• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Anyone using BTRFS, how is it?

R

Red Squirrel

[H]F Junkie
Joined
Nov 29, 2009
Messages
9,217
I've been using md raid for a long time, but I recently found out it's limitations with raid 10, it can't do live expansion. BTRFS can according to the documentation. It's also better as it has more features, error checking etc...

I'm just wondering who here is using it for production and if it's stable or if I should wait? I'd be using it for all sorts of things from just misc files, to VMs. Mostly via NFS shares.
 
I'm just wondering who here is using it for production and if it's stable or if I should wait?
Click to expand...

I have used it in production for several years at home and work. However I do not use the raid5/6 features with btrfs. For that I use btrfs on top of mdadm raid6. However since early this year I have moved a few servers to using ZoL so I am migrating away from btrfs. With ZoL I fully trust that the raidz(123) levels are stable and I will be fine when hardware fails..

I've been using md raid for a long time, but I recently found out it's limitations with raid 10, it can't do live expansion.
Click to expand...

Hmm. I did live expansion many times with mdadm raid5 or raid6 but I don't think I ever tried that with mdraid10.
 
Last edited:
Yep raid 5/6 works fine, I figured 10 would too, but went to expand a volume and said it can't do it. I also looked to see if raid 0 can be expanded (I could just do a raid 10 manually) and that does not work either. Kinda a big bummer as I am moving more towards raid 10 instead of 5 given the availability of larger drives.

I'm guessing ZOL = ZFS On Linux? Is ZFS actually better than BTRFS? From what I'm reading it sounds like BTRFS is even better, though I guess ZFS is more mature too.
 
I'm guessing ZOL = ZFS On Linux?
Click to expand...

Yes.

Is ZFS actually better than BTRFS?
Click to expand...

In my opinion ZoL is not as fast as btrfs on top of mdadm raid6 but it is more stable and also I believe fixes and development is faster on ZoL versus btrfs. Although the only real problem I remember on btrfs is extremely long mounts (several minutes) when the space_cache needed to be regenerated.
 
Last edited:
btrfs development seems to be in permanent beta for a few years now. It mostly works, but for dependability and reliability it fails in comparison to XFS or ext4. And the development is agonizingly slow. Chris Mason is just not a good lead developer for a project like this.
 
I've got an experimental btrfs system at work. Raid on btrfs is pretty far from something I'd want to use, but btrfs on an md raid is working pretty well. Compression works well, performance is sub xfs, scrubs work fine, I wish sending and receiving was a little simpler, but overall it's decent. I'd hold off on "production" use, but it's defiantly usable if you wanna jump in.
 
Good to know, my goal was to use it's raid features as well but guess I'll wait a bit longer.

Going to stick to md raid for now and just deal with the fact that I can't expand raid 10's. I'll just have to plan my raid 10's better so I don't need to expand them. I do have one array I did with intention of adding more drives later but guess I'll have to figure something else for that.
 
I used it for quite some time on top of dmcrypt on top of mdraid. I have since moved to a ZFS raidz3 pool. That was around kernel 3.2 to ~3.8.

The reasons are that there were some performance issues with dmcrypt and most of the linux filesystems at that time, that for some odd reasons did not apply to ZFS at all (probably because it has an own file cache). The mentioned extremely long pool import was also part of the transition.

I'm still using it for the OS partition of my file server, because this runs from SSDs and the aforementioned problems are not as important for a very fast backing device.
I miss some of the btrfs features like file level deduplication and on demand defragmentation, but the heavy write caching that ZFS uses (burst write all data every 5 seconds, read in the remaining time) makes it much more enjoyable to work with.

The ZFS on Linux project is aiming for a production implementation that priorizes stability over everything else, while btrfs still implements new features in basically every kernel release.
Just recently they changed the on disk format again.
 
Last edited:
I use BTRFS on my server's OS disk and backup disk (on kernel 3.14 with btrfs-tools 3.14.1) so I can easily perform incremental backups of the live OS via snapshots.

RAID5/6 on BTRFS is simply not ready to be used outside of testing yet but RAID 1 should be fine.

See: http://marc.merlins.org/perso/btrfs/post_2014-03-23_Btrfs-Raid5-Status.html
https://btrfs.wiki.kernel.org/index.php/RAID56

See the many limitations and the various important missing pieces.

This is as of kernel 3.14 so it's pretty recent.

I stick to ZFS on Linux for my currently 12-disk 2-vdev RAIDZ2 and it's been working perfectly for 8 months now. I've had some bad disks, bad power and SATA connections and other things and ZoL has always detected the errors and fixed them automatically and perfectly for me.
 
Last edited:
You must log in or register to reply here.
Back
Top