yes it does

in usb config

not sure how to do it for sysctl

what's the device id? usbconfig -d ... dump_device_desc

specifically idVendor / idProduct

the realtek?

yes

usbconfig -d ugen3.2 dump_device_desc

what freebsd version?

386 13.1

hello I am in the chat now with this computer

as well as my laptop running natively freebsd

FreeBSD 13.2-RELEASE releng/13.2-n254617-525ecfdad597 GENERIC

sixpiece: as far as I can tell there's no driver for it

that usb device id isn't in the usb device list or in any of the builtin driver sources

ok so just return it?

what a headache

hopefully they take it back without questions

ok thank you it's a done deal correct?

as far as I know. my knowledge in this area is not perfect

there's a patch available supposedly

I mean is it worth doing or it's just a headache and try to find the one that works?

Title: Solved - Getting TP-LINK Archer T2U AC600 to Work | The FreeBSD Forums

do you see any hope there or just return it?

your device id doesn't match the one there?

ok thank you

what terrible sales people they told me I wouldn't have any comptability issues with this one they wanted to sell the more expensive gadget

but at least they are supposed to return it for 15 days

painful day today...

they also sold me two bad vga cords so I have to return that also and buy a new monitor at the second hand shop

thank you RhodiumToad

antranigv: Your sales pitch was very effective. :)

ted-ious :)))))

is there an equivalent to FreeBSD's du of the --max-depth option ?

I want to get just the size of a directory, not its sub-directories

du -hc `find . -depth 1 -type f`

. is the directory (. for current), f is file

that only calculates files in the current directory

I want the size of all directories ( and subdirectories ) but to only show the first level name size...ugh, not sure if that makes sense

so you DO want the subdirectories too?

ah

right now du -kc for example shows me dir1/subdir/subdir , I want it to just show dir1 = 8G

I want subdirectories to be calculated as part of the main dir, but not be shown

e.g.

200M  ,/a_file

20G dir1

exactly

hmmmm

du -d 0 -hc `find . -depth 1`

I got it

du -hd1 | sort -hr

last1: does that do the trick for you?

du -hd1 does only show entries for the directories, not the files in the current directory

we could use some long options in du(1)

-A looks like --apparent-size; -B / --block-size, -c / --total, -L / --dereference, -P / --no-dereference, at a minimum

This is so freaking strange. My WM exits when I close out a (seemingly) random terminal session.

last1: Do you have to stick with du or are alternatives an option?

last1: If not, then take a look at freshports.org/sysutils/erdtree

Title: FreshPorts -- sysutils/erdtree: Filesystem and disk-usage analysis tool

erdtree looks nice

thank you

I don't have to stick to du but an already existing utility is nice

if zfs status says "use -v to see errors" ... and then zfs status -v has no output below the errors line --- what might the error be, then?

its might be time to use ufs

concrete_houses: Are you the opposite of glass_houses? :)

Or are you by chance into dome houses?

last1 du -d 1

well, he doens't have to dress in basement

Hi, I was wondering if anyone is experiencing issues with firefox-esr lately, I just updated after installing RStudio and I am getting JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Can't find profile directory. The result is that Firefox does not connect to any website

well I ran pkg upgrade now and it is upgrading 400 packages

How often do you usually upgrade?

Seems that not as often as I shall

am I expected to do pkg upgrade from time to time>

?

I you do not want 400 package to be upgrade together, yes

s/I/If/

uskerine: are you on Quarterly or on latest?

I think I am in Quarterly, I am googling how to check taht

Is that like 13.1 vs 13.2?

no.

13.1-RELEASE

shall I upgrade the release?

("13.1-p8") Patch 8 is still current

ted-ious: all releases get the same versions of packages (so long as they build) but some people are not comfortable upgrading everything every day (which is what the "latest" pkg branch gives you) and prefer to go slowly… that's the "quarterly" pkg branch, which only fixes severe bugs and security issues.

by default, a freshly installed system (RELEASE, STABLE) uses "quarterly". A CURRENT system uses "latest"

quarterly is my thing

uskerine, Do you not update the OS per security or errata announcements?

