: The tools generally come down to zrepl, zxfer, and sanoid/syncoid, or whatever's built-in, such as with TrueNAS. You definitely want some snapshot retention to mitigate human error or ransomware, and "no array is a backup". Ideally replicate to a second pool or two.