fbsd still says p3 on uname after installing p4...

if it doesnt require anything kernel

it will remain

byt freebsd-version will show p4

ahh i get it. would the kern say p4 if i had built it from source or does even that say p3 still?

if freebsd-version says p4 youre good to go

stop worrying and go play scorn

not worried, just OCD and it annoys the shit out of me

dont know what OCD means

go play scorn then ;)

did not like the game ;/

i've not even heard of it, so you're one up on me

indie horror game

oh boy, have to build rust again... guess I'll check in on portmaster later lol

pkg why compile

cuz i run ports, tbh i don't even know what pulled rust into the mix. don't need it that i'm aware of, i just let it do what it says it needs to do

gotta customize eh?

ports is 90% the reason why i'm using freebsd

i do have custom installs for postfix, dovecot, etc

have 2TB of space. worry about 5 MB worth of deps

r0ni: did you reboot after installing the kernel? and where does it say p3?

he mentioned uname

if a virtio device reports 512 byte sectors in diskinfo, gpart, etc, but zpool creates with ashift=12, should i trust zfs or force ashift to be 9?

isnt 512 still the default

cpet: not sure, but i'm second guessing myself because this existing pool has ashift=12 but now new pools have ashift=9 as expected..maybe the existing pool was somehow created with 12

just going to go with 512-byte then for the new pool

well once you install with 512 cant upgrade to 4k

depending on the device it may help or degrade

yeh it's a virtual disk though so never going to change i assume. not sure what the host intends or what the actual underlying size is but if it's reported at 512 then seems the best

but if this is a desktop doesnt really matter if this is a server then yeah but only if you have a ssd or nve drive

wonder if there is a good way to test performance of random read/write from within the install cd before i install and get along with the setup

do you have ports ?

i suppose i can set that up creating a temporary system on disk just for this purpose

is there a port?

theres bonnie and bonnie++

nice..will try thanks

jmnbtslsQE: if it's for windows server I'd recommend 4096 block size, you'll run into all kinds of issues with databases otherwise

I honestly dont know the difference in 512 and 4096 I just install GPT and it defaults to 4k

OK, well it's just for this freebsd VPS, i'm inside the VPS though, i'm not the provider, so just gauging based on what i see inside. haven't asked the host yet but i will test performance with the tools with both 512 and 4096 to see

vortexx: ^

UEFI + GPT defaults to 4k

OK

I would just reinstall if its new and use the web base VNC

the underlying virtio disk is reporting 512 i guess so hence supecting that the host intends 512 (e..g maybue throttles on the basis of iops for 512 byte blocks)

had a provider default to UFS rather ZFAS despite the server having 24gb of rams

so :)

yeh this is a new node so just testing it out before i actually install my zpool

good idea

zfs is more of a ram issue

eh.... no it isn't

if the VPS has small ram not reallt recomended without tuning the hell out of it

pretty much it is if you arent using say hammer which doesnt hog all the rams

with 24gb i think there's plenty of ram

this is my personal ded server I really doubt his VPS has 24gb of ram

so far i just set arcmax and use ufs on zpool for certain pieces of data that are operated on intensely, and it seems fine, the performance of another node on this host was pretty bad but i think it may be because they throttle iops (not intended to b ea fast disk) and possibly bvecause i used 4096 instead of 512

welcome to the woprld of VPS

elliot@phil:~$ sysctl -a hw.physmem

hw.physmem: 34321428480

you share IO cpu

heh this one has 1GB memory

cpet: well never had this issue on other hosts at least to this extent

hw.physmem: 16427266048

every host is different

indeed

in my personal opinion andf again my personal opinion anything from the US is garbage for what you pay

i pay 46/m for a full ded server with 24gb of ram and 2 x 512 SSD and 1TB HD

nor running zfs with 1gb is frowned upon

europe isn't better is it?

but you can tune it

this server is in germany

ah i see

it's definietly a lower price than what you would find in the US i agree

Title: ZFSTuningGuide - FreeBSD Wiki

may need to play with that

but you can make zfs run decent with 1gb

desktop has 32GB ded server has 24GB and server has 16GB so I have some play

you dont :)

ye hi've read that a few times, honestly i think the issues i had with the other node here was throttling + possibly wrong sector, or at least in any case, it's specific tot his provider not my config

dont think they throttle you just share

if another is compiling world

you will feel it

this is why you look at bare metal discount server

Title: Discount Dedicated Servers | Contabo

chepa isnt the way to go when it comes to serving

it's a good question i'm not too familiar with virtualised environments but the disk can be quite slow, to the extent that i wonder what this underlying storage actually is - maybe not even a directly connected spinning disk. pretty much always the same performance over 12 months

again you are sharing IO with x people on the same server

even if they mention SSD or NVE having 20 VM ..

