Bcachefs Under Review With All Known Blockers Resolved

[phoronix]

Phoronix: Bcachefs Under Review With All Known Blockers Resolved

Kent Overstreet who developed the Bcachefs file-system out of the Linux kernel's block cache code has sent out the latest patches for review and to also serve as a possible pull request for mainlining the code...

http://www.phoronix.com/scan.php?page=news_item&px=Bcachefs-December-2020

[anarki2]

Fingers crossed!

[mdedetrich]

Hope this gets through, bcachefs has real hope of being an actual competitor to ZFS for Linux rather than btrfs which was rushed out too fast (giving it a bad name with all of the reliability issues in the early days) and it still cannot handle certain features (such as RAID 5/6) due to fundamental design issues.

[pal666]

Hope this gets through, bcachefs has real hope of being an actual competitor to ZFS for Linux rather than btrfs which was rushed out too fast (giving it a bad name with all of the reliability issues in the early days) and it still cannot handle certain features (such as RAID 5/6) due to fundamental design issues.

i see nice intersection between bcachefs and rust fanbases. i guess it's "gullible fools". how stupid you have to be to hype some unmerged alpha-state code because btrfs had bugs decade ago?

[zxy_thf]

i see nice intersection between bcachefs and rust fanbases. i guess it's "gullible fools". how stupid you have to be to hype some unmerged alpha-state code because btrfs had bugs decade ago?

I think it's nature to look for alternative solutions since RAID 5/6 on btrfs have been broken for almost a decade, and apparently its developers don't care about it.

[timofonic]

i see nice intersection between bcachefs and rust fanbases. i guess it's "gullible fools". how stupid you have to be to hype some unmerged alpha-state code because btrfs had bugs decade ago?

What's so bad about Bcachefs? I see some good points:

1. New approach at a filesystem, innovative.
2. Not named after a personality, so no issue if the original developer was a killer or a raper.
3. RAID support.
4. Caching.
5. It's based on a mature code base.
6. Replication.
7. Encryption.
8. Scalable.
9. Tiering/Writeback support.
10. Full data and metadata checksumming.
11. Snapshotting.

And it's a competitor to both ZFS and Btrfs, aiming at EXT4 speed.


Seriously, I would like a well explained argument instead only bashing against new projects.

[JackLilhammers]

i see nice intersection between bcachefs and rust fanbases. i guess it's "gullible fools". how stupid you have to be to hype some unmerged alpha-state code because btrfs had bugs decade ago?

Wow, it's amazing how toxic you are. I've been reading your comments in the last few days and in each thread you posted on you managed to be awful to someone.
Do you work on this kind of attitude or are you just a natural?

BTW, I'm in no hurry to try bcachefs, and once it's mainlined we'll see if it's really good how its fans think. In the meantime if you have any data on why other people shouldn't be hyped, please share it

[Anarchy]

Nice. This is a big step up for bcachefs & Overstreet, I know he's been working hard on this for a long time.

Looking forward to future updates, I hope the project expands and develops more in the future. It will be much easier for him & testers if this is part of the kernel.

[rekrek]

Getting into the kernel even as experimental will allow a broad spectrum of testers to find any possible bugs. I find it nice that it tries to do more/different than other FS and is a great playground for new ideas. This can only benefit users. If it makes substantial win in certain categories, perhaps BTRFS and ZFS will look to add similar features and that in turn help evolve the whole FS ecosystem ! I really look forward to caching/tiering/encryption/snapshots and ERASURE CODING. Finaly a FS that offers that without needing to be distributed like CEPH !

I will test it up once it is in the kernel, as I don't compile kernels any more since I left Slackware more than 10 years ago.

[skeevy420]

I want you to think about what you've done, Kent.
And for now on, stop playing with yourself.
Y'all better have seen that movie and know that quote.

Seriously though, I like how it tries to do all the things I use ZFS for. If I wasn't so invested in ZFS I'd be a lot more excited about this. I hope he gets the Zstd stuff worked out since high-level Zstd compression is nice for write once read many datasets and directories like, well, most everything a desktop distribution uses by default aside from /home, /tmp, /etc, and /var. Those are what LZ4 and Zstd-fast are for.