😬

No I do not, unless I see something crazy around ssh

I use it as a desktop inside a LAN

Oh ok so freebsd tests groups of patches and releases them together.

so if I do pkg upgrade frmo time to time I am fine I guess

?

or am I expected to do something, like, when I shall move from 13.1 to somewhere else

or add a patch to that

pkg-upgrade fixed everything, both rstudio and firefox

quarterly is cut from the ports tree every quarter, so it is likely, that you get many updates after the quarter changes, but in the quarter only security fixes and bug fixes should get in there, so in the quarter you shouldn't get many updates, for latest the builders build the current ports tree

thanks nimaje

I have a FreeBSD(PowerPC architecture) kernel core dump. "file coredump" gives "ELF 32-bit MSB core file, PowerPC or cisco 4500, invalid version (embedded)". However, if I try to open it with "gdb coredump", then I get the "not in executable format: file format not recognized" error. Any suggestions, how to analyze this further? If I open this core dump with hex editor, then for some reason it also seems to

include some kernel ring buffer messages, content of the TLS cert files, etc

might have something to do with kTLS in the kernel. last time ive ran on a G5 i had to strip that kernel down quite a bit to get it going but that was long ago

eventually a faulty network card conflicted with the kernel and coredumped randomly from every 3-10

mrtnt: which FreeBSD version is that?

as for gdb and the file you might have to set "osabi"

use kgdb

meena: it should be FreeBSD 6

an oldie but a goodie

CmdLnKid: "show osabi" shows 'The current OS ABI is "FreeBSD/kernel".'

RhodiumToad: I tried with "kgdb coredump", but this does not seem the be different from gdb..

you're opening it on the same system that crashed or on a different system?

RhodiumToad: sorry that I didn't mention that. I am trying to open it on FreeBSD 13.1 for PPC. I am emulating it with qemu-system-ppc64

ah.

perhaps I should try with qemu-system-ppc32

Just a sanity check: Upgrading Samba 4.13 to 4.16, is that as simple as stopping the Samba service, uninstall 4.13 and install 4.16 and start the service? Or would I miss something crucial?

zfs snapshot ftw. remove and install seems to be enough. I don't use anything LDAP/AD here.

is there a way to set a comment/label next to a rule and/or table in ipfw ?

I mean, I see this in the manual: -b Only show the action and the comment, not the body of a rule. Implies -c.

but not how to add a comment

I tried: ipfw add 5 allow ip from 1.2.3.4 to any //comment but that errors out with: ipfw: pathname: //comment: No such file or directory

what version?

?

This is on 12.2 , but I mean, it's in the local manpage, it should be supported, no ?

oh, I think you need a space after the //

let me check

yeah that was it

damn, that's strict

thx!

if I plug my USB 2.0 stick into a USB 3.1 slot, then I get: "ugen0.3: <SanDisk Cruzer Glide> at usbus0, cfg=0 md=HOST spd=HIGH (480Mbps) pwr=ON (200mA)" and write speeds that are appropriate for a USB 2.0 stick

but if I plug it into a USB 2.0 slot, then I get: "ugen0.3: <SanDisk Cruzer Glide> at usbus0, cfg=0 md=HOST spd=FULL (12Mbps) pwr=ON (200mA)" and very slow write speeds (below 1 MB/sec)

why is that?

ugen.... does that present at a 'da' device?

a da* device gets created, yes; otherwise, I wouldn't have been able to test write speeds

spd=FULL is usb 1 speed, not usb 2

possibly the port is faulty or damaged

or your port isn't actually 2, but 1

that seems unlikely, unless you have a very old hub or something

RhodiumToad: no, I don't experience this kind of problem with my other USB 2.0 stick

could still be physical incompatibility between the stick and the port (just not making a good enough contact)

seriously?

have you tried a different usb2 port? or a usb2 hub or extension cable between them?

