Thread 106558895

>intentionally breaks zfs compat
why are people like this? there seems to be a lot of intentional sabotage in the kernel lately.

Why would they remove shit? Just keep it for backwards compatibility. It works for Windows (90% market share btw).

>>106558895 (OP)
Because there are "people" who see that software freedom is the largest gate towards greater and broader freedoms.

I remember kernel devs also trying to sabotage the Rust work.

>>106558975
no that's just the rust community

>>106558895 (OP)
I remember that commie faggot Linus claiming ``you never break userspace''.

>>106558895 (OP)
fucking based. non GPL modules should be sued anyway. nvidia, openzfs, all of them are violating the GPL.

>>106559052
>use out of tree code
>it breaks
what did you expect?
Linus would call you a stupid faggot

no, not like this... my reiserfs out-of-tree work

>>106558895 (OP)
Exactly how does this change affects the license you are using for your software? I don't get I don't what non-GPL has to do with removing a function

>>106559078
Hi Linus

>>106559078
literally every post with that character combination is actually retarded.

>>106559106
it means that the licences are incompatible so you can't distribute the module with the kernel, you need to compile the module seprately and then dynamically link it afterwards, but they are going to remove a dependency so that the dynamic link will fail for just one specific module with their middle finger up because it uses an incompatible license anyways

>>106559106
I don't get it I don't understand*

>>106558965
OpenZFS is free and open-source software licensed under the CDDL, which is recognized by the FSF and a free software license
breaking the kernel to make it more difficult or impossible to use OpenZFS does nothing for the cause of software freedom, and anyone who says otherwise is a delusional zealot

>>106558895 (OP)
Imagine not locking your kernel version when Linux handed it over the a tranny. Imagine still not doing it once they started to RIIR.
My 4.19 is still working just fine.
I could probably go down to 4.14, maybe 4.9 with my hardware, but I'm fine.

>>106559165
>does nothing for the cause of software freedom
That's because they aren't working for, but rather against all freedoms.

>>106558895 (OP)
Because there are the stallman type retards that think it's their duty to make it harder to make non free software run on a "free" os

Meanwhile bsdchads gets to use zfs without any violations.

i hope truenas goes back to bsd

>>106558895 (OP)
Incredible, this is textbook harmful big corpo behavior. The sort of thing that Microsoft, Apple or Google would do.

>>106559052
Kernel drivers aren't user space, dummy.

>>106558895 (OP)
>stupid out-of-tree driver uses deprecated interface
>deprecated interface gets removed
>surprised_pikachu.jpg

Go back to your non-GPL, ZFS-built-in OS, retards.
Oh wait, you can't!

>>106560094
>back to bsd
lmao

>>106558895 (OP)
it's not as if they didn't see that coming. Every year that passes I get the feeling that there should be a stricter control on what gets merged in the kernel or not. I feel like in the last few years the number of contributors grew way too quickly for the maintainers' abilities to filter out the cruft. The only reason why I have some home is that I know that Windows will only get even worse with time, so I think Windows is safe against that, but MacOS seems to be taking over. With Apple using iPhone hardware on the I don't think it is much longer until the release a $500 laptop that absolutely crushes the competition and completely takes over the low-midrange laptop crowd, which is probably the highest volume sector of the consumer PC market.

>>106558895 (OP)
Gold/Platinum Linux Foundation sponsors already license their kernel code as GPL, so it doesn't affect them
It seems to be just to ruin any competition and make things more miserable
You WILL use btrfs
You WILL lose data on power outage
And you WILL like corporational freetard software

>>106561449
ext4 just fucking works, what’s the deal with these other meme filesystems?

>>106561498
Will be removed soon, mark my words

Less than 10 hours before I can preorder my iPhone Air. We hype apple bros?

>>106558895 (OP)
Based, non-gpl out of tree faggots must eat shit.

>I want to use Linux to play gaymes or something
Choose btrfs, ext4, or xfs. That is more than enough.
>I want to use ZFS for my home server
Run a real Unix like Illumos, FreeBSD, or NetBSD.

Simple as.

>>106558895 (OP)
>Christoph Hellwig
kek, of course it's that retarded autist.

