EXT4 In Linux 4.15 Gets Online Resizing When Using Bigalloc, Corruption Fixes

[Zan Lynx]

The day Btrfs is able to surpass the rest Linux available filesystems in all ways and gets really robust even at RAID stuff, then it will become more interesting.

Ever had to reindex an entire Postgresql DB because ext3 and MD corrupted the index files? I have. Really robust Linux filesystems? Is to laugh.

And did you ever experience the joy of 4K kernel pages with stacked MD/LVM/XFS devices? For a while (maybe kernels 2.6.9, around there) we were lucky if it ONLY crashed once a week.

[xorbe]

Raise your hand if you're actually running ext4 + DAX. The number of btrfs problems through the opensuse forums is definitely non-zero. Why are btrfs advocates so touchy, and always lack any technical data supporting why btrfs is so desirable?

[timofonic]

Raise your hand if you're actually running ext4 + DAX. The number of btrfs problems through the opensuse forums is definitely non-zero. Why are btrfs advocates so touchy, and always lack any technical data supporting why btrfs is so desirable?

What are they doing wrong? After all those many years of invest, I wonder what's wrong with Btrfs. Are there any hope for it? Why?

[starshipeleven]

I wonder what's wrong with Btrfs.

It's a next-gen filesystem without compromises. So far the only other nextgen filesystem that works is ZFS, and it did make many compromises to get it in production.

Btrfs isn't under the same pressure to get a product out no matter the cost. Mostly because people needing a nextgen filesystem NOW and can pay its price are using ZFS.

Are there any hope for it? Why?

It's still being actively developed by SUSE, Fujitsu and Omnomnomnomnomracle, Facebook, plus other third parties.

And if you get past the learning curve (and don't use unstable features) it's usable without much issues.