There are two USB 2.0 slots in the front and a couple of usb 2.0 slots in the back

could it be a ... an immitation usb stick? there are some sketchy fakes on amazon

if it works on usb2 in any port, then it has working usb2

if it's a fake, why do I get appropriate speeds (for a usb 2.0 stick) when it's plugged into a usb 3.1 slot?

fakery is usually about capacity - having the device report itself as much larger than it is

it's a usb 3.2 slot, actually

doesn't matter

but, as I said, I get normal usb 2.0 write speeds when that stick is plugged into that 3.2 port and very slow speeds when it's plugged into either one of those front usb 2.0 ports

that suggests some incompatibility between the stick and the internal hub, perhaps

I have some usb 2.0 slots in the back too, but never plugged the stick into any of those slots

USB is a terrible, terrible, terrible non-standard interface. Everyone who works on USB hardware and software ends up making hacks to make it _mostly_ work. It's likely some incompatibility between those particular bits of hardware.

i think what rwp meant to say is, usb is a standard, but it's just really bad

If implemented to the USB standards one will not have a working USB implementation. Therefore all implementations are mostly standard but then with a long list of modifications for practical functionality.

but it's like any hardware standard. just because there's a standard doesn't mean the standard is good, or that the hw manufacturer follows the spec faithfully

uefi for instance

Let's throw IPSEC in there too as one of those things that would be great if one could implement it from the standards documentation.

But yes UEFI is another one of those with an awful implementation history from vendors.

all the patents on firewire expired. maybe someone could pick that up again.

I don't myself know much about firewire other than that the interns at Intel who invented USB didn't understand it and so did their own thing.

But either way so much time has passed now that I think improving the USB situation with Firewire is no longer viable.

intel understood firewire perfectly well. they created usb to be cheaper and host-centric in their own interest.

(and succeeded)

firewire was a threat to intel because it didn't need a host.

Created for detachable disk drives, used for keyboards and mice, and now expanded to be a power distribution system!

well at first it was to get rid of rs-232

but then other stuff later, yes.

"Brother can you spare an ampere?" Only if you can negotiate the power protocol for it.

mice were an intended use-case for usb from the start

In the end we have this somewhat compatible collection of things that mostly works but sometimes does not. If you find a combination that does not work then avoid it. If it works then keep using it.

they're the main reason why there's a low-speed mode at all

As for Oleg it would definitely be good to try the rear slots too. Those will likely be a different chip driving them.

rwp: since those additional slots are located in the back, it would be hard for me to see if they are usb2 or usb3 slots. I know that usb3 slots look blue and usb2 slots look black. I wish the front part of this tower had more than one usb3 slot.

what FreeBSD command can I type to find out if the port is a usb 2.x or 3.x if I am not able to see that "blue" or "black" thing with my own eyes?

unfortunately there's no way to identify which physical port is which

the only reliable way to detect whether a port is usb3 is to plug a usb3 device in and see what speed it reports

if a usb3 device is plugged into a usb3 port, then the message has to be something like " usbus0, cfg=0 md=HOST spd=SUPER (5.0Gbps) pwr=ON (0mA)"?

yes

Hi! Has anyone here been able to run MedusaJS on FreeBSD? I have set up a vanilla FreeBSD 13.1 system with git and PostgreSQL 15.3, and I can create a new store, but when I try to run 'medusa develop', I get a lot of SQL errors, like "QueryFailedError: relation "staged_job" does not exist" and others.

Seems like I have forgotten something, but according to medudsjs.com, this is what I should need.

that sounds like it expected you to do some setup

Of course the guides are for Linux, but I figured it should be about the same for FreeBSD.

what was the _first_ error you got?

RhodiumToad: You're asking me? If so, that was the first error I got during the "Initializing defaults" stage.

what exact command did you run and what guide are you following?

Sorry, got distracted by a phone call.

Title: Install Medusa Backend | Medusa

That's the guide I'm following.

