So since noone has any ideas it's safe to assume that none of the many fixes btrfs has received within the last two and a half years have been backported to the aging sunxi 3.4 kernel. Therefore btrfs supports in this versions can be considered unstable/experimental and btrfs should be avoided.
bronco replied at 2015-11-20 12:33
So since noone has any ideas it's safe to assume that none of the many fixes btrfs has received with ...
You are right, no fixes are ported to 3.4.39 kernel.
But I run OPI-PLUS on btrfs for couple of months now, without any problem. I mostly use it because of compression an snapshots, and it works good.
You only have to be carefull when using the format (mkfs.btrfs) not to use unsupported options (always use mkfs.btrfs -O ^extref,^skinny-metadata ... ).
We had a very 'funny' data loss (36 TB in a few minutes) at a customer a few years ago with brtfs (excessive use of snapshots and send/receive) and then switched back to Solaris x86 and ZFS (mature fs code is always good!). Since the btrfs code lives inside the kernel I would only go with the most recent kernel versions and always take care that you've a backup. But since our data loss occured while doing exactly this (taking snapshots and sending them to a second host) you would've to rely on more inefficient approaches to do backup then.
Regarding btrfs it's kind of a dilemma: You both want to use the most recent kernel for stability fixes and also mature fs code at the same time. We always end up looking at ZFS when we think about it. But that's a different scenario compared to home useage of single board computers
Show IP Card
Radar Card
Published in 2015-11-18 04:42:32
Bump Card
Stick Card
Silent card
Open Thread Card
Highlighter card
Jack
Anonimous Post
Author
Published in 2015-11-20 21:55:59
