yeah ddr4 price is wtf i just looked

yep

i have a ton of 32gb modules

could even part with 64gb for right price

barely any size and price is wohoow

r430 and r720 mem

so does zfs work now

yep ty

it may be going up due to failures in existing modules

i read it was about the industry retooling for next gen mem mfg

so it's like eol

no doubt insiders made a bundle

speculate -> supply shock -> profit

ah, uch

i know what I wouldn't do if I was the rationing chief

eat all

update: if I turn on AMD DASH in the BIOS it now loads the kernel module without crashing. And then times out. And does not work.

then _unloading_ it causes a kernel panic! cool

oh well, my remote management approach is to shove a USB null modem serial cable into a spare port anyway

Desktop and mobile Architecture for System Hardware

i think that naming sucks

now that this wee gadget is on my desk instead of in a cupboard it is very obvious how hot it gets

maybe I should go find another mini PC for my freebsd server needs

then again I'm in Scotland, there's very few parts of the year where running a mini heater in the corner of the room is a bad idea

zip, how «in Scotland» are you? Edinburgh, or highland and islands? is summer weather severe enough to merit air conditioning?

i doubt any uk is tropical place

but why it has to be in cupboard

without a vent?

Some things only make sense in Scotland.

scottish summers are pretty oppressive lately

ketas a lot of the UK has borderline tropical summers. yes winters are rainy and cold, but summer is a riot of excess heat. not that I'd know, I've not been there in years

yeah really summer does get a little toastier than I'd like

ketas: honestly it's off most of the time because I'm tinkering rather than using it for what I planned to use it for

I don't feel like 'borderline tropical' reflects the reality of UK/Irish weather. It is mostly miserable weather, and then a couple of months that are ok/warm.

misterable

this year estonia is bust again

bust how?

weather wise

weather is warm

but it's not good

everything sucks due moisture

I like tropical weather with an AC that dials into UK weather. It's the best of both worlds.

:p

tropical outside is horrible actually

Agreed. Too hot for me.

not only will mold grow but also all sorts of nasty parasites

True

which will die in winter

i like the old +25'c ... -20'c

those are cycles too of course

anyway i'm now proud owner of old touchscreen thinkpad

should be fun

Hard to go wrong with a thinkpad

yeah

hopefully lasts

but then

just recently someone had new laptop, suddenly didn't charge, took usb c out and replugged, got cloud of smoke and brick

pulled a wrong ticket eh

I've been blessed to avoid horror stories like that, but they do happen

i still have my first, was t60, was n.o.s., did serve me well, started failing and eventually froze and gone

have to look what it did

served past 10y actually

i mean

lid is perfect and carries freebsd the power to serve sticker :p

yeah whatever

i replaced it's ccfl which busted the inverter eventually

i mean they actually are kind of used up goods

any easy way to add bsdinstall scripting to the first time a fresh install boots? i wanna run the freebsd-update fetch, install, shutdown -r now sequence

i use scripted bsdinstall and i always forget to run that manually after first boot :(

( echo freebsd-update fetch install; echo shutdown -r now; echo rm -f /etc/rc.local ) > /etc/rc.local

?

tried reading rc.local in man page but can't figure out what it is

wouldn't that run on every boot or?

oh you rm it

fetch install can't be chained tho

yes

it's old thing

oh so that won't actually work?

now executes via /etc/rc.d/local

it does work

well now you can't shutdown hah

:p

no

the /etc/rc.shutdown.local is for shutdown

that hopefully executes once

i never tried

you can also write your own

that checks a file in install

what do you mean fetch install can't be chained? maybe i misunderstood

like can't run both at once

also isn't this time for pkgbase eh

huh?

oh

well i'm not on 15

what do you think about using firstboot service?

wait you could maybe still

i never used it

and now i never will

oh yeah there was that too

you'll never use what?

really hard to understand what you say

kerneldove: it's possible to script sysinstall, but i last did it quite a while ago

and with it, you could drop a rc script to do what you're after

kerneldove: never had reason do try freebsd-update

rtprio, by systinstall do you mean scripted bsdinstall?

yeah, man bsdinstall talks about it

i said like 10 lines up i use scripted bsdinstall.

what i want is, upon first boot, to run the freebsd-update sequence and reboot again

so should i use the firstboot service to do that or?

