Btrfs Gets Talked Up, Googler Encourages You To Try Btrfs
Phoronix: Btrfs Gets Talked Up, Googler Encourages You To Try Btrfs
This week at LinuxCon North America in Chicago is a presentation by Google's Marc Merlin that's entitled "Why you should consider using btrfs, real COW snapshots and file level incremental server OS upgrades like Google does." The presentation does a good job at looking at the state of Btrfs on Linux and comparing it to ZFS...
Been running it for ages.
</smug Arch user response>
Been running it longer
Originally Posted by kaprikawn
</smugger Fedora user response>
Honestly, I cringe at reading something like "it typically doesn't just corrupt itself or crash in recent kernels (>3.10)".
On my experiences with btrfs some years ago I always hit bugs, I have to confess that I'm still weary about btrfs from those times. The talk with btrfs always seems to be "it's stable, don't worry" followed by "...usually... but use a recent kernel just in case...."
To those of you proudly using if for some time, what are your experiences? Is the low-disk-space handling decently fixed yet? How long since you've last hit a btrfs bug (even if corrected)?
/dev/sda2 on / type btrfs (rw,noatime,ssd,space_cache)
Originally Posted by Ericg
Am i doin' it rite? </smuggest Gentoo user response>
I've been using it for a really long time (around kernel 3.7 or so) and where I live the power goes out rather often, the only problem I've ever an into (few kernels ago) was that chrome's preferences gets corrupted every few power offs.
Originally Posted by [Knuckles]
But I can't compare it to anything else since my root have been btrfs for years.
I used it for some time on our home server. It runs great until it doesn't. Even with recent kernels (3.14), I ran into trouble in combination with NFS. Sometimes it decides that it is now read-only until I do a reboot, it constantly had trouble with free space that was not really free. I had to run regular balances in order for it to not lock up eventually requiring a reboot - balance - reboot. I had only about 30GB free on the disk/ssd, and this seemed to be not enough for a stable experience. On the larger disk, with 500GB free, it ran without trouble.
I now converted everything back to LVM/ext4 and it just works.
I still use btrfs on my laptop and there it runs fine, but it also has a lot more free space, which is what I would strongly recommend.
I haven't had any btrfs corruption that i could pinpoint to be the filesystem's fault. I've got one drive (unfortunately a home server drive) that is throwing a few btrfs errors but I'm thinking thats because the drive is like... 8yrs old and probably dying of its own accord.
Originally Posted by [Knuckles]
Any advice to "use a recent kernel" is a smart move for ANY filesystem that is still being maintained, even EXT4, since you never know what bugs will be fixed. The low-disk-space bug I can't comment on as I've yet to fill up any btrfs drive completely or near-completely. Any other BTRFS bugs... If I've hit them, I didn't know it at the time.
Quite honestly BTRFS has been NOTHING but a pleasure to use as far back as I can remember.
I still suffer from the ReiserFS nightmare, because I jumped onto ReiserFS while it was new and the world was still a happy place, until the day a sad truth made the news and people started making cruel jokes about it (calling it MurderFS). I am completely burned when it comes to adopting a new file system and I will probably stick forever with Ext2/3/4.
I'm using it on my home server, and so far had no (visible) problems with the FS itself. The home server used to be a Raspberry Pi, but the Raspi couldn't really handle 2 external HDDs connected through USB for my btrfs raid1, it would lose the connection to one of the disks every other day or so. Even though I was using a powered hub and one of the drives is pretty old with its own external power supply. This caused a few scary messages from BTRFS in the journal so I eventually bought some better hardware that can actually cope with light USB load like in this case. I'm still using the same two external disk drives without reformatting, I just "scrubbed" the FS and so far everything seems nice, no more journal errors and it seems like no files were lost through the various USB troubles that repeatedly disrupted the kernel<->disk connection.
The RasPi used the 3.12 kernel IIRC (Arch Linux ARM), now I just use whatever most recent stable kernel is packaged for Arch, right now 3.16.
I still feel a little uncomfortable with BTRFS though, I do hope it won't start eating/corrupting files silently and without me noticing. Ironically, that's exactly the reason why I have decided to use BTRFS, because it should notice hidden file corruption through the use of checksums. Everything stored in my BTRFS could be restored from other places in case of emergency, but when you don't notice there's something wrong you may be unable to restore these files before the redundant copy on a different FS is inaccessible (disk failure etc) and you want to restore it from the BTRFS…