>>106563160
>real Unix
Unix is dead and buried.
>ZFS on [anything but Linux]
No such thing.

>>106558895 (OP)
GPL faggotry is unironically a mental illness.

>>106558895 (OP)
It's understandable. If nothing in-tree uses it anymore, why should they be forced to maintain it for the sake of out-of-tree users who refuse to even comply with the GPL?

>>106558895 (OP)
>yet again fucking with zfs on linux but this time because of muh licensing mental illness
>muh butter still corrupts data when using RAID5/6
zfs will just continue to work around the next attempt to fuck them over like they have the dozens before meanwhile all license trannies will inevitably get the gas chamber

>>106563740
This. I'd sooner drop linux for *BSD than I'd drop zfs for anything else.
No ceph is not the same and not a replacement either, I run both ceph and zfs pools, they're meant for completely different use cases.

>>106559790
that makes no sense.
cddl is copyleft. if bsd-chads would actually allow zfs inside their kernel instead of using a shim, their bsd kernel would become cddl. bsd-chads would henceforth be known as cddl-chads.

I fucking hate freetards man, they ruin everything

>>106558905
There is no backwards compatibility broken here.

>>106558895 (OP)
first reiserfs
then bcachefs
now zfs

i'm starting to notice

why do people use meme FS like ZFS instead of Btrfs?

>>106561498
>Ext4 just fucking wo--
>truncated files on crash
>weird behavior like renaming over a copy of a file causing implicit fdatasync()
>not recommended by multiple database engines and RHEL
only a fucking retard uses Ext4 in 2025.

>>106563923
ya? that's why BSD just lets idiots make their own implicit cddl "forks" because the licensing term of BSDs allow it.

>106564532
>Btrfs
>still has data corruption issues with RAID5/6
>literally added warnings against using RAID5/6
>meme FS like ZFS
bait used to be believable

>BTRFS
>Wayland
>GNOME (Vanilla, ZERO Extensions)
>Fedora Silverblue
>Flatpaks (from Flathub)
If you are not using these then you are NOT a true gigachad. In fact, if you are NOT using these then you are a retard. It is that simple.

>>106565161
no one gives a fuck about raid 5/6
raid 10 is where its at and btrfs works good with that
zfs is a meme with their cuck licensing
anybody using zfs over btrfs is a fucking idiot and thats a fact

>>106558895 (OP)
No real suprise here. RedHat has been pushing Linux to kill off zfs for a better part of a decade now to make people switch over to xfs + stratis

>butterfs is so garbage they have to actively sabotage other filesystems to force people over to it
You can't make this shit up lmao

>>106558895 (OP)
>>106558905

>>106563494
>>106564010
Even though this sounds like a gpl thing, and practically it is, this is all probably linus being a major dick against the bcachefs dude after making his fs out of tree.
>>106564532
zfs is superior to btrfs

>>106565401
I think redhat is moving towards btrfs since fedora has been defaulting to it.
If they started testing it then it might showup, atleast as an option, on rhel11.

>>106565473
Highly doubt. btrfs got added into fedora for atomics where they require it to be able to easily have the snapshot and rollback functionality. btrfs was fully moved in RHEL 8. No way they are bringing it back. RHEL is heavily invested in XFS + Stratis

>>106565461
>Even though this sounds like a gpl thing
It's not. The whole "GPL" issue only applies to allowing these filesystems be distributed with the kernel (in-tree). Exposing functionality like "write_cache_pages" doesn't break GPL in any way. Removing it is just a direct move to breaking any out of tree filesystems on purpose. There's zero reason to remove it from a licensing perspective. It's purely a power play to only allow things in house and lock down linux.

https://arstechnica.com/gadgets/2020/01/linus-torvalds-zfs-statements-arent-right-heres-the-straight-dope/
Linus made it very clear that he doesn't want people using zfs and intended to kill it 5 years ago.
Also this affects companies like TrueNAS big time. We're either going to see a shift of companies like them forking linux, going to BSD, or being forced to btrfs (which is likely what linus wants)

>>106558895 (OP)
WTF IS WRONG WITH THE RECENT KERNEL RELEASES
Every 6.16.x release has some major thing breaking, 6.15 just kinda works (i am still unsure about networking, it does feel a bit off).
I consider downgrading to 6.14 or earlier, depending on what version was the last one to actually work, and then freezing that.

