[sheepdog-users] Multidisk & ECC with one sheep daemon

[Liu Yuan]

On Wed, May 20, 2015 at 08:25:22PM +0200, Marcin Mirosław wrote:
> W dniu 2015-05-20 o 17:51, Liu Yuan pisze:
> > On Wed, May 20, 2015 at 12:19:32PM +0200, Marcin Mirosław wrote:
> >> Hi!
> >> I'd like to use one sheep daemon with Multi disk and use ECC for
> > 
> > Yes, it is possible with undocumented disk mode, try following:
> > 
> > $ ./configure --enable-diskvnodes ...
> > 
> > then run a single sheep with multiple disks. and every disk will share one piece
> > of data, either parity or real data. This sheep will act as a RAID controller.
> 
> I tried with --enable-diskvnodes. But without success:
> # sudo -u sheepdog sheep -n
> /mnt/test/sheep0,/mnt/test/sheep1,/mnt/test/sheep2,/mnt/test/sheep3,/mnt/test/sheep4
> # dog cluster format -c 2:1
> Number of copies (3) is larger than number of nodes (1).
> Are you sure you want to continue? [yes/no]: yes
> using backend plain store
> # dog node md info
> Id      Size    Used    Avail   Use%    Path
>  0      850 GB  0.0 MB  850 GB    0%    /mnt/test/sheep1
>  1      850 GB  0.0 MB  850 GB    0%    /mnt/test/sheep2
>  2      850 GB  0.0 MB  850 GB    0%    /mnt/test/sheep3
>  3      850 GB  0.0 MB  850 GB    0%    /mnt/test/sheep4
> # dog vdi create vditest 10G
> # sheepfs /mnt/sheep/
> # echo vditest >/mnt/sheep/vdi/mount
> # echo fooo >/mnt/sheep/volume/vditest
> -bash: echo: write error: Input/output error
> 
> and in sheep.log is:
> May 20 20:18:36  ERROR [gway 10920] gateway_forward_request(507) There
> isn't enough copies(1) to send out (3)
> 
> Did I miss something obvious?
> Marcin

Ah, it needs some modification to the code for the whole idea to work. Well,
for simplicity, just run multiple sheep.

Thanks,
Yuan