Really simple.

Maybe I should reinstall FreeBSD in a new VM.

I'm running this on AWS.

the error is nothing to do with freebsd

what did you do at step 2 of that guide?

I did exactly what it said. "medusa new my-medusa-store"

and what happened?

No, it's not FreeBSD specific. But when I'm googing around I don't get a lot of hits on the combination of FreeBSD with MedusaJS. So I figured I should ask here and maybe someone has a hint.

That step worked fine.

It's when I go to my-medusa-store and run 'medusa develop' that I get the SQL error messages.

it's expecting some database table to already exist, and it doesn't exist. so the problem is likely to be with an earlier step

what other errors were there?

Hm. When I installed PostgreSQL, I had to create a super user. That is the user I gave to Medusa. Maybe the problem is there somewhere.

I'm running it again, it takes several minutes.

how did you install postgresql?

pkg install postgresql15-server postgresql15-client

Then added postgresql_enable="YES" to /etc/rc.conf

And of course started it: `/usr/local/etc/rc.d/postgresql start`

Oh yeah, I had to run `sudo /usr/local/etc/rc.d/postgresql initdb` too

Then I started it as root and ran "CREATE USER pg WITH SUPERUSER CREATEDB CREATEROLE PASSWORD '...'"

Then I gave the pg user to MedusaJS when it asks for PostgreSQL credentials.

er, medusa requires superuser?

kevans: I assume not, but I don't know how else to do it. I gave it some other user, but apparently it requires permissions to create a database and whatnot, so I figured I could try with the superuser first.

This guide isn't very specific.

kevans: Do you have experiences with Medusa?

nope

kdehl: btw, service postgresql initdb and service postgresql start are the approved way to init and start pg

running the rc scripts directly is deprecated

i'd probably run away from it as soon as it wanted me to give it a db superuser; you get a database with all privileges on that database, and you'll like it, damnit

RhodiumToad: Oh. Okay.

did you have to specify what database to use?

RhodiumToad: I didn't. Which I found weird.

did it create one?

I found this guide: dev.to/yinks/how-to-setup-a-medusa-server-locally-44e3

Title: How to Setup A Medusa Server Locally - DEV Community

It tells me to create a database named medusa_admin.

I'm trying that.

RhodiumToad: Yes, it did create a database named something random.

those instructions look reasonably sane

note that it's creating a non-superuser :-)

Exactly. Which is nice.

Just strange that this info is not part of the official guide.

yeah now that one looks better

And whenever I ask around people are like "huh? You're not using whatever-favolrite-Linux-distro-of-the-week?! Then I can't help you."

RhodiumToad: okay, it looks like I don't have this problem with my two rear USB 2.0 ports. That stick gets "spd=HIGH (480Mbps)" messages when it's plugged into them. But for some unknown reason, it doesn't like my front USB 2.0 ports because it behaves like a USB1 device when plugged into them. However, as I mentioned earlier, my other USB 2.0 stick behaves identically no matter what USB 2.0 port

it's plugged into.

is the other stick from the same manufacturer?

no, the other stick is PNY and this one is Sandisk

could well be a bad interaction of some sort between the hub behind the front ports and the stick hardware or firmware

but the front usb 3.2 port is attached to a separate hub? I just don't know if usb2 and usb3 ports can ever be attached to the same hub.

the way that mixed usb3/usb2 hubs work is weird and I'm not sure I understand it

usb2 and usb3 are actually very different beasts

i.e. it's not just cranking up the clock speed, which is more or less how usb1 and usb2 relate

Does anyone seem to see that people claim that ports in FreeBSD are one of the best features, yet they are, in terms of servers, very hard to work with?

no

I've done cd /usr/ports/dns/bind918 ; make config ; make clean, I think 5 hours ago, and it's still going

which is infuriating to say the very least

it's probably compiling clang or something like that

was there another make in there? just doing make clean shouldn't take 5 hours

no other make in there

right

it's compiling texlive, x stuff, python stuff, and all kinds of other dependencies

make install not make clean, sorry.

ah

and it's just dependency after dependency and it just never ever seems to end

a lot of that will be to build docs

there are some shortcuts

because this is a two core, 768 MB RAM VM, I just sit and suffer

I don't see texlive on the list, though

x stuff?

