Talk Of "EXT5" File-System; Should EXT4 Be Frozen?

[phoronix]

Phoronix: Talk Of "EXT5" File-System; Should EXT4 Be Frozen?

In the discussion that followed when it was found a nasty EXT4 file-system corruption bug hit recent Linux kernel stable releases, one user proposed that EXT4 be put in a feature-freeze mode and future work then be put towards an "EXT5" file-system, to which Ted Ts'o did respond...

http://www.phoronix.com/vr.php?view=MTIxNTE

[r1348]

Did people give up on btrfs already? Apparently Oracle is just unreliable when it comes to open source.

[Teho]

Did people give up on btrfs already? Apparently Oracle is just unreliable when it comes to open source.

They did not. BTRFS is actively developed by many different companies (Oracle, Red Hat, Fujitsu, Intel, SUSE, STRATO, Fusion-io among others). The author and the lead developer of BTRFS left Oracle for Fusion-io but he still works on it.

[89c51]

Did people give up on btrfs already? Apparently Oracle is just unreliable when it comes to open source.

The comment was about adding new features. And btrfs is "not ready yet" -at least thats how its perceived in a way. No big distro uses it by default.

[ryao]

The comment was about adding new features. And btrfs is "not ready yet" -at least thats how its perceived in a way. No big distro uses it by default.

I have asked about this in #btrfs on freenode. The btrfs developers claim that anyone currently using btrfs should expect to lose data.

[emptythevoid]

After reading Ted's comment about ext4 vs ext5 and the rest... I just gotta say that he seems like a super nice guy, completely understanding, and real cool for taking the time to explain why the idea of forking to ext5 isn't the best idea (instead of just saying, "that's a bad idea, idiot" or something like that). That was real neat.

[uid313]

Any headline which ends in a question mark can be answered by the word no

[kernelOfTruth]

After reading Ted's comment about ext4 vs ext5 and the rest... I just gotta say that he seems like a super nice guy, completely understanding, and real cool for taking the time to explain why the idea of forking to ext5 isn't the best idea (instead of just saying, "that's a bad idea, idiot" or something like that). That was real neat.

quoted for truth


there's always some experimental, new and shiny features that I try out from time to time but this clearly shows that if you depend on your data - you should refrain from using too new stuff (even on stable kernels)

will probably go on with ext4 without metadata checksums

[coder111]

What a guy. I wish some of my coworkers were developers of his caliber.

And I agree with him completely- given how FEW data corruption bugs made it into ext4, and even given the fact that this one got caught before it made into any distributions, I just don't think it's worth the major pain and resource expenditure to fork ext4, at least not at this moment.

[89c51]

I just gotta say that he seems like a super nice guy, completely understanding, and real cool for taking the time to explain why the idea of forking to ext5 isn't the best idea (instead of just saying, "that's a bad idea, idiot" or something like that). That was real neat.

Not all kernel devs are expected to have the same personality. And thats good. Ie. Torvalds acts as a dick some times but he is very good at it and his dickery gets things done in the right way. Which is a benefit for everyone using linux.