I am setting up a new test server. Doing a fresh install from CD onto a couple 4TB drives. Would like to try btrfs in a RAID 1 format. Are there any how to’s on how to do that?
I was under the impression that upstream was deprecating BTRFS.
brtfs was only a technology preview in RHEL 7 and will still remain in RHEL 7 (without support). it will be removed in the next major release.
best regrads Ulf
I don’t recommend it for roots. First there are udev, systems, and Btrfs limitations that prevent automatic degraded boot. Second, any bugs you find have a really good chance of already being fixed in upstream kernels. Btrfs has has hundreds of thousands of line changes from a CentOS 3.10 whatever kernel to a Fedora 4.14.8 kernel.
So if you were to test on Cent OS, at least you’d want to church any bugs against a current elrepo or Fedora kernel and try to reproduce the problem. And if it doesn’t reproduce, now what? Might as well stick with the newer kernel.
For data drives, raid 1 is fine. Just be sort aware of the fact Btrfs raid1
is not well named, is not going to behave like mdadm or LVM raid. There are known issues like no spares, no auto resync if a drive is temporarily missing and degraded writes happen to one drive, no faulty (kicked out)
drives when they misbehave, etc.
Anyway, Btrfs has been my primary filesystem for roots and data for years. And I’ve experienced no unplanned data loss. But I still keep many backups
(up to seven copies, five are independent, the of which are Btrfs based).
Upstream being Red Hat, yes. Upstream Btrfs development continues unaffected, Red Hat was not a major contributor to Btrfs development the last few years.