I'm not joking when I say that

did you select any non-default options in the config?

I just wanted BIND 9.18 with 64 bit file support and a few other add-ins

no just hit enter on all of the blue make configs for the myriad of deps

ah, which add-ins exactly

let me check

`make install-missing-packages install` would probably really speed things up for you

OVERRIDECACHE, FIXED_RRSET, TCP_FASTOPEN, QUERYTRACE, LARGE_FILE, IDN. and DOCS

Maybe DNSTAP as well

and it's just been going and going in what I can only describe as dependency hell

Hi, I'm getting "Too many open files". This happens in a 13.2 Jail after running a Python program that open/close thowsands of files residing in a Windows filesystem mounted with smbfs_mount.

I would like to know how can I see what files are open, but fstat doesn't show that info, or I can't find which param should I pass.

martinrame: fstat on its own, as root, should list all open files

martinrame: sysctl kern.openfiles should give you the number of open files

martinrame: sysctl kern.maxfiles is the system-wide limit

signalblue: ok, I see that with all default options, bind918 has about 234 deps, including such horrors as llvm15, texlive, and a bunch of x11 client libs

horrors is an understatement

thank you for the confirmation

RhodiumToad: thanks, kern.openfiles shows 535, now I wold like to know how to list those files, because I can't see them with fstat.

freshports.org helped me verify that and yes, I dread the X stuff along with OpenJDK for things like Nginx

this is a server serving hundreds of critical DNS zones and I end up with all kinds of junk that I need to run an pkg autoremove on at the end

can't wait

signalblue: a few careful settings can prune the dependency list a lot

and don't get me wrong, I love FreeBSD but this is annoying.

Mmm. kern.maxfiles: 517260, kern.openfiles: 544, something is wrong.

martinrame: I could be wrong, but that sounds reasonable?

signalblue: no, of course not.

This is a Jail, can it be the host having the rest of files open?

what blocksize should i set for zfs. i'm guessing this is related to zvols

in proxmox

16k good enough or should i go larger?

Oh it is a jail, sorry I missed that part.

signalblue: 147 of those dependencies can be removed with one option.

(turn off DOXYGEN option of devel/protobuf-c)

RhodiumToad: which option, please do tell.

that gets rid of all the X stuff and llvm and texlive

which option though>

s/>/?

the DOXYGEN option of devel/protobuf-c, like I just said

ohh of that dependency

it's a bit to late for it at this point

I think I will just go out for a walk and let it do it's thing

honestly, given that llvm is involved, which takes nearly two hours to build on all 4 cores of my box, I'd abort it and start over

seriously?

unless it's already finished with llvm

Installing texlive-texmf-20230313... is where we stand

(pkg info should tell you what it installed so far)

yep

let me initiate a new SSH connection

texlive is nothing to build but it's a huge mass of files, so it takes forever to extract and stage

on Windows, TeX is the same way :)

things that build rustc, that's what ruins my day

oh yeah

and that new python horror

(py-poetry)

because that pulls in _both_ gcc12 and rust

looks like LLVM isn't installed according to pkg info

kill it then

you'll struggle to build llvm at all on two cores with only 768M

yes

if you have plenty of swap set up then it might work on one core, but then you're looking at, what, 6-8 hours of build time

killing it as we speak

you might create an /etc/make.conf with OPTIONS_UNSET_FORCE=DOCS to remove some more deps

not sure how much that helps in this case

not as much as I'd hoped unfortunately

just going to not do DOXYGEN and save the X, llvm and texlive stuff

ah, you need to force off MANPAGES as well as DOCS to really cut down the deps list, but that may be undesirable

that DOXYGEN option on protobuf-c is a real menace though