I don't understand any of this
but wouldn't they try to get R5/6 stable on BTRFS before bullying a working solution out

>>106565666
>but wouldn't they try to get R5/6 stable on BTRFS before bullying a working solution out
They did try, turns out they're not just exceptionally malicious they're also exceptionally incompetent.

>>106565666
R5 has been broken in BTRFS for basically a decade. It's never getting fixed. Just accept it.

>>106558895 (OP)
>This patch from Christoph Hellwig
That's the guy that resigned from the DMA team after trying to block Rust integration and Linus told him to go fuck himself kek

>>106564010
why are they doing this?

>>106566107
idk but the sequence of events suggest its not mere chance.
it sure does look like an effort to reduce choice.

>>106566107
>>106566141
nobody was maintaining reiserfs
bcachefs main dev kept breaking CoC and no one, including Linus, wanted to put up with him anymore
ZFS has always been hated by linus so it's really no surprise. This isn't the first time he tried to kill it btw.

>>106566163
whatever.
fact remains that choice is getting slimmer..
i honestly dont give a fuck about what linus likes or not.

zfs is a great fs.

we also need continued development of bcachefs AND good integration within the kernel.

by reducing choice the users pay the toll

>>106564552
>>106564552
>>not recommended by multiple database engines
Which?
>and RHEL
You mean, the only company on Earth developing and using XFS?
Gee, it's almost as if they were the least qualified to have an opinion on filesystems or something.

>>106566107
redhat is trying to take over the kernel. embrace, extend, extinguish.

>>106566163
>bcachefs main dev kept breaking CoC
omg shut him down
>ZFS has always been hated by linus so it's really no surprise. This isn't the first time he tried to kill it btw.
what's his problem with ZFS? I've been wanting to use it for a while.

>>106566226
>i honestly dont give a fuck about what linus likes or not.
yeah well he's in charge unfortunately
>we also need continued development of bcachefs AND good integration within the kernel.
bcachefs is still maintained, just externally now. Also the change mentioned in OP doesn't affect bcachefs as it can use the GPL exports.
>>106566426
>what's his problem with ZFS? I've been wanting to use it for a while.
It uses a custom non-compat GPL license. Period. He has no technical arguments against it. He thinks it's a meme (literally) and his core issue is the license is incompatible which means it will never ever ever be in the kernel. Period.

>>106563160
I use linux for all my computer activities and none of the filesystem you mentioned come anywhere close to zfs.

>>106566459
> he thinks zfs is a meme
> but btrfs somehow is not
absolute_retardation.jpg

>>106566459
whatever. its not a good year for file systems. i hope this is the last issue for a long time.

>>106566498
He uses Fedora and ext4. What do you expect? I'm not even joking he doesn't give a fuck about non-ext4 fs besides having a say in their code reviews.

>>106565666
>>106565702
What's more funny is btrfs devs literally copied what bcachefs did for raid5 and still got it wrong.

>>106566533

i currently use btrfs because kernel updates are handled automatically.
otherwise i'd use zfs.

is anyone running bcachefs here?

I'm currently going all-in on ZFS, so while somewhat worrying, I don't think this will affect me until I have to upgrade to, worst case scenario, RHEL 11. I'm still on RHEL 9 and intend to be on it for another couple of years at least. I should hope there will be a solution to this long before it can affect me.

ugh
why can't we have ORACLE LINUX with FIRST PARTY ZFS SUPPORT via UEK??
the filesystem situation in Linux is in disarray (haha), why can't Oracle step in and seize the market?

LARRY!!
ANSWER ME LARRY!!

>>106566625
sorry sir, larry is busy counting his money and racing boats. he'll get back to you.

>>106566625
Go back to the original ufs and make the occasional backup.

