Bcachefs Fixes Deadlock In Recovery, More Fixes Coming

[phoronix]

Phoronix: Bcachefs Fixes Deadlock In Recovery, More Fixes Coming

Following the revised Bcachefs code making it into Linux 6.9 and a warning for Bcachefs multi-device users to move past Linux 6.7, a batch of fixes was merged overnight for Linux 6.9 while further fixes to this file-system are expected soon...

https://www.phoronix.com/news/Bcachefs-Fixes-For-Linux-6.9

[Britoid]

I think people are going to find out bcachefs isnt the btrfs killer that's its been hyped up over the years as.

[waxhead]

I think people are going to find out bcachefs isnt the btrfs killer that's its been hyped up over the years as.

I doubt it will be a btrfs "killer" , but I am sure it will become a worthy alternative at some point in the future.

It will probably take a while before it becomes decently stable and btrfs will obviously continue to evolve during that time. Maybe then the feature gap will be so small and the storage devices will be so fast the the only god reason to choose one of the other is personal taste or the specific implementation of very weird features that not that many rely on.

[dralley]

I think people are overreacting a tad to a bug in a complex feature (multi-device filesystems) in the very first public version of a filesystem explicitly marked "experimental" that was fixed in 2 weeks and doesn't result in data loss.

People ragged on BTRFS because it was marked stable by the authors way too early and yet had multiple serious data loss bugs, some of which remain to this day. Bcachefs isn't even to the point where it can fuck up that badly yet.

However I do think he should probably drop that one snipe from the website ("the COW filesystem that doesn't eat your data). It was added years ago, BTRFS isn't that bad anymore, and it feels a bit like tempting fate.

[pWe00Iri3e7Z9lHOX2Qx]

Maybe then the feature gap will be so small and the storage devices will be so fast the the only god reason to choose one of the other is personal taste or the specific implementation of very weird features that not that many rely on.

I think the one big differentiator will be tiered storage. Even that won't matter for a lot of people who only have a laptop. But for any desktop users that have room for at least one HDD in addition to one or more SSDs, being able to combine them into a tiered pool will be nice.

[varikonniemi]

exactly as i previously wrote, this "catastrophic" bug has not eaten anyone's data unless they personally commanded it to be so. It still is the COW filesystem that won't eat your data.

[waxhead]

I think the one big differentiator will be tiered storage. Even that won't matter for a lot of people who only have a laptop. But for any desktop users that have room for at least one HDD in addition to one or more SSDs, being able to combine them into a tiered pool will be nice.

The BTRFS extent tree v2 should (as far as I know) lay the foundation for being able to do things like that. The first step would be separate storage profiles for subvolumes. As you can guess I am quite biased towards btrfs, but I admit that bcachefs at least on paper have a couple of really interesting features.

[lowflyer]

make sure no one looses data,

Yeah, block all these insideous guys that go on loosing other peoples data...

[intelfx]

make sure no one looses data,

Yeah, block all these insideous guys that go on loosing other peoples data...

I chuckled. Nice.

that said, who are all those guys on the insid_e_?

[horizonbrave]

I think people are going to find out bcachefs isnt the btrfs killer that's its been hyped up over the years as.

I think people are going to find out you just post meaningless shit seemingly for the purpose of stirring shit or anyway adding NOTHING to the conversation. Goof effort, congrats!