Is BTRFS still relevant?

Software & Operating Systems Linux
21

That would be weird since ZOL is still catching up and sad as you need to patch the kernel to have it fully working.

22

Not really.

The kernel license is ZFS hostile, so ZoL has to be implemented in an arse-about way rather than the way it was implemented within Solaris or FreeBSD - where it is treated as a first-class citizen within the kernel and the rest of the OS.

1 Like
24

Its not in the shipping kernel, the shipping kernel is not written around ZFS. On both Solaris and FreeBSD, ZFS is deeply hooked into the kernel, and kernel modifications are made with ZFS in mind.

Then why just last week was i hearing about some data-loss ZoL bug?

1 Like
26

I’m not going to bother to argue with you other than to leave this snippet of the Linux kernel-developer attitude towards ZFS

Linus Torvalds’ second in command, Greg KH, had to say the following about his views about ZFS On Linux with the current issue at hand:

My tolerance for ZFS is pretty non-existant. Sun explicitly did not want their code to work on Linux, so why would we do extra work to get their code to work properly?

ref: https://www.phoronix.com/scan.php?page=news_item&px=ZFS-On-Linux-5.0-Problem

Thus, my view is

  • if you’re running ZoL in production: Good luck! Given the kernel team’s attitude torwards it.
  • If you’re deciding on a platform to run ZFS for serious use, I’d suggest either Nexenta, FreeBSD or some other solaris variant. Where as i said above, ZFS is treated as a first class filesystem.
1 Like
28

Anybody who cares to be able to keep their kernel patched without random delays because the kernel team “have no tolerance for ZFS” an broke compatibility with it.

ZFS may not be “catching up” in terms of functionality - my point is that the integration and level of support vs. any of the other platforms where it is native is… sub-par.

I would not run ZoL in production on anything i seriously care about (i.e., will lose money if it breaks) yet. Not until it is treated by the kernel team as actually important.

But hey, the cavalier “i don’t like it so fuck that guy who wrote it, and fuck the users of it” attitude seems to be endemic within the Linux community.

2 Likes
30

Aslong as you need to patch the kernel to get all the functionality out of ZoL you have a problem.
And this attitude towards ZFS could just be the beginning of compability issues. And you know the kernel dev wont help you.

2 Likes
31

Isn’t the write hole only on raid 5/6?

ZFS is a great filesystem and is widely used everywhere but I also don’t see where all the hate for BTRFS is coming from. If you could go more in depth with your explanation that would be great.

32

Sure.

The point though is that BTRFS was supposed to be better than ZFS and (more?) open source originally.

The fact that it is broken with RAID5/6 kinda shoots holes in that.

I’m not sure why you’d bother to run BTRFS rather than just use regular mirrors and say ext4, given that snapshots also have issues and can cause corruption if the system goes down mid-snap…

I guess the point is “is BTRFS still relevant” given that there’s very little point to running it instead of a regular FS on a raid group, given the issues it has with data integrity.