Linux 6.8-rc4 Released With Bcachefs & NTFS3 File-System Fixes, Transmeta Crusoe Fix

[fitzie]

The speed of ext4/xfs with reliability and robustness ahead of features is supposed to be the sales pitch as far as I can gather. But as I see it there is a lot of (admittedly) cool features talked about, but not so many reliability features implemented yet as far as I can tell. Scrubbing for example is missing.

I am myself biased towards btrfs, but bcachefs plans for many of the features I as a btrfs user would love to have so it will be interesting to see how it eventually turns out in the long run. Regardless of what filesystem you choose: have tested backups if you value your data.

it's not really promising the speed of xfs, but it is promising to have all the features of modern filesystems with the smallest performance penalty possible. and performance isn't always measured on a simple benchmark but also just avoiding pathological slowdowns, like operations when you have tens of thousands of snapshots.

I think you can expect the next two years to be early adoptors to shake out some of the kill your filesystem bugs, and get some of the essential features in place. then another few years after that to get some serious interest from a hyperscaler. if a hyperscaler "adopts" bcachefs, then probably another 3-4 years to get it in shape.

As much as kent focuses on performance, the top priorities were correctness and maintaining the codebase, there's plenty of things that will have to be optimized over and over as the codebase matures.,

[EmanuC]

it's not really promising the speed of xfs, but it is promising to have all the features of modern filesystems with the smallest performance penalty possible. and performance isn't always measured on a simple benchmark but also just avoiding pathological slowdowns, like operations when you have tens of thousands of snapshots.

There's nothing exclusive here about Bcachefs, I think any filesystem developer has that as their goal.
If you are comparing with Btrfs, this will be fixed:

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.

https://github.com/btrfs/btrfs-todo/issues/25

[Quackdoc]

So, what actually is bcachefs's sales pitch? Is it the same but better than btrfs? EXT4 killer? All new and shiney tech to reduce staleness for future proofing?

so far for me it's btrfs that doesn't eat my data, compression and dedupe are working as expected which is very nice. I have it running on my game and VM NVMe for a bit now, working better then expected, but I am also using a DKMS for it which is rather risky I realize