Btrfs With Linux 6.13 Delivers Performance Improvements & Other Features

[Britoid]

Would be really interesting to see how OpenZFS fares again BTRFS since it wasn't included in the last round (6.11)

Looks like the last ZFS benchmark was done in 2019? Huh

• https://www.phoronix.com/search/zfs
• https://www.phoronix.com/search/btrfs

Post was comparing Linux-filesystems, which ZFS is not

[Jonjolt]

I'm wondering why CachyOS defaults to Btrfs? It seems to be the slowest of all the file systems to choose from?

Snapshotting and roll back, also these benchmarks https://www.phoronix.com/review/linux-611-filesystems AFAIK have the Copy On Write Enabled, usually the database already has those functions making it redundant.

[Kjell]

I'm wondering why CachyOS defaults to Btrfs? It seems to be the slowest of all the file systems to choose from?

ZFS might not make much sense when you have to wait months for it to be compatible with mainline kernel

[pkese]

I'm wondering why CachyOS defaults to Btrfs? It seems to be the slowest of all the file systems to choose from?

Extra features (compression, deduplication, snapshots, subvolumes, data checksumming, etc) more than outweigh the 10% loss in performance.
Anyway, ever since we switched from mechanical drives to SSDs, I/O performance is not an issue for a normal user any more.

https://www.youtube.com/watch?v=4DKLA7w9eeA

[muncrief]

ZFS might not make much sense when you have to wait months for it to be compatible with mainline kernel

The CachyOS devs work closely with the OpenZFS devs so that ZFS is compatible on day 1 of kernel releases. I was concerned about this at first but after running CachyOS with ZFS data disks for almost a year on my workstation and media server I've never had any errors or problems related to it. As I understand it all new features may not be available immediately, but ZFS remains compatible with kernel changes. However this will have to be confirmed by a CachyOS dev as I'm not certain exactly how they do it, but it has been truly amazing.

The bottom line is that if you want to use OpenZFS with a rolling release CachyOS is the way to go. I don't know of any other distro with such well maintained integrated support. With its ZFS support and optimized packages it's basically Arch on steroids

[pWe00Iri3e7Z9lHOX2Qx]

The CachyOS devs work closely with the OpenZFS devs so that ZFS is compatible on day 1 of kernel releases. I was concerned about this at first but after running CachyOS with ZFS data disks for almost a year on my workstation and media server I've never had any errors or problems related to it. As I understand it all new features may not be available immediately, but ZFS remains compatible with kernel changes. However this will have to be confirmed by a CachyOS dev as I'm not certain exactly how they do it, but it has been truly amazing.

The bottom line is that if you want to use OpenZFS with a rolling release CachyOS is the way to go. I don't know of any other distro with such well maintained integrated support. With its ZFS support and optimized packages it's basically Arch on steroids

No. CachyOS has no magic power to make ZFS work with mainline kernels. The only thing they can do is cherry pick early compat changes, just like everyone else trying to use ZFS with mainline. This is the same thing that the OpenSUSE ZFS maintainer does, and the same thing that people do when rolling their own. And ZFS does not "remain compatible" between kernel changes. Those early compat changes also can / do have bugs. If you care about your data, you shouldn't be relying on them, especially on a server.

[EmanuC]

AFAIK it was abandoned. There were several changes introduced and that's mostly it. Therefore it seems to be dead-end now.

Is that your belief, or do you have a reliable source to back up what you're saying?
As far as I know, Josef is still working on it.

As for extent tree v2, yes I'm actively working on it. With the scope of the project the design has had to change while I was developing it and discovering flaws in certain areas. I hope to be code complete in the next couple of months.

[muncrief]

No. CachyOS has no magic power to make ZFS work with mainline kernels. The only thing they can do is cherry pick early compat changes, just like everyone else trying to use ZFS with mainline. This is the same thing that the OpenSUSE ZFS maintainer does, and the same thing that people do when rolling their own. And ZFS does not "remain compatible" between kernel changes. Those early compat changes also can / do have bugs. If you care about your data, you shouldn't be relying on them, especially on a server.

Well, it can't be any worse than btrfs or bcachefs, which are both known for data corruption and other errors. And my goodness, before switching to OpenZFS I tried btrfs on all my data disks and it was so incredibly slow that I had to switch back to ext4. But thankfully CachyOS came along and solved my COW file system dilemma.

And hey, if it's good enough for OpenSUSE I'm confident it's good enough for me. The fact is that OpenZFS simply has no competition when it comes to features and performance. And while it's regrettable that Linux will not allow it into the mainline, if you want or need those features and performance OpenZFS is the only way to go, and CachyOS makes it effortless to do so.

[ultimA]

No. CachyOS has no magic power to make ZFS work with mainline kernels. The only thing they can do is cherry pick early compat changes, just like everyone else trying to use ZFS with mainline. This is the same thing that the OpenSUSE ZFS maintainer does, and the same thing that people do when rolling their own. And ZFS does not "remain compatible" between kernel changes. Those early compat changes also can / do have bugs. If you care about your data, you shouldn't be relying on them, especially on a server.

For server that's not a problem in practice. If you are running a server, then you're also not running CachyOS or any other distro that lives on a bleeding edge kernel.

[polarathene]

Well, it can't be any worse than btrfs or bcachefs, which are both known for data corruption and other errors.

OpenZFS has had a data corruption issue in the past year IIRC related to reflink support (which I think they reverted). Errors wise I recall something related to hibernation or resuming from suspend? Neither of these may be relevant to you, but none of the FS are 100% immune to these concerns.