seems like accident waiting to happen anyway

how so?

i'd just write your own that does the update, deletes $0 and reboots

hmm

why not use the builtin firstboot to do this?

i didn't use it

but it would probably work

what the cis

hence the rc.local && rm

ketas are you saying you don't trust the firstboot service to only run once?

the fuck is firstboot service

i think there is firstboot somewhere there indeed

kerneldove: do you not manage updates elsewhere in config management?

also poking around it looks like there's the package firstboot-freebsd-update for this exact reason?

i only found firstboot lately

Run 'freebsd-update fetch' when the system first boots; and if updates are

downloaded, install them and request a reboot.

indeed

looks good

that pkg i mean

know where i can look at the files in the pkg without installing it?

i'm away from freebsd system

it's a tar

look at the makefile, via freshports or git

ketas: no it is not, it's a single rc script

hmm freshports don't have it

i mean pkg is a tar :p

but if you meant without downloadinf

hmm

it's right there

yeah in this case yes

otherwise i don't know if there's pkg explorer

but that doesn't mean it's not a tar if it's made into pkg itself :p

Got an intel fiber NIC, fiber switch, can't seem to get them to work together. Both SFP modules are the same intel model, and I can see each one has light coming out of one of the ports before I insert the fiber cable. I believe polarity is correct, light should go no no light, and vice-versa.

The light for the switch is on for the SFP port, but there is no light on the NIC. I've tried both ports, tried swapping NICs (I have two).

And I got it. Swapped SFP modules between NIC and switch, connection is active and NIC is lit up.

Connection is active, but it's slower than 1gbps ethernet, should be runnig at 10gbps...

Well, 2.5gbps is the limiting connection my my client, but I'm seeing transfers less than 1000kbps

How are you measuring this? With some bandwidth measuring utility such as "iperf"?

rwp: it's what scp is showing. Things seem to be moving faster with KDE, which is weird.

If it were me I would try testing with iperf between two systems as that is pretty raw data transfer and does not have any interactions with encryption or anything. Try with both UDP and TCP.

On one system run "iperf -s" and let it sit. On the other run "iperf -c othername" and let them spew between them for a minute and it will say something.

[ ID] Interval Transfer Bandwidth

[ 1] 0.00-11.68 sec 6.50 MBytes 4.67 Mbits/sec

server interface shows: media: Ethernet autoselect (10Gbase-SR <full-duplex,rxpause,txpause>)

The only thing I might note is that the SFP modules are multimode, my cable is single mode - the vender I bought the modules from listed them incorrectly, and I only noticed when I went to buy more that the model didn't line up.

But if the cable was an issue, would it not fail to negotiate at 10g?

that's different wavelengths, it'll connect, but as you're seeing, won't be quick

single mode is 1310nm, multmode is 850nm

multimode can be 1300nm, which is close, but still barely works

In terms of cost, my best course of action seems to be to buy everything multimode and eat the cost of the cable. I've read that single mode is the way to go, but these runs will be between a few feet to a few dozen feet around the house. I just figured fiber would be better for power use, along with power isolation between nodes.

wipt, That's pretty awful numbers. :-(

when i get a chance to do it i need to reboot my nas and check the bios to see what SOL is set for so i can enable it

My GigE copper two systems say this: [ 3] 0.0-10.0 sec 1.09 GBytes 936 Mbits/sec

i'd love to be able to use it as a makeshift ipmi since i'm not going to be running anything graphical on it (just vanilla fbsd) but i'd also like to be able to see it boot

Macer, I only ever use serial IMPI sol because the serial terminal for booting and installing and controlling a server is totally sufficient to do everything.

singlemode fiber is higher quality than multimode, but for under 200m, it won't matter much

Firmware Build Time : Aug 27 2013 15:47:45

rwp: how exactly do you even connect to it? the web interface causes all types of problems since it's so dated for me

tomorrow i'm going to load it up with ram from an old 1u i'm no longer using.

so i'll probably check on it then since i have to be local without the web based kvm

Macer, "ipmitool -I lanplus -H $IPADDRESS -UADMIN -PADMIN sol activate" Replacing ADMIN and ADMIN with the account name and password that are actually used. ADMIN is the factory default for both.