>>106566773
At that point, I'd rather lock the kernel version

i'm using linux 20 years now and soon as you want to do something thats not out of the box, you have to jump at least one hoop.
its not only a linux problem but on the other systems they make it harder deliberatly. it shouldn't be like that on linux. there should only be technical reasons for having to screw with it and not for political reasons.
its tiresome.

>>106567470
The technical reason is
>nothing in-tree is using these functions so we’re going to remove them and we have no responsibility to care about out-of-tree modules also this one legally can't be in-tree because the maintainer refuses to use GPL
but I’m sure you already knew that going in. (the real reason is Linus’ hate boner against ZFS)

>>106565461
Nope. This change doesn't affect bcachefs at all.
It was already once mainlined, the codebase is already GPLv2.
It's purely a GPL issue. They removed code that mainline filesystems migrated away from that ZFS depends on.
ZFS cannot yoink the old code because it's incompatible with GPL.
Linus is just cleaning up the mainline repo, and ZFS takes collateral damage. That's all it is.

>>106567557
of course. its obvious that there is some kind of undercurrent of change going on in linux especially file systems. i can feel it.
linus knows something we don't. with file system stuff, he does things but does not elaborate except the usual reason which is some kind of spat between developers but i think thats only half of the story.

they are pushing major changes for how memory is handled.
then you have valve developing on linux from now on.
the corpo-take over is amplifying.
there is a lot going on.

umm so why can't anyone take the to-be-removed code and make it into a GPL kernel module which you would then load if you wanted to use ZFS?

>>106566107
just use btrfs
stop complaining you entitled bitch
we do all this work for you for free and somehow its never enough
just use it
just use btrfs bitch

>>106565494
>Exposing functionality like "write_cache_pages" doesn't break GPL in any way.
Well I don't know how this exact situation works, but I read discussions about the GPL, and it seems I'm basically not allowed to do anything with it. Like, these retards even say I can't design a program that loads another GPL program as a library and use its functions (meaning I would only use its interface without any of the actual code) without pozzing my own program too because muh ????.

>>106568005
xfs + stratis is better than btrfs. There is zero reasons to use btrfs.

>>106568005
>no ARC
no thanks

OpenZFS devs already found a workaround. Linus can't silence ZFS.

>>106568042
Meanwhile in reality their lawsuit would get dismissed immediately anyways because boomers. Even assuming they got a court hearing somehow just think of the average boomer, now imagine them being 2000x as disconnected and sheltered from reality and about 500x less tech literate than you average boomer and you have your average judge.
>So you're saying they are using your code... that you published online for free... and that is... against the rules.. you made up?
They wouldn't even get laughed at, the judge would straight up order them to be sectioned immediately before they can hurt themselves or others. And before anyone starts coping about the three ""successful"" 2000s lawsuits go ahead and read them or better yet have a lawyer explain what they actually mean because they don't have the legal implications you think they do. Not even close.

>>106568005
btrfs is buggy. the ONLY linux fs I've ever had self destruct.

it has been fixed already
github /openzfs/zfs/compare/master...robn:zfs:linux-6.18-compat
by this handsome gentleman
social.lol @robn/115189135969338619

>>106568005
When bcachefs got kicked out of the kernel last month even its dev was making fun of how btrfs is still a buggy piece of shit and other people in the thread reluctantly agreed with him to a degree lmao. Facebook apparently uses a modified btrfs to even make it work.

>>106568005
sorry not using an fs that had data-losing bugs ship to distros, >>106568250 i would trust even bcachefs over the buggy piece of shit that is btrfs

What even is the point of stuff like ZFS and btrfs anymore?

f you have data that actually matters, you want multi-host replication at the application or storage subsystem level, ideally geographically separated, and for boot volumes, you should be using immutable snapshots and for that you want something read-only and rock-solid like plain old multiple partitions or even redundant boot devices that don't require complicated drivers. The only use cases I can imagine for them is decrepit organizations that refuse to update their software, in which case, they are already paying IBM or someone to maintain some propreitary mainframe architecture specifically for them, so they'd have no interest in FOSS projects, as well as hobbyists and tinkerers who haven't taken it upon themselves to git gud and build a homelab storage cluster yet.

