Does FreeBSD have anything similar to Linux's /proc/pressure/memory or MacOS's NOTE_VM_PRESSURE, where a program can be notified when the system/cgroup is low on memory

skyenet: not that I'm aware of

clod: hmm, is there some API that could still give similar information? perhaps via polling

does FreeBSD have any kind of protection that prevents it from showing something as associated with a privatekey in a file?

closest option according to searhes are vm.stats.vm.v_free_count and things like that

skyenet: polling with sysctl

sysctl vm.stats.vm.v_free_count vm.stats.vm.v_free_min vm.stats.vm.v_free_target vm.stats.vm.v_inactive_count

nm, i realize what i did wrong

i am planning to post my issue with my sound card to the FreeBSD forums. what information should i include?

at minimum release number (uname) and your chipset type. also kldstat i'd imagine to show what sound drivers the kernel has loaded

glenny2: putting /var/run/dmesg.boot on dmesgd.nycbug.org/dmesgd and linking that would also be a good idea

patch to -wireless

rwp, I have tried ``camcontrol apm da0 -l 254'' -- no effect.

thoughts?

installing arcanist

so I got a null modem adapter and I still cant seem to figure out how to get serial to work

I see the interface server side

client side I have a DB9 M/F NM adapter plugged into a through cable plugged into my laptop usb port

I dont see anything in dmesg

and ls /dev/tty* shows no changes

this is starting to feel helpless

polarian, Have you tried the mailinsg lists, the official issue tracker?

ant-x: this isnt a bug, this is a me issue

and I dont like using mailing lists as I make a fool of myself

lol

What? Did you develop all the hardware and components involved :-?

no

the point is, I cant figure out how to get it working

It may not be a 100% you-issue, then...

Better make fun of yourself and get it working than vice versa, then.

serial port issues are crossed RX-TX lines, 80% of the time

psycorama: elaborate, you want to cross them so that RX is connected to the TX on the other end, and vice versa

do you mean that you have crossed it again so that you have RX crossed to RX?

no matter how much attention i pay for wiring it up. 80% of the time i'll have to cross it again

usually you'll net to connect RX to TX (as both sides are labeled as seeing themself)

It worked fine when I was using usb ttl cables

soldered together RX-TX TX-RX

that worked flawlessly, issue is I cant decrypt a FDE using that

the null-modem adapter, does it expect rs-232 signal levels?

but your real problems seems to be, the adapter not being recognised.

psycorama: JCOM port on motherboard, connected to it is startech.com/en-us/cables/plate9m16 which shows up within the server dmesg, so success there

then connected to that is a startech.com/en-us/cables/nm9mf

with a generic through DB9 --> USB cable

I cant for the life of me plug this into any device for serial access

I wonder if I plugged in the adapter the wrong way to the motherboard

there was no marking on the motherboard which is pin 1

so I just yoloed and plugged it in one way

and it showed up

So I assumed it was plugged in correctly

maybe I should flip it around and see?

wait dumb question I should totally do that xD

what can possibly go wrong? :)

I blow the motherboard :p

that might be expensive

but apart from that I honestly dont know what more I could try

the connector plugged into the motherboards JCOM port is totally passiv. so the OS should see the port regardless of anything plugged into it.

psycorama: weirdly enough it doesn't...

ttyu0 didnt exist until I plugged into the JCOM port

or maybe im just blind

but I doubt it

but yeah I assumed it would be passive, its just an adapter

there might be some shenanigans whith a sense pin against the gnd/shield.

lemme flip it around and see what happens

is there documentation for the board?

nothing about the JCOM port funny enough

and honestly I forgot which board it is

and im not stripping the server to find out

if I dont need to ofc

yeah, understandable

polarian: whats the serial problem?

f451: well the problem is nothing shows up on the client side

is the client side freebsd?

yes

and the server side is freebsd

whats the cable? null modem or ttl <> usb

listed above

i cant see it as i had to reboot

startech.com/en-us/cables/plate9m16 --> startech.com/en-us/cables/nm9mf --> some generic through DB9 --> USB cable

ok lemme look

dumb q but is the port turned on in the bios?

f451: I dont see anything about it in the bios lol

however: bpa.st/UHPA

and theres a /dev/ttyu0

server side

does connecting RX-TX on server (and client) give you echo on the consoles?

ok - that bpa.st/UHPA thats server?

and it's /dev/ttyu0 not /dev/ttyU0

?

f451: yes

U0 would be usb based

afaik

ls -lah /dev/cua* ?

no - U) the caps means carrier is high

U in caps in that context means the carrier is high

ah

makes sense

i have 2x serial ports connected /dev/cuaU0 /dev/cuaU1 both with init and lock

wait did you want this on the server or clie nt

this is still server side jsyn

I said to psy that I would flip around the connector on the mobo

see if it helps

yes those things i posted were from the client part

there is no pin on the board

f451: well the server has cua and tty

let me check my server side

im currently mid make installworld, in 1 min I will pull the server apart and flip the connector on the mobo

see if that does anything

theres no pin markings so I just plugged it in one way and hoped it would work

and then forgot about it while waiting for the nullmodem adapter

hehee

if that makes no difference then im clueless

I dont see much more what I can try, server side looks fine

what i was going to suggest was if, on booting you saw, once removed, still, the 16550A uart

that card thing

its an adapter

so it might still show up

if you pulled that do you still see 16550A when the server boots in dmesg

but I dont remember seeing a ttyu* without having the header in use

so maybe theres a pin there to detect if a adapter is connected?