Note that the system must be configured to use that as the serial console. So for example for me on FreeBSD on a Supermicro X9 motherboard it uses the 3rd serial interface which is not the system default and therefore I must configure comconsole_port=0x3e8 console="comconsole,vidconsole" in /boot/loader.conf so that the system will use the 3rd serial port.

Otherwise you might connect to a serial port that the system is not using for the serial console and that's not going to be useful.

Using that terminal to login will be using ttyu2 on my Supermicro FreeBSD system.

and will that, for instance, give you serial access to the bios if that's necessary?

Yes.

On a server type system such as a Supermicro which uses IPMI. (Dell uses iDRAC for example which is similar but different. HP uses iLO which is similar but different.)

wow... thank you so much. i'll have to look at that

this is an X8 soooo.. pretty old

now i want to reboot to see it work lol

let me try it

Good luck!

hm. so i reboot and didn't see anything. i just saw !, in ipmitool

so it seems like it reboot without the SOL working properly. i'll have to go downstairs and check the bios. i was hoping this was all already set

I don't know anything about the X8 series. Good luck!

yeah.. it's from the days of yesteryear

but knowing i have to use ipmitool is a step in the right direction

Over the network the serial port speed is virtualized and so shouldn't matter. But I found that it still needed to be used consistently through from power up to console login. By default 115200 is used in the BIOS and I would then not ever set it anywhere else and let it be inherited through. That gave me the best result.

The reason is that it isn't actually virtual on the Supermicro motherboard! It's an actual hardware serial port on the motherboard. Which is connected to the BMC and then converted to network. So if the serial port speed is mismatched then you get all of the expected serial port fail when speed is not correctly matched.

gm

how do i get rid of these errors

ACPI Error: Aborting method \RTMP due to previous error (AE_AML_LOOP_TIMEOUT)

i cant even login for minutes when it boots

nomia: what is the software version and hardware platform ?

this is 14.2 and a super old old pc

OK

you can try to book without ACPI

s/book/boot

linux's work fine but i cant ever get freebsd to disable acpi and work

try never FreeBSD, 15.0-BETA

s/never/newer

mzar: disable like this? lists.freebsd.org/archives/freebsd-current/2022-November/002838.html

it looks like if i edit loader.conf it will stop booting

can i just do this > 2. Type set hint.acpi.0.disabled="0" on the loader prompt

rwp: comconsole_port and console="..." are separate lines in loader.conf right?

Macer, Yes. I just joined them for the IRC line.

You can set all of those manually, if tediously, at the FreeBSD boot loader screen.

ok. yeah i double checked. i guess remote access was set to com2 instead of 3 (bmc) and it's the same adress you have at 3e8

that's ok. I have it in a position where it can be reboot at will for testing

But if you are not seeing the BIOS screen then things are not set up right from the start.

Are you seeing the BIOS screen?

yeah. i'm seeing it now. i also saw the fbsd boot screen but, of course, it stopped afterwards

yes. seems so. i'm going to adjust the loader.conf and see if i get the full boot

Sounds like you have things very close to fully working. :-)

one thing i noticed as well since i was in the bios was that aes-ni was disabled lol

I enable that option.

like... why? i also disabled hyperthreading because i guess... exploits?

yeah i did

i was wondering why it was off to begin with tbh.. .let me adjust this and reboot and see what happens. hopefully after loader.conf i can finally have some sort of remote access to this thing during boot since ipmiview / java is essentially useless :)

Whenever I benchmark systems I always find them faster with HT disabled. Real cores are faster than fake cores. HT does allow faster context switching but the workload mix is critical to actually being faster.

HyperThreading is a great marketing term for Intel since it is a trademark and therefore AMD can't sell it.

ok. i'm rebooting now

At the boot loader screen you can interactively select "Escape to loader prompt" and then type in these "set comconsole_port=0x3e8" and "set console=comconsole" and "boot" and it should work. Also must be typed in SLOWLY because if copy-pasted in it will garble things (for some reason).

ok yeah. i'm seeing it boot 1990s bbs style lol

mzar: is it suppose to restart the pc when i do OK set hint.acpi.0.disabled="1" and then do boot?

now i am back at geli

The above can be done for an install ISO booting system and then can use the serial console to install FreeBSD. at some point I want to create an install ISO that has this for the installation so that I can simply boot the ISO and then install over the IPMI serial.

nomia, issue command "boot:

"boot"

i think i did that

i choose 3 and then at OK i type set hint.acpi.0.disabled="1"

IIRC you can disable ACPI from the loader menu

and next line is boot

hm. it stalls out after the fbsd boot screen after it loads the kernel

Macer, One more thing from my notes... In order to accomodate the IPMI SOL for the BIOS the xterm size must be 100x30. But the standard terminal is of course 80x24. Set the environment variables to override. export COLUMNS=100 LINES=30 Because there is no SIGWINCH on the serial port console.