it's designed to be a large non-fast disk which is one clue to me that it's throttled specifically or something similar

well depends on what you will use it for

a small site or a enterpsie site

not doing much serious with this but i do want to put poudriere on it

I was lucky enough to get this server in a discount

so going to put poudriere ina jail and do RCTL

do you have a reason to compile by hand or just benching IO

cpet: that you one linked is theone you mentioned before?

yeah

typically compilking so i can choose options

ah nice

I dont use anything from the US all my services are from EU

well, don't need any extra now but if i do i will take a look at that

you get way more for the buck

well think about that as freebsd doesnt like to grow it's FW

it works but doesnt like it

FW ?

FS

OK

what exactly are you using this for ?

miscellaneous

most of the important things are on the bare metal servers in the safehouse(s)

if thats the case why do issue with 512 or 4k ?

well, i'd like to put this to use since i've paid upfront, and if i have a chance to fix the performance issue i experienced, worth checking the sector size. can't run poudriere on it if it's performing the way it was

160 hours building rust then it timed out because i forgot to raise the timeout beyond 10 days.. heh

no need to build rust spefifically but it's a dependency

pkg install rust

why the hell do you need to compile it ?

especially on a 1gb VPS ?

have not set up poudriere's use of repo-installed pkg

is that possible?

i guesss, just need to tell it to skip it

but there is also the possibility of the freebsd repo not being in-sync with my system, same ABI, etc. forgot how that works

again why do you needf poudriere for rust ?

i think it only matters if my minor version is not the latest, right?

nuking the hell out of this

rust is just the dependency

then you can just instal it and worry about the actual program

i've always been cautious about mixing repo-installed pkg and ports-built

pourdiere is a build system as far as I know it cant fetch and install pkg

it build once then uses the cache to make it quicker

if you dont install portd with options doesnt matter

just a waste of time if you dont need too

yeh i do use options, that's the main reason

rtprio: yes uname still says p3 as the kernel has no update(s)

well you do you then

OK

r0ni: you have to reboot after installing your new kernel

r0ni, freebsd-update if it says p4 youre good

yes, i'm aware that one needs to reboot and yes its also at p4... just a kernel thing I find annoying

if you dont need to update the kernel why update it just to see the pX

why was the graphical logo removed from the efi loader? it's been replaced by the ascii logo.

been replace with a perdy logo

the text-based logo was replaced by a graphical logo a long time ago. so, why did they decide to go back to the text-based logo?

version

today's version of freebsd-current

freebsd detect the system and uses either UEFI or syscons

what?

older systems will likely use syscons which is ascii while the uefi uses the newer BMP/PCX

I just said that the text-based logo is a part of the efi loader.

ok

you also fail to provide actual issues configs and other files

so ...

is there a tunable that I can put into loader.conf to enable the graphical logo?

should be auto based on what the system detects

According to this: forums.freebsd.org/threads/boot-menu-logos.81216 , the uefi loader always displays the graphical logo, so, why does my efi loader display the ascii-style logo?

Title: Boot Menu Logos | The FreeBSD Forums

booting in legacy mode

can someone please explain this mystery to me?.. if my system uses the uefi mode, why does the boot loader display the ascii logo, instead of the graphical one?

any reason why you dont post configs ?

cpet: I have zfs_load=YES in my loader.conf

conole isnt the same as a FS

maybe there is something in my uefi bios settings that prevents the orb graphical logo from showing up.

cpet: I have only one one line in loader.conf. why would it prevent the graphical logo from showing up?

I guess I'll reboot and take a look at my uefi bios settings

this is so strange

How can I change key mappings e.g change left alt with ctrl. Change /swap capslock and Esc key?

icky

ok

bootcamp is tricky

and windows 11 can't be installed

Anyone who was "ever" able to crash Linux AND a freebsd OS?

haven't killed a FreeBSD install yet, but have murdered lots of Linux installations

mostly due to the package managers, though

ayecee which os has least crash ration? I heard freebsd is "hard" to crash even intentionally by mere system load but crashing linux with intentional heavy load is just implicit to happen? true?

hackkitten ^

hackkitten do I make an sense/

what do you mean by crash unable to use the os, kernelpanik or need to manual repair the instalation?