>>106569409
Because even if al you said is true, you are talking about a higher level.
File system is lower level.

>>106569409
>build a homelab storage cluster yet.
Why would i introduce all that overhead instead of just using zfs

>>106569456
Sure, but why would you want anything except the simplest and highest performance fs in that case? What purpose do you even have for a complicated mulit-volume, checksumming filesystem like ZFS if you could just use a simple, well-supported fs on top of a storage system that provides those already? Why bother with a complicated journaling system if you are doing WAL shipping and synchronous replication at the application level (e.g. postgres) where you can delete and reinitialize an instance if anything fails?
>>106569499
Because Linus is trying to kill it, if the anons in this thread are to be believed.

>>106568177
Yeah, it's absolutely possible it would get thrown out, but we have seen some retarded ass decisions about software being made in court. It also doesn't help the FSF and their cultists try to force their retarded views about the interpretation about their Bibl--I mean the GPL. There's a QnA page on the website of the FSF detailing their mental gymnastics, and honestly, they are so strict about dynamic linking, library loading, and IPC that even if they wouldn't hold up in court, the only thing one can do is not touch anything GPL with a 10 foot pole. This is why I swore never to release a single code under any FSF license.

>>106569542
*single line of code

>>106569409
>f you have data that actually matters, you want multi-host replication at the application or storage subsystem level, ideally geographically separated,
Something ZFS can do with replication since for fucking ever
>and for boot volumes, you should be using immutable snapshots and for that you want something read-only and rock-solid like plain old multiple partitions or even redundant boot devices that don't require complicated drivers.
ZFS is not complicated and provides exactly that for over a decade now. Just about every NAS company uses ZFS for good reasons, and best of all it's self healing if you have multiple disks or multi-copies without any intervention.

The rest of your post is irrelevant. Using ext4 with backups isn't going to prevent ext4 from fucking up your data, or your backups on the other hosts if one of many situations occurs where ext4 can corrupt data. You're an absolute fucking muppet if you actually use ext4 for backups.

>>106558895 (OP)
what's wrong with doing things the Windows way and just munging through undocumented internal data structures to do what you want
git 'er done, doesn't matter if you're supposed to or not, you run in kernel mode so you are the king

>>106559106
because third-party modules have access to a reduced number of "exposed" symbols from the kernel, if symbols are exposed as gpl symbols then they can only be used in gpl-compatible modules.
it's not really an issue for proprietary modules that are never released, change the linux code and export symbol as non-gpl ones then compile it yourself, who gives a fuck anyways, that's what we do at work and I know for a fact that most if not all globohomos also do it.
https://docs.kernel.org/core-api/symbol-namespaces.html

gpl and freetardism is pure cancer, it serves no purpose other than to throttle progress and to prevent people from doing anything useful with others code.
linux is gplv2 and it's a mistake, it succeed because unix at the time was proprietary, if one popular unix system was open under a permissive licence then linux would not have existed at all and we would have much more advanced OSes

>>106568130
thank god

>>106561498
>>106561505
if ext4 goes away f2fs is basically drop-in replacement without retarded fluff on top. It has similar performance characteristics while at the same time has wears out SSDs and other flash storage less.

>>106568005
>we

>>106561505
no its fucking not
at best ext2 and ext3
>>106571178
kys mouthbreathing retard i wont elaborate why fuck you

I use Fedora and btrfs just works.

>>106570503
>gpl and freetardism is pure cancer, it serves no purpos
Then why do you use it? Why is there no proprietary software you can buy and enjoy instead?
Shouldn't the superior commercial softwate totally crush it? Why can't it compete? Isn't this the free market?

>>106568057
RedHat shill spotted.
Enjoy you broken crap no-checksums-for-metadata xfs.
Reminder that even RedHats own Fedora decided against RedHats xfs and went for btrfs instead.

>>106568130
wow that took long lmao

>>106569537
>Sure, but why would you want anything except the simplest and highest performance fs in that case?
Because at enterprise level, those things are a fundamental problem. Manuallly unfucking the file system in userspace is BAD for performance and troughput, plus FS bugs.
But again, at enterprise level its good enough that your benchmark can beat vanilla Win XP NTFS.

