Boston Linux & Unix (BLU) Home | Calendar | Mail Lists | List Archives | Desktop SIG | Hardware Hacking SIG
Wiki | Flickr | PicasaWeb | Video | Maps & Directions | Installfests | Keysignings
Linux Cafe | Meeting Notes | Blog | Linux Links | Bling | About BLU

BLU Discuss list archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Discuss] On Btrfs raid and odd-count disks

Its funny, but I never considered an odd number of drives as viable for
raid1. I guess its simple enough Dn + D(n+1 >= N ? 0 : n+1), but doesn't
"feel" right.

> Today I ran into a problem that I hadn't expected but I should have
> expected.
> Remember that Btrfs "raid" replicates file data and metadata, not disk
> blocks. If you have three disks in a raid1 configuration then any file
> written to one disk will have a replica written to another disk. If you
> have 3 times 500G disks then you have ~700G usable capacity. df reports
> this as 1.2T since it doesn't fully understand Btrfs.
> Mostly.
> Say you have 500G disks in a 3-disk raid set, and you've stored 150G of
> data. df will show 300G used and 1.1T free. That's 550G usable after
> dividing by half for mirroring. The largest file that you can write is
> still only 400G. This assumes even balancing of that 300G across all of
> the disks in the set. If that 300G is a single 150G file which is
> replicated across two disks in the set then the largest file that can be
> written is 350G -- the space available on those two disks.
> And if you do fill up one of the disks, such as by using dd like I did,
> then you will start getting file system full errors despite df showing
> plenty of usable space.
> --
> Rich P.
> _______________________________________________
> Discuss mailing list
> Discuss at

BLU is a member of BostonUserGroups
BLU is a member of BostonUserGroups
We also thank MIT for the use of their facilities.

Valid HTML 4.01! Valid CSS!

Boston Linux & Unix /