"stalls out"?

Yes. I see the fbsd logo and the countdown and the kernel loads and nothing

comconsole_port=0x3e8

console="comconsole,vidconsole"

It's probably booting, try pinging it, using ssh to it.

that's what i put in loader.conf

i can ssh to it just fine

so i mean it's booting afterwards. but it doesn't seem like the console is getting passed to ipmitool

My "sysctl kern.console" says "kern.console: ttyu2,ttyv0,/ttyu2,ttyv0,"

Macer, My /boot/loader.conf file: termbin.com/4zw7

And unfortunately I must run off now. It's the top of the hour and I must run. Good luck! You are getting close.

yeah mine is essentially the same with regard to the serial console

hm.. i'll look around a bit more. but just the fact i can see it booting is better :)

so i'm sure it's just a config issue with fbsd to use the serial connection

hm. still not working.

pastebin.com/NdGcfTri hangs at kernel load... let me look at the handbook and see if i can sort it out

mzar: it crashes as soon as it loads the kernel

i think the issue is that it is using COM3

(: at least im not the only victim

wait just a second... why is there /boot/loader.conf and /etc/loader.conf?

are they the same file?

OK... guess that had to be in /boot/loder.conf lol

no

they are not

so have i been doing things wrong all this time? i always thought adding kernel options at boot was /etc/loader.conf

and now my SOL console works after putting the console settings in /boot/loader.conf

(: yay?

yeah. it's awesome. but still. that's an hour i'll never get back lol

let me make sure that's the case

going to reboot one more time

i think if i edit /boot/loader.conf i will be finished with freebsd

Yeah. /etc/loader.conf isn't a thing.

it will take me all day to add 1 line

what in the world?

how did i even get that wrong? lol

maybe i edited it using muscle memory one day then it stuck in my brain that's where it was meant to go

i guess we learn by doing (or not doing) ... seriously with that.

Could be! You could always "ln -s /boot/loader.conf /etc/loader.conf" just in case it happens again.

Sounds like you have that all cleared up now, though. Likely won't forget in the future.

i sure won't lol

but yeah i might want to do that. you're right

how can i edit /boot/loader.conf in single user login?

it says read only filesystem

rwp: ok so that was my problem. we'll just try to look past that :)

it's working now

which is absolutely awesome. now i have makeshift kvm with ipmi using the serial console... no more need for crappy java based web interface!

nomia: perhaps you can run an older version of FreeBSD on this hardware instead newest one

mzar: thats why i tried 14.2 so i could learn how to upgrade

(: i learned how to make zfs readonly=no

now i wait 5 more minutes and reboot

incidentally, how does one load a module from the loader prompt?

acpi errors freeze the system so i can only type for 5 seconds every few minutes

I tried doing a `set ipmi_load="YES"`, then I tried unloading and reloading the kernel ...

oh it's load

actual java?

like java applet?

i have hp printserver like this

entire webui is made of several java applets

and then i have dlink ipcam where login is activex object

mzar: do u rly think older fbsd will work?

could work

also, newer

idk why this works fine as single user but not regular

ketas: i get this error with the 14.2 and 14.3 installers too

and latest 13 i think

some supermicro server i read

getting tired of spending dl's on this tbh

hmm multiuser doesn't work?

i would be curious if CURRENT works

because this doesn't seem hw that the support is removed

also, what works in single user?

if i am in multi user it boots but i get errors all over the screen every few minutes

wait

and between everything is frozen

who has which hw

my hw is a super old pc

ACPI Error: Aborting method \RTMP due to previous error (AE_AML_LOOP_TIMEOUT)

yeah Macer had supermicro :p

lots of these errors

i also get acpi errors

but does it boots?

ya

when does it stop?

single user is perfect

so no need to download anything new

if single user works

the freezes start as soon as it finishes booting and gets to login

that's... fun

so i can type root and wait 2 min for it to show up

then hit enter

wait it still works?