FreeBSD in my view has an advantage because it's a singular distribution, not just another variation of packages tossed around a kernel variation

means less chance of those sneaky kernel space bugs

now thinking about it, I was able to crash linux and freebsd (under all my definitions)

I need an OS which does not freeze regardless of load. It can decide to performe limit as per hardware capacity but it can "not" freeze

or an OS close to above said

get an IBM Power mainframe with their UNIX variant

costs a bit, but it's virtually bulletproof

plus you get to hotswap CPUs

hackkitten ok which os?

z/OS, for z-architecture

:D

Title: IBM Z - Wikipedia

the alternative is to use fall-over services within the same network

Use rctl(8) to throttle things.

Or just use cpuset(8) to reserve at least one core for only FreeBSD to use.

That's effectively what mainframes do.

Is there backup of old FreeBSD Perforce data somewhere?

Stuff that used to be in perforce.freebsd.org:1666

Maybe on ftp-archive.freebsd.org ?

I'm pretty sure I've seen it _somewhere_

What, in particular, are you looking for?

is freebsd performant on a laptop/desktop or should i go with linux?

if i want performance

performance of what? try it and compare for yourself

no one here is going to stay, "sure folgers, linux has better performance if you're playing tuxracer"

rtprio for things like multithreading etc

you mean to say performance is subjective or we can't know?

for example they say the network stack of freebsd is much more performant, is this subjective also?

I have not benchmarked either performance or power but I would guess that performance would be very similar on both FreeBSD and Linux. But I think Linux has more power saving features. Linux is definitely better for WiFi.

would performance even matter when talking about laptop hardware, they tend to be inherently inferior to a proper workstation

the point there is mobility not performance

IMO it's all pretty meaningless unless you're testing a certain use case on both freebsd and linux on the same hardware. saying "performance on laptops is better on x" doesn't really make much sense

a one year old laptop OCcan cream a ten year old workstation.

s/OC//

the comparison is basically subjective and asking for finite resolution where there's none to be found from any of us. only trying the various environments with the intended tasks will the real answer become apparent, I think.

aside from say nixos with a declarative/reproduceable environment, giving up boot environments is not enough to sway towards linux, performance be damned

tsoome: code.illumos.org/c/illumos-gate/+/2510 typo, line 196

on smb_node.c

wrong channel but thanks!

test

ok, it works

hi all

'lo.

I have a new 13.1 install with ZFS, and it looks like ZFS is going off of the device numbers, which change when I connect additional drives, causing the machine not to boot. I'm trying to figure out the appropriate way to configure this zpool so that it will always boot properly

For whatever reason, when I add or remove disks from the PCI-X sata controller, they show up as ada0,1,2,etc and offset everything from the builtin controller by that much >_<

Basically this problem but I don't see a satisfactory solution here: forums.freebsd.org/threads/how-find-numeric-id-of-drive-zfs.62475

Title: ZFS - How find numeric id of drive zfs? | The FreeBSD Forums

so what's the official place that developers hang out these days? is it the discord server? is it an IRC chat? do we have any of that documented?

some of those more specific dev channels mentioned on the freebsd IRC channel list are fairly active

rtyler: pci-x?

like, the old high-bandwidth pci?

PCI-E I think I mean, can never keep those straight

express (newer) vs extended (older), totally not confoofing

pcie is rarely written with a dash

The trademark is PCIe

Or PCI Express

rtyler: what kind of pool? i would suggest putting zfs boot blocks on all drives so it doesn't matter the order

There are efi and gpt boot partitions on all the drives, the problem is that if I attach or detach drives to the other controller, that throws the boot because it then thinks that this other device is part of the pool when it's not

Basically I installed, I had two drives that were not part of that pool, the show up as ada0, ada1. If I remove those drives, the four in my zpool have their numbers decremented by two, and then there are problems booting correctly

can you not set one of the correct drives in the bios? or still zfs gets confused?

What do you mean by setting them?

twsed them as the boot device

set them as the boot device

Am I getting this right that when you run `zpool status` the members show up as ada2 and ada3 (or some partition thereof)?

they're already set as the boot device, the machine will boot but then will fail out to single user mode because it gets confused about the status of the zpool if the numbers of these drives aren't exact, meaning two drives have to always exist on the other controller taking up ada0 and ada1:

Title: zpool status · GitHub

Okay, that's inline with what I thought. I made a similar setup work by switching the members to diskid/DISK-<ident>p<#>

oh, that's a good idea

I got the idents by running: geom disk list | grep -E '(Name|ident): '

what command were you using to adjust the pool? did you just detach them and then reattach with their disk ID?

and you should be able to zpool replace the drives one at a time

I don't know how to change that in place, so (with raidz1) I removed the spare, then re-added it using the diskid/DISK notation.

s/spare/sparable one/

Resilver, then repeat the process.