˜/40

741

404

you can create a new ZFS data set within an existing ZFS pool, and give that data set different properties (e.g., different compression), can't you?

*different properties than the pool / "main file system" has

jemius: yes, you can. think it is called a subvolume

or maybe that is just the term for btrfs in Linux lol

terminology in ZFS's empire is indeed a bit challenging

but I know for sure what a pool is o/

I think you mean inside another dataset, a pool can also have properties, but those are pool-wide and you can't override them in datasets (but compression is no pool property)

A pool is what you create with zpool, i.e., a pool of volumes, for example configured as a raid. A data set is something created with `zfs create` on a pool

AFAIU

zpool create will create a first dataset too, as you can only create child datasets with zfs create

can you mount parent dataset A to /home, then create a child data set with different properties in A, and mount this new dataset B to /home/bob/my_diffferent_properties_folder ?

yes, but you could make the home for each user a dataset too and give them create permissions and profit from the default mountpoint for datasets as that will be the last part of the dataset name inside the parents mountpoint

the costs of datasets are negligible, just create more of them and profit from inheritance of properties

If I copy from one set to another that won't result in more than the links to the blocks being copied I suppose

because right now I have >100GB on there already and would have to move stuff to create new sets

between diffrent datasets I'm not sure, but I think it will result in a full copy, so remember to create more dataset for next time (or decide that it doesn't matter as much and try now if it is fast or not) (maybe someone else here knows for sure)

kk, thx

Anyone know what would cause realtek-re-kmod to not show the interfaces? /boot/loader.conf has if_re_load and if_re_name set, after reboot I see the module loaded in kldstat, the NICs show in pciconf -lv but I don't see the NICs in ifconfig output

Ne'er mind. Built the kmod, installed it, and now works!

hi, everyone! does anyone have FreeBDS + BSD on different datasets of the same Zpool (so only one disk)? native encrypted zfs ? suggestions? thank you!

clapont, do you mean two copies of freebsd? that's basically what boot environments are

futune: oh, sorry! I meant FreeBSD + Linux (any distro)

I'd like to have FreeBSD and Linux (Arch/Debian) on the same ZFS pool, separate system datasets ofc, sharing the same /home dataset; encrypted would be the last step

he

hello, I am getting this error "can't create backend" when launching sway

I am following the handbook

what could be the reason for this error?

did you walk through all of the Wayland install and setup procedures 1st?

SponiX: I did "pkg install wayland" like handbook says

I went through Wayland install, I think

I also searched the error but the freebsd forum says it could be an issue with my GPU driver

ghodawalaaman: maybe it is an issue with your GPU driver

I have skylake GPU, (which comes in-built with cpu)

okay, and do you have a driver installed, configured, and loaded?

SponiX: here is the output of dmesg

I couldn't find GPT in dmesg, so it could be that GPU isn't detected

clapont, I have run multiple linux distributions like this from the same pool. My main worry would be that freebsd and linux would have slightly different zfs versions, and a zpool upgrade from one system would break another

secondly, if you want to mount stuff using the mountpoint property, freebsd and linux have slightly different hier conventions

like, you'd have to find a way to mount the home dataset under /home on linux and /usr/home on freebsd (or other problems along that line)

I think legacy mounts using fstab files could solve that though

that being said, I have not actually tried freebsd + linux on the same pool myself, maybe something else breaks...

futune: thank you for comments, I'm thinking over...

The start of a 'swap_pager: cannot allocate bio' - mostly graphs for now dan.langille.org/?p=8139&preview=1&_ppp=c8daebad9e

clapont, if you try it, and it works, please report back. I'm very interested.

I'm trying to read single sectors from a geli device, e.g. /dev/ada6.eli. dd if=/dev/ada6.eli gives me dd: /dev/ada6.eli: Invalid argument

i was expecting this to work because using dd to write to the device works fine, e.g. dd if=/dev/zero of=/dev/ada6.eli

does anyone know what is wrong, and how to read single sectors from geli devices?

futune geli block is 4k

do you mean that adding something like bs=4096 would make it work?

tsoome_,

i would guess so;)

woah, thanks, by golly it appears you are right, tsoome_

thanks

the thing is, encryption does add some data, and you do not want to shrink your data into too small chunks

futune: sure thing. at this moment the root zfs encryption option seems quite far; I had some play on freebsd 14.2 today with root zfs + geli/native encryption...

tsoome: thank you, interesting thing with the bs=4k for dd on geli; maybe once/twice I needed to backup a smaller encrypted partition to some other/remote place but it was luks+linux, it simply worked..

hello

how is easy anticheat support on freebsd?

i am considering booting freebsd for desktop use in the future and game compatability is a slight concern.

most of the games i play on steam either run natively or through proton with no issues.

the main game i play is the finals but there are no records on the internet that someone has gotten it working.

the finals uses easy anticheat + it's own solution on top of it.

Hi. I am using unbound from ports for DNS over TLS, recursive caching DNS from ports. I like to switched to unbound default in FreeBSD. It will work the same?

hey folks

we have a FreeBSD system, which during boot says "zio_read error: 5", which is reached at src/stand/libsa/zfs/zfsimpl.c line 2306

5 is EIO, which usually does indicate some sort of unexpected input

are there any other messages?

this line number is from either we were reading from gang blocks or we were reading from vdev. But physical IO error should have related [additional] messages.

tsoome_ ZFS I/O error, all block copies unavailable

tsoome_ we did set the BE back to the old one (which was 14.1 latest patch0

and rebooted, and we still got the zio_read error 5, however it did boot properly

yes, thats from zfs reader, but was there other errors? I guess no?

all block copies onlymeans we did try to read from all available blocks (and failed).

no not at all, but we're wondering why it would fail when the kernel was updated to 14.2 (we did do gpart bootcode manually via live system btw)

btw, does lsdev -v confirm all pool disks are visible?

tsoome_ yes we can see all disks and it also printed the ZFS pool name (zpool)

and, if you got system up, you better run scrub.

I should mention we are on GPT + BIOS mode

does the pool have redundancy?

we did run scrub using the live system, yesterday, didn't return any errors, it's made of 2 vdevs, mirrors of 2 disks.

how large disks?

4 disks, 3TB each, there's only 50G used, 2 minutes to go with the scrub to finish.

well, 3TB and BIOS... I suppose it is UEFI + CSM? you *may* hit the buggy int 13 read from large LBA issue.

I'd check with UEFI boot from the same new BE.

tsoome_ agree, my thought as well, but it does boot with 14.1, just not with 14.2, which got me worried. bug? changes? not sure.

if bios boot gets the error and uefi boot does not, then its buggy bios

HP MicroServer Gen8... well according to reddit, there's no UEFI support :(

the fact that you have multiple BE's means the data from new BE is stored in different blocks and can happen to be just far enough.

well, other option is to use zdb and verify all block pointer offsets are below the 2TB line;)

ok, so in that case you can test by creating boot pool partition small enough not to pass 2TB line. Altho, it may be the actual wrapover is not exactly on 2TB line...

ok, and other thing you can test is to copy /boot/loader from older BE (to loader.old or like) and start it from boot: prompt

if it does get the same errors, its LBA problem.

actually, you can switch BE from loader to see if you can load the kernel and modules

no need to copy anything

so after scrub, we rebooted, we still see zio_read error 5, but it boots all fine!

well, it only meant that we failed to read something non-vital

also, where the new BE fails - during the kernel load? or during the loader startup (while its reading the lua scripts etc)