ZedHedTed: you can apply this directly with sysctl

thank you CrtxReavr

I already have an old clonezilla pendrive I got to shell but I didn't try any nt commands. will investigate more.

using 4GB of RAM for: ZFS, DE/WM, browser media player

ty

jauntyd: cool

:)

pretty amazing in these modern times, which is why I pointed it out

what is amazing ?

only uses 4GB RAM

yeah

if it works, nice

and you can always upgrade

jauntyd: whats DE WM ?

enlightenment

cool

FWIW, I had the issue of creating raw ZFS (no partitioning) on disk without previously cleaning up GPT and its backup from the disk; in my case, the complicated procedure with dd described in forums.freebsd.org/threads/gpt-table-corrupt.52102/#post-292341 was not necessary as gpart -F removed remaining GPT backup without destroying ZFS, kudos for having such useful tools in the OS

I don't know why wblock@ recommends using dd over gpart, but it _is_ an almost decade-old post - so that could have something to do with it.

On the other hand, overwritng sectors with dd _will_ work.

The previous utility for manipulating GPT was gpt(8), but that was removed way more than a decade ago, so I've really no idea.

vedranm: gpart(8) only manipulates the partition tables, not the filesystems laid out inside the partitions

this is sometimes annoying and other times very useful

you can "move" partition ordering, without losing the data, for example.

vedranm: `zpool labelclear` probably is the thing you want to use for that, but it wasn't around in 2015 for that blogpost

dch: it was added to illumos in 2009 (illumos/illumos-gate 096d22d) and openzfs in 2013 (openzfs/zfs 51a3ae72d)

dch: the issue was dmesg reporting backup differing from main GPT, zpool import worked fine tho

zfs doesn't really care about disk partitioning, so long as there's no encryption like GELI being used

that is very annoying when the gpt backup headers don't match what zfs has/sees

ack

i've noticed this problem before and never got around to working out the cause - does ZFS not automatically clear the backup header when you create a pool?

i'm not sure it should, as there's a theoretical way to recover data from a disk where the labels are intact and none of the data has been overwritten

having zpool-labelclear(8) be an unambiguously destructive administrative command prevents last-ditch footgunning

or it at least has the potential to work that way

that's a reasonable argument, but i think it should print a notice when creating the pool then. or does it do that nowadays? i haven't created a new pool for a while

it always has, if memory serves

what is the lightest live image of freebsd i could find to test if my wifi works

ober: thanks! i'll read the man page for sysctl. my RAM's coming in today anyway.

sibyla: if you're looking to test your hardware, try nomadbsd. it's live and based on freebsd, but kinda large. 4GB. it loads all the drivers for you and starts X so you can try connecting to an SSID w/ their GUI applet.

thanks

I try to run a binary and it exits with 'Bus error'

what is that exactly ?

the program is trying to access memory that's not permitted or that doesn't exist

great, I guess welcome to openssl 1.1 vs 3.0 hell

what binary?

check_mysql from nagios plugins, nvm, recompiled it with openssl 3.3

works fine now

weird because I wasn't checking an SSL host

if it was linked, it would be accessed when the program was ran

even if it didn't use any functions within

I am pretty sure that Bus Error indicates invalid opcodes. Doesn't it? Not a segmentation violation.

rwp: that would be SIGILL

SIGBUS is usually a memory error that's not specifically an access violation, things like accessing misaligned data for example

Hmm... Misaligned data access.

(might be some newer memory protection mechanisms also raise SIGBUG, not really sure)

last1, On what architecture?

SIGBUG!?

am64, 14.1

just upgraded the server

ivy: excellent news, it turns out that the hp elitedesk mini that died on me when I tried setting it up with freebsd is not dead, it was just booting into a stupid bios mode that my shitty old monitor did not want to talk to

on a hunch I attached it to another monitor and it was like "please enter these four digits to disable secure boot", which I am sure seriously improves the security of it or some shit

9991

anyway now I guess I can play with FreeBSD on x86

as a real computer that can do shit like hypervisors

I wanted to use this machine as a base for VMs anyway so I hope that bhyve is less annoying than connecting libvirt to systemd or whatever

as far as hypervisors go i have a mental model of it and the networking setup which is largely in tune with how it really works

sounds helpful, I hope bhyve is that

maybe, but the reasons for using freebsd on workstations do not involve bhyve for me.

i used kvm until i discovered bhyve

i like bhyve quite a bit better

one of the reasons i'm not a windows sysadmin

Never ever again

oh, I don't intend to use this as a workstation

I intend to use it as a little server in my data closet

a workstation is not dissimilar to a server. in just does personal things.

ok

Workstation = what consumer boards used to be like, with enough PCI(e) lanes etc

I feel like it's a crime when motd doesn't fit in 80 columns.

Well having an motd is a crime initsself.

What, you don't want people accessing your system to know that they could be subject to law enforcement if they're not entitled to access, or some legalese nonsense like that?

I've lost count of how many times I've seen that.

you must logoff immediately if you are law enforcement

lolol

hah