I dont know how JCOM works so yeah

f451: will pull it boot it and then I will flip it around then

id like to make sure its the card being detected which is why im saying pull it

yeah

btw on a rpi1b 'server' - the ttl end of the serial is plugged into rpi4 gpio pins to get console, the other end into a usb port on the rpi1

has /dev/cuaU0

same with /dev/ttyU0

i guess the rpi1 is also a 'client' thinking about it

brb

f451: I cant even decrypt my server at this point as all the keyboards are defective lol

congrats

still there with flipping the connector to the mobo

theres also uart: ns8250: UART FCR is broken (0x1)

but this afaik means nothing

I read in some bug that this wont cause any issues

nope still nothing

lemme dmidecode the mobo and see if I can find anything in the docs but im clueless on what to do

dmesg: bpa.st/AIIQ

interfaces

this is with NOTHING plugged into the jcom port

there is a keyboard ethernet and power, nothing else plugged into the mobo

lemme dmidecode and see if I can find the mobo

anyone is using Anubis ? I've added -difficulty 4 to the rc.conf but I'm still getting 2 ... any idea?

I've set anubis_args="-difficulty 4 -target "0.0.0.0" -bind "192.168.0.50:8923" -hs512-secret xxx -cookie-dynamic-domain"

(there is an HAProxy in front)

f451: omfg, maybe both cables are nullmodems, but wait no that means one would counteract the other

so I dmidecoded and found the mobo, and found this

the pinout for it

the pinout for the plate adapter is a through adapter, cool

thats the rs232 pinout

2 is receieved and 3 is transmit

so a nullmodem will cross 2 and 3 over

and the pinout from the mobo shows pin 2 is rx and pin 3 is tx

so with through, you would have rx-rx tx-tx

no pinout but if startech is to be trusted, the rx/tx pins are flipped

wait I wonder if it matters what way around the M/F is

because its plugged in F/M

maybe this is going to need some electronics testing

using a multimeter and some paperclips

back

polarian: in an x2 box theres a standard serial port, it's that old

i use this:

Hi,I don't suppose anyone could tell me how to list the open files/locks an nfs client has on a freebsd based nfs server?

We've a freebsd nfs server and a load of linux clients. The server shows nfsd at 5580% cpu, 10-15k of open files and a few hundred locks for most of the clients according to nfsdumpstate. The linux nfs clients don't seem to think they have any open files/locks.

hi rp, id be interested in that too

I'm open to any pointers on how to debug something like this...

Honestly I would probably start by rebooting the nfsd server so that it would start from a known good state. nfsdumpstate immediately after boot. Then watch to see if there is some resource leak happening.

The server does this every Friday, we reboot the clients every Monday. I suspect it builds to the point it breaks

so yes, I can reboot it but that isn't really going to help me

Hmm... Well... I have no more ideas. Good hunting!

thanks! :)

Reboot more often? :-)

I thought bsd was better than that! (he says laughing) :)

It should be, but sadly the lack of resources sometimes shows

I know the feeling. FWIW I'm from the Yocto Project and the server in question sits in the middle of our CI which is currently dead in the water due to this :/

RP: Which NFS protocol are you using? That might matter for this.

Ooh... And you're not running straight FreeBSD on your NFS server.

(At least, that's what I'm gathering from "freebsd based".)

mason: nfs 4.2

There are a couple knobs noted here, at least one lock-related: man.freebsd.org/cgi/man.cgi?query=nfsv4(4)

mason: it should be a standard freebsd install of 14.4

mason: I hadn't spotted that page, I'll have a better read. I think we just use file locking, not byte range locking though

I'd be curious what locks are being held. Might be possible to explore with DDB. docs.freebsd.org/en/books/developer…kerneldebug/#kerneldebug-online-ddb

Unfortunately this system is doing other things I really don't want to interrupt so that kind of debugging may not be a good idea. Is there any other way to find out the locks being held?

RP: There's lockstat, but you'll want to set that up in advance.

RP: Hm, I've never used this, but I just noticed /usr/sbin/clear_locks

lockstat might not be useful - wrong locks :P

so many different kinds of locking...

RP: Might be worth sending an email here to get some targeted advice: lists.freebsd.org/subscription/freebsd-fs

RP: Also looks like you can run rpc.lockd with debugging turned on.

thanks, I'll send something and ask

should "mount_udf" be able to mount a win 10 install dvd/iso?

I'd welcome critiques here. I'm very, very new to pkgbase: wiki.freebsd.org/MasonLoringBliss/PkgBaseChroot

everyone is new to pkgbase.

pkg-version doesn't work with pkgbase?

jbo: pkg search confirms that there is FreeBSD-set-base-jail-15.1 and some related jail packages and I bet I should be using those instead. Will test them this weekend.