The reason its a problem is in the end that
1.FS makers seem to be generally insane people
2. From a enterprise background you might go "power outtages? system crashing? lmao"
3. A lot of t he deeper intricacies of scaling and troughput are things you don't encounter at the hobbyist level
4. If something really fancy comes around, its going to be patched into the next Ext system. Rember that
"ext4 was initially a series of backward-compatible extensions to ext3, many of them originally developed by Cluster File Systems for the Lustre file system between 2003 and 2006"

>>106561382
It will have 4gb of ram, and to get 16 gb will cost another $400, and to get 32 gb will cost $800.

Hackers will find a way to add ram for like $20 and apple wont care because 0% of the users will do a gray market upgrade.

>>106564010
>not using Storage Spaces™
What are you a poorfag or something?

>>106569409
>if you don't escribe your data into diamonds and burry them in a vault 2km under the lunar surface you obviously don't need that data and should just delete it now

Wut?

>>106565598
That has always been a thing. There is always something broken in each release, so wait to update till several patch releases are out.

>>106571250
>Then why do you use it?
no alternative
>Why is there no proprietary software you can buy and enjoy instead?
linux exists and it would be better with a permissive licence
>Shouldn't the superior commercial softwate totally crush it?
why
>Why can't it compete?
because linux is free
>Isn't this the free market?
exactly

not sure what point you're trying to make, I agree with you and still think linux adoption happened because it was the only alternative to windows at the time. it's the most used and supported kernel so that's what I use, I'm not a luddite.

>>106565461
>zfs is superior
>can't be used on latest kernels
>may corrupt data at anytime, including when upgrading kernels
>low quality, out of tree codebase even Oracle doesn't touch

Why do people believe the zfs memes?

>>106558895 (OP)
Not my problem
t. ext4 user, don't plan on migrating

The last version of truenas on bsd runs great on my old file server.
For newer hardware running a newer version of bsd, XigmaNas may be the easy solution to install and play if you don't want to manually setup bsd with zfs and all the different services.

am I a retard for using jfs

>>106575536
i've heard of it, but i've never heard of someone using it. i'm curious what made you chose it

>>106575536
>>106575799
-- i mean, from my perspective it was at best a competitor to ext3, but i can't imagine a reason to use it over ext4, or any of the newer options

>>106575799
I think at the time like 10 years ago because I heard it was good if you had many small files

>>106575536
>>106575833
We used it many years ago because ext3 wasn't coping with 1 million files in a single directory.
Then ext4 came out and we reinstalled all the Debian/jfs servers with Ubuntu 8.04/ext4.
Never seen jfs used anywhere after that.

>>106575536
I thought that was XFS

>>106568250
Slightly OT but didn’t bcachefs remove the bcache part? Why was this still an interesting option?

>>106567745
Get cracking.

>>106572680
I’m building a new 100TB+ NAS with a bunch of Optane devices for caching and was going to go with mirror accelerated parity on storage spaces because it had the best chance of performing well for my use cases vs ZFS whose devs tell you to fuck off when you want a real write-cache solution. Unfortunately, REFS fails to correct bad blocks in certain situations making their version of checksum (integrity stream) self healing unreliable. Another feature I would prefer is their scheduled dedup instead of the realtime-only dedup with zfs which destroys performance for very little gain in most cases.

Microsoft was made aware of the self healing issue (an msft project manager communicated with one of the people that tested it) and chose to delete the feedback post describing it after 3years without reason.

>>106568250
And yet bcachefs literally kept catching heat from Linus for breaking the filesystem and shipping them. Taking the word of an lolcow about anything is pretty retarded.

>>106578469
Because it was GPL and the closest working replacement to ZFS they've had to date. It's much better than btrfs, but unfortunately the main dev is a complete dick and no one likes hims. He also told Linus to fuck off when he asked to review the bcachefs code quite a few times and that's why Linus told him to take a hike after dealing with it for these past few years.

>>106566089
>he's now conspiring to kill linux from the inside
kek based

>>106571250
kek the wagie could do absolutely nothing in the face of your simple and elegant logic