regis: yeah, I don't know

if I drop mic=0.00, still feedback screech and no audio

oh, interesting. actually putting it on, mic is infact routed to the speakers on it right now

jbo PR#283229 seems reasonable to me. I don't bother patching jails because generally the project doesn't issue errata that require it often

kernel is a very different thing

sorry about that, folks

:)

debdrup: Meh, you're allowed to have a life outside of IRC.

Has there been any recent luck in 14.2/15 with getting the genet driver working on the RPI4/5?

I'm still running fine with wifi although curious....

does the interface come up, or

Tenkawa: thanks!

sjk: no problem.. did that help on the os-release?

I'm trying to (via ansible) generate a zfs jail template. I unarchive base, fetch and install patches with freebsd-update -b. But then I'd like to create a snapshot of the template named the current version and patch level

like zfs snapshot zroot/jails/template⊙12

but I can't seem to figure out what the patch level is unless I boot the template jail

Tenkawa: the rc.d script you mentioned uses freebsd-version, which requires the jail to be booted

can the underlying files be mounted but not "booted"?

if so you could chroot and run the equiv function

I'm still trying to figure out if the RPI image can be rewritten to use ZFS and not UFS

(for everything besides boot/msdos stuff of course)

Hmm, interesting idea

i would just tag them with the major version and date and move on

rtprio: to answer your q for my genet one.. no.. the driver isn't seen at the hardware level. It acts like the acpi/dt still isn't quite right

s/driver/interface

Its been a known issue since day 1

Going to try ZFS install again... its theoreticly possible...

(Here's where having several spare units is helpful)

yes, you can do it but have to install by hand

hit me for not doing guard partition sizes to avoid disk too small problem

unsure why the fuck i created this pool like this knowing it beforr

e

ketas: what are guard partition sizes anyway?

Ltning: not ignoring your e-mail, will respond soon-ish

yuripv: when you make partition smaller than whole disk on purpose. you lose 0.001% space but can replace disk with any other disk with approx the same size

i think i somehow forgot that after getting 4 totally different maker/model hdd's with same exact bytes capacity :p

or just use the same drives. this isn't a drobo or other fakeraid

hello, could you guys invite the bot to #freebsd-pulse again ?

rtprio: never heard of it

but i had this idea to have them aged differently

fits this purpose i guess

i get that if you add enough redundancy, that issue disappears :p

different ages is fine

but keep them the same model

you surely mean performance here right?

as i recall, 4tb drive from different vendors might be slightly different sizes

rather than putting as partition table and rounding them all down

yes, you can use drives from different vendors, but in such a setup you usually create partitions for ZFS not covering the whole space

yeah, i'm not doing that

mzar: worked perfect :)

what worked ?

my install attempt to get a ZFS root on my RPI5

zfs on a nvme is a nice thing

yes, it is

ufs was just too restrictive...

4 t = 4000787030016 b, 4 * 12 t = 12000138625024 b, 160 g = 160041885696 b, 160 g = 160000000000 b, is what i see here

either i repool or just have bigger ones :p

the stock FreeBSD installer cannot use ZFS if you not give it whole drive - that's serious drawback IMHO

a serious drawback is running zfs not on the whole drive

that could be changed maybe0

?

but such a setup could be easily complete by hand

mzar: yeah I discovered that yesterday

i know what it does with two different size ones

just gives it to zfs and it becomes pool sized to smallest diak

atm now I have lan & wifi running great using usb adapters too

rtprio: if you use geom, ale FreeBSD does, it's not that importat to have ZFS on whole drive, especially that you want boot loader too

yes, obvously i make an expection for my boot pool

does solaris boot from raw?

sun servers did?

where the zfs originated anyway

I don't know maybe tsoome_ can answer that ^^^

btw, openzfs can be eventually updated to shrnk, right?

ketas: you could shrink from a mirror to a single drive

but that's about it i think

i read that

could try

i mean it's a fs and volume manager, it knows exactly where all data is

it just needs that feature somehow implemented

solaris does not boot from raw (partitionless) disk. I would not recommend partitionless for boot disk (with UEFI this is not possible anyhow).

and with bios, you get CSM, which may be with nasty bugs -- having partition table does allow us to use partition size to determine disk [partition] end.

and therefore avoid reading past disk end, which may hung some systems;)

kevans: No worries. I'm ignoring emails for a living, according to recent news I've received. :D

there are worse ways to make a living :-)

rtprio: regarding tagging major version and date... good idea

yeah, it's one of those things, not ideal, but spinning up the jail to check is way worse

We're spinning up a Nomad cluster for "production" use this week, with any luck. Anyone else using that for anything but futzing around?

rtprio: yeah, I agree

rtprio: why is it worse? Its morse consistent and it takes about 2 commands that easily could be scripted...

much more reliable..

"tagging" becomes invalid on a jail soon as that jail is used again

morse :)

You have no way to ensure consistency

Tenkawa: you'd either have to round down

it's an extra step. perhaps i'm just old and tired and no longer scavenge whatever hard drives i can get these days

not sure about anyone else but zfs sure seems a "lot" faster than ufs

at least on a RPI5 with NVMe

it could do that

how much ram you got on that, boss?

8gb

what's the pool size?

i mean what's data size

know the cmd to check offhand? (rusty on zfs cmds)

it could fit into ram entirely

zpool list

oh... this is showing my drive pool

zroot 460G 5.04G 455G - - 0% 1% 1.00x ONLINE -

that's a beautiful cap

if you tried read, it cached it likely

I did...

so does ufs

I was scoping data

I had ufs on here earlier and it felt much slower

who knows

zfs adds overheads

compression is on or off?

I definitely like the added flexibility though

off

I definitely set that

if a game runs swell through wine on linux, can i assume the game will run without any problems through wine on freebsd?

On an rpi5 you can safely run with lz4 compression at least. It's likely to offset the cpu usage with lower transfer volumes. That CPU is beefy enough.

genshin impact works through wine on linux but i'm not sure what its state is on *bsd

Ltning: with already offloading some cpu for my lan and wifi adapters I'll keep extra cpu overhead off....

Wait, genshin works on linux? My kid is pissed he can't get it to work on his M-series mac..

yup

silent change mihoyo made to their anticheat

Tenkawa: I'd be very surprised if it actually matters in a positive way. Data transfer will almost always be worse than a bit of cpu spent.

w00t, he'll be happy to hear. They did this to silence complaints, or what's the deal?

Ltning: true.. but the io scheduling "could" get fickle enough considering I have no scheduler/temp/pwm control either

don't want to cook it

i got genshin working pretty easily with lutris

genshin should work out of the box with no patches

i wonder why i get pissy zfs write speeds in just one mirror, in addition system freezes up for seconds

hdd went bad?

there are no errors anywhere

no dedup or anything

shows frag 75% cap %76

that's the problem?

s/%76/76%/

one of old disks there now refuses to complete long test funnily

running zfs on old hw eh

used to run it to finish, now only short test complete

only has 4g ram but somehow 12t/4t pools write 10g file at stable 150mb/s and there is no slowdown or anything locking up