3

I had a ~1.6TB BTRFS volume on a single partition, to which successfully I added another freshly created and formatted partition (on another device) of exact same size with the intent of mirroring them. The size of the volume now appears to be ~3.2TB. However, when I try to convert the volume (mounted to /mnt/mnt0) to a RAID-1 with

btrfs balance start -dconvert=raid1 -mconvert=raid1 /mnt/mnt0

(from here)

I get the following error:

ERROR: error during balancing '/mnt/mnt0' - Inappropriate ioctl for device
There may be more info in syslog - try dmesg | tail

dmesg output is pretty uninteresting (and none of this appears to have been produced by running the above command in any case):

root@Xeon:~# dmesg | tail
[26712.029432] device label Storage devid 1 transid 3760 /dev/sdb1
[26712.029464] device fsid 7c7bb4a7-f403-4dad-b7e3-92fa1a30c764 devid 1 transid 175588 /dev/sdb2
[26712.029494] device fsid 56a9e829-c964-4f16-a37f-3bde05fb22a0 devid 1 transid 172361 /dev/sdb3
[26712.029524] device fsid 29181083-c9d6-4d8a-bd3b-1f17b925bb21 devid 1 transid 70197 /dev/sdb4
[26712.029582] device fsid ff65812c-9515-465d-8536-e8d00cfb2193 devid 1 transid 3677 /dev/sda1
[26712.029641] device label Storage devid 2 transid 3760 /dev/sdc1
[26712.029886] device fsid 44c395d9-df4e-4ab2-8a41-5b06e584e854 devid 1 transid 4 /dev/sdc2
[26828.213983] device label Storage devid 1 transid 3760 /dev/sdb1
[26967.116350] device label Storage devid 1 transid 3760 /dev/sdb1
[27079.571213] device label Storage devid 2 transid 3763 /dev/sdc1

btrfs fi show gives (in case this is of relevance):

Label: 'Storage'  uuid: 77a43f33-908b-468b-9570-55964547d1a1
        Total devices 2 FS bytes used 744.68GB 
        devid    2 size 1.62TB used 0.00 path /dev/sdc1
        devid    1 size 1.62TB used 947.54GB path /dev/sdb1

Both partitions passed btrfs.fsck.

How would I make it work?

EDIT: This is on an up-to-date Debian Wheezy installation with a 3.2 kernel.

UPDATE: Strangely enough, btrfs balance start /mnt/mnt0 worked fine - however, btrfs balance cancel /mnt/mnt0 failed with the same error. After balancing, both volumes appear to have the same data on them, but the data is still in RAID0; btrfs fi show gives

Label: 'Storage'  uuid: 77a43f33-908b-468b-9570-55964547d1a1
        Total devices 2 FS bytes used 744.66GB
        devid    2 size 1.62TB used 901.01GB path /dev/sdc1
        devid    1 size 1.62TB used 901.01GB path /dev/sdb1

And btrfs fi df:

Data, RAID0: total=1.72TB, used=742.51GB
System, RAID1: total=8.00MB, used=144.00KB
System: total=4.00MB, used=0.00
Metadata, RAID1: total=19.00GB, used=2.15GB
jaymmer - Reinstate Monica
  • 161
  • 1
  • 7

1 Answers1

3

Solved by upgrading kernel to 3.8.4. Had to compile my own, but stock settings seem to be working fine (I built btrfs statically rather than as a module, though I'm unsure if this was necessary).

EDIT: I can't believe I missed this... according to the BTRFS wiki, changing raid levels after the filesystem has been created isn't supported for 3.2 series and older kernels. That must have been my problem.

jaymmer - Reinstate Monica
  • 161
  • 1
  • 7