(: i dont call that working

nomia: have you submitted bug report about it ?

i was about to ask if any caps bulge in that hw

mzar: no

plase, don't hestitate to do so

the other bug reports i saw werent fixed either

so such bugs were reported ?

ketas: nah this pc works great with linux

does it always take 2min for text to appear?

and on physical console, right?

FreeBSD is different OS

fun, definitely

what's the error message?

nomia: perhaps BIOS update will help

i missed that part

yeah bios update too

ketas: no i get a few seconds to type like normal every few minutes

hmm

then it freezes again

that's right on boot?

i recall that from somewhere

like during the fbsd boot screen?

Macer: its all the time

yeah but i mean is it before or after it starts booting?

o no the boot screen w the red artwork is fine

able to run top -qaHSPs1

maybe?

oh so it actually gets far enough where you login then just freezes?

ketas: gimme a min and i will try

Macer: freezes start before i can login

but then waiting 2min is fun

yeah but you are able to run top

i think so

you don't get like a kernel dump or something? that sounds pretty bad

if it were me i'd be thinking hardware

it will have error messages on it but i think it will run

Macer: its messages like this

if it responds every few minutes for a few sec you can have top from there

but i guess if it runs in single user mode without issue then it's something getting loaded

ACPI Error: Aborting method \RTMP due to previous error (AE_AML_LOOP_TIMEOUT)

hm

but single user mode works is weird

on my old boxes i get i look for any bulging caps in psu or mobo (could do more), update all the firmwares and run memtest86+... then would be nice to run buildworld as test maybe

but that doesn't apply here maybe?

maybe does

oh and new battery :p

sometimes that's fun

is hint.apic.0.disabled="1" still a thing in loader.conf?

ya i added that

it doesnt do anything i dont think

i ran memtest for 3 days and 0 errors

right now i cant type at all

still waiting to login and run top

hell of a bug

since kernel works

single user works

wtf is it :)

so any hw the kernel accesses there, or anything else it runs

of course the system performs much more actual work in multiuser, it would be fun if it pushes power system over limit

no idea about this tho

i'd probably yank whatever i could out of it and disable zfs just to sort out where it's starting

disk works hopefully?

i can run linux on this hw no prob

oh

fancy sw error then

i have an ancient A10 that fbsd just won't run on :)

same sorts of problem

a10?

yeah. ancient AMD APU

now i cant even ctr alt delete

in linux i have to disable the on die gpu power management

cant fbsd run on amd?

i think i recall strange bugs being reported

sure it can :)

i run it on my ryzen 3700x

ok

a10 is fro, 2012

from

ketas: yeah

sounds about right

lol

still works great for debian

you can run much older hw

my macbook pro that i use as a daily driver with fedora on it is from 2013

i yanked out the ssd and replaced it with an adatper + nvme and replaced the wifi module with an intel one so it works out the box

yeah

i have to unbust the psu but i had 2006 amd64 machine

actual amd too

i think fbsd still has the old 32bit stuff around

what are all these methods that keep aborting

yeah 32bit is now going away

:/

so you can pull the 386 out the closet and run it

oh really?

\csms \rtmp \rbyt \_tz.thrm._tmp

say it isn't so freebsd! i mean i get it lol

probably not actual 386

why not?

have to recompile kernel

for a 386?

can i disable whatev is causing this?

it was upped to minimum 686 cpu iirc

hm... i saw something like that...

i mean i have ran fbsd on actual 486 tho

the first machine i got

i did too back in the 1990s :P

back when the ports tree was so innovative lol

i have no idea how it evolved

66mhz cpu

too bad bsd had to deal with that whole sco unix nonsense lol

nomia... i saw something earlier about that .... something about adding some sort of hint

#include <lawyers.h> #include <policies.h>

:p

i only installed 4.6 as first one

Macer: i put set hint.acpi.0.disabled="1" in /boot/loader.conf

wait no

no set

only hint.acpi.0.disabled="1"

i think i have to kill power to reboot :(

i'm not sure if that is related tho and that's very old

bits /6

err

hint.apic.0.disabled="1" <- it's not that?

