Bcachefs Continues Making Progress - Finishes Big Allocator Rewrite

[phoronix]

Phoronix: Bcachefs Continues Making Progress - Finishes Big Allocator Rewrite

Bcachefs as the next-generation Linux file-system born out of the kernel's block cache code is aiming to possibly go upstream in 2022 and as a result has been trying to work through its remaining invasive changes and other big ticket items before proceeding. Bcachefs lead developer Kent Overstreet has put out another status update on this open-source file-system effort...

https://www.phoronix.com/scan.php?page=news_item&px=Bcachefs-March-2022

[cynic]

It's nice to see competion in the field of new generation filesystem.

However I won't be in a hurry to convert my btrfs volumes untile bcachefs has been battlefield tested for at least a couple of years.

[NobodyXu]

It looks to me that Bcachefs is still quite young, as it still need to release breaking changes in its ondisk format frequently.

[skeevy420]

It's nice to see competion in the field of new generation filesystem.

However I won't be in a hurry to convert my btrfs volumes untile bcachefs has been battlefield tested for at least a couple of years.

I'll consider using it for root volumes whenever it gets upstreamed. I keep most all my important information off my root on a zpool and a folder with all my tweaked /etc files so my roots are basically throwaway. It's why every other month y'all hear me talking about running a different distribution. I'd try out bcachefs now but it's already a pain in the ass keeping up with one out of tree file system for my data disks let alone a 2nd FS for root (that's the same reason I haven't tried Resier4/5 for root).

It looks to me that Bcachefs is still quite young, as it still need to release breaking changes in its ondisk format frequently.

That's why it's still out of tree. He's trying to get all the major bits and pieces in place before it's available for the masses for consumption.

[cl333r]

That's why it's still out of tree. He's trying to get all the major bits and pieces in place before it's available for the masses for consumption.

I just hope it's not another KDBUS/BUS1 project where there's a lot of hype for a few years, video presentation talks and then it gradually dies into oblivion.

[cynic]

That's why it's still out of tree. He's trying to get all the major bits and pieces in place before it's available for the masses for consumption.

that's the idea, but we all know that a lot of bugs will be only discovered when "the masses" start to using it.

[flower]

I'll switch once it is available in a LTS kernel.
esp tiered storage and different replication methods per folder is what interests me.

but well... i am prepared for bugs and i do have good backups.

[⤐⤐⤐⤐]

I want BcacheFS to add a missing feature to Linux ecosystem, a third kind of file link, an absolute-relative bridge link concept (relatibolute link), so if the users have ~ (tilde) starting symbolic links (tilde starting symbolic links) that point to the home directory using tilde as part of file path they work if the user put them inside this third kind of link, as instance a link containing/pointing to ~/file can become portable to any account the user have independent of the point the home folder is itself positionated inside filesystem hierarchy/extructure (deepness).

Other thing I want to see tackled by BcacheFS is to accept "/" as part of filenames as escapable (or especial case) caracter in some way.

Levarage it with more capabilities beyond of what is already implemented by others FS and it will gain traction.

[M@yeulC]

Usual reminder that his bcachefs work is funded on patreon: https://www.patreon.com/bcachefs

[geearf]

I just hope it's not another KDBUS/BUS1 project where there's a lot of hype for a few years, video presentation talks and then it gradually dies into oblivion.

That's kind of what he did with bcache, but since bcachefs is suppose to render bcache useless, maybe that's ok.