(: oh

well i changed it but its still freezing

ok i was able to run top finally

top -qaHSPs1

but it is frozen after a few seconds

top process is [idle]

oh hey

[acpi_thermal] is using the most time

can i disable that?

debug.acpi.disabled="thermal"

i tried that with the apic and it is still giving error messages but not freezing yet

top -qaHSPs1 is working too

hmm only 1 set of errors and no more?

you have a lot more patience than me. i'd have given up by now lol

o nope it did error messages again

but no freezing i dont think

not yet :D

oh go fuck yourself This may be because upgrading from this platform (amd64)

or release (14.4-RELEASE) is unsupported by freebsd-update. Only

platforms with Tier 1 support can be upgraded by freebsd-update.

See freebsd.org/platforms for more info.

last thing i want to do is hassle with this

Macer: (: i like winning

i am glad this is working now

mostly

hmm it looks like it still freezes sometimes but much less

nomia: that's fun

thermal is unsure what

fan control?

just temps?

freezes less eh

idk

what runs there?

according to top

ye it is still freezing sometimes but i figured out how to switch consoles

load is not high?

so now the error messages arent a big problem

hm load is up to 3 sometimes

there we go

what exactly runs there?

i cant run pkg updated cuz the xfer keeps freezing

wdym

well load is 3

like what apps?

so wtf runs

idk

3 times over the cpu time

i dont see anything in top

and load is 3?

just zfs things and powerd but no user apps

those top args also show all kernel threads

this system is brand new i didnt even install tmux on it yet

something runs if load is 3

why it runs long no idea

io or cpu

how old is it :p

well it has acpi :)

smth is running and taking all the cpu for a minute i guess

like periodically?

ya

unsure

service crond stop?

but it should not run anything

pkg is updating quarterly data.pkg

so that runs eh?

it is taking 10 minutes

downloads a lil bit and then freezes

freezes whole system?

i am running pkg -d now so i can see progress

yep

disk is bust or something

or it swapa?

swaps

i doubt it

it runs linux w no problems

i put these disks in a friends system to install

oh right

bc of the freezes

i tested them overnite

top says cpu is running 99.9% system 0.1% interrupt

that means the kernel is using all the cpu?

i have to yank all this ddr3 ecc out of my ancient supermicro 1u to toss into my ancient isilon lol

mmm yes

i'm considering turning the old 1u into a router

but below it shows where it goes

if pkg finishes downloading im going to try removing the acpi disable and reboot

no vacuum cleaner noises?

data.pkg 78% now

if it's zfs, kernel shows it in top

shows as [*zfs*] something

you used zfs right?

ya

no errors in zpool status

thats good but

unsure where cpu goes

maybe kernel errors

zfs uses all resources isn't new

disk is not full?

that also causes slow io and high frag

outside of pkg, it runs?

or wait you told it won't

so load is 3 after boot too?

and keeps there?

it goes up and down

up when it freezes and then back down when its not

hmm

and no single user issues?

unsure what periodically freezes the system in multiuser

no way to get actual top outputs?

i can run the command u gave me

here it gives me

PID USERNAME PRI NICE SIZE RES STATE C TIME WCPU COMMAND

11 root 155 ki31 0B 32K RUN 1 254.7H 96.78% [idle{idle: cpu1}

]

11 root 155 ki31 0B 32K CPU0 0 249.4H 96.78% [idle{idle: cpu0}

]

which is like 2core and idle

i'm curious what's below ir

it

as load has to come from somewhere :)

even if that thing comes from kernel, it has to come from somewhere within kernel

what does dmesg say about this?

i just realized fbsd doesn't have a dmesg -w equivalent

dmesg is full of ACPI Error: Aborting method

and ACPI Error: AE_AML_LOOP_TIMEOUT

oh -w is followtail

but you can pipe most of kernel into syslog

and tail it

at least

whatever fancy acpi bug is it

oh there's kern.msgbuf_show_timestamp

i found

tmux freezes even worse than a regular login :(

crazy resource exhaustion0

?

or just something blows

what, i have no idea

i think its just random freezes maybe

idk

i keep finding tons of pages about AE_AML_LOOP_TIMEOUT for linux