Gluster users - Mar 2010 - how to export 4 disks by only one subvolume name?

Hi,

i try to create a server vol file which exports 4 local disks via 
distribute translator.
all examples i find use the distribute translator on the client side.
the thing is, i want to be sure that the client doesnt use the disks in 
a wrong way
like misconfiguring a primary disk as a replicate or such.

so my primary goal is to export 4 disks as just one brick and all files 
are distributed (not striped)
over those disks.
later i will add more disks, but i dont want the client vol file to be 
changed!

using the following vol files i get an error:
root at hostname:# ls /media/storage/Media
ls: cannot access /media/storage/Media: Input/output error

the vol's are started like this
server:
/usr/local/sbin/glusterfs --log-level=DEBUG --disable-direct-io-mode -l 
/media/74fd89bf-b8c7-4f3d-9092-826613fe1e85/glusterfsd/glusterfs.server.log 
-f /media/74fd89bf-b8c7-4f3d-9092-826613fe1e85/glusterfsd/server.vol 2>&1

client:
mount -t glusterfs 
/media/74fd89bf-b8c7-4f3d-9092-826613fe1e85/glusterfsd/client.vol -o 
log-file=/media/74fd89bf-b8c7-4f3d-9092-826613fe1e85/glusterfsd/glusterfs.client.log,volfile-name=Media,log-level=DEBUG
/media/storage/Media

the server doesnt log anything after the startup
and the client logs from the 'ls' are

[2010-03-14 17:04:04] D [client-protocol.c:4908:client_lookup_cbk] 
serverMedia: LOOKUP 0/ (/): inode number changed from 
{5445936622006697985,1} to {0,1}
[2010-03-14 17:04:04] W [fuse-bridge.c:722:fuse_attr_cbk] 
glusterfs-fuse: 2: LOOKUP() / => -1 (Success)

any idea how i can export the server data with only one sub-volumename, 
distributing files over the disks
and having the possibility to expand the storage by adding new disks, 
maybe including replication, in the next
months?

thx ll and
kind regards




#################server.vol#############################
volume posix_68d5f83c-bbae-40c7-9a3d-e2a52c3c0ecd
  type storage/posix
  option directory /media/68d5f83c-bbae-40c7-9a3d-e2a52c3c0ecd/storage/Media
end-volume
volume locks_68d5f83c-bbae-40c7-9a3d-e2a52c3c0ecd
  type features/locks
  subvolumes posix_68d5f83c-bbae-40c7-9a3d-e2a52c3c0ecd
end-volume


volume posix_74fd89bf-b8c7-4f3d-9092-826613fe1e85
  type storage/posix
  option directory /media/74fd89bf-b8c7-4f3d-9092-826613fe1e85/storage/Media
end-volume
volume locks_74fd89bf-b8c7-4f3d-9092-826613fe1e85
  type features/locks
  subvolumes posix_74fd89bf-b8c7-4f3d-9092-826613fe1e85
end-volume


volume posix_ed14c1d5-0560-46a5-a8cd-b67bbb77386a
  type storage/posix
  option directory /media/ed14c1d5-0560-46a5-a8cd-b67bbb77386a/storage/Media
end-volume
volume locks_ed14c1d5-0560-46a5-a8cd-b67bbb77386a
  type features/locks
  subvolumes posix_ed14c1d5-0560-46a5-a8cd-b67bbb77386a
end-volume


volume posix_f2db5144-ee3f-4047-b845-2e08c6bc4a2e
  type storage/posix
  option directory /media/f2db5144-ee3f-4047-b845-2e08c6bc4a2e/storage/Media
end-volume
volume locks_f2db5144-ee3f-4047-b845-2e08c6bc4a2e
  type features/locks
  subvolumes posix_f2db5144-ee3f-4047-b845-2e08c6bc4a2e
end-volume


volume distributeMedia
   type cluster/distribute
   subvolumes  locks_68d5f83c-bbae-40c7-9a3d-e2a52c3c0ecd 
locks_74fd89bf-b8c7-4f3d-9092-826613fe1e85 
locks_ed14c1d5-0560-46a5-a8cd-b67bbb77386a 
locks_f2db5144-ee3f-4047-b845-2e08c6bc4a2e
end-volume

volume Media
  type performance/io-threads
  option thread-count 16
  subvolumes distributeMedia
end-volume

volume storageServer
   type protocol/server
   option transport-type tcp
   option transport.socket.listen-port 6996
   option auth.addr.Media.allow *
   subvolumes Media
end-volume

#######################client.vol#############################
volume serverMedia
  type protocol/client
  option transport-type tcp
  option remote-host 127.0.0.1
  option remote-port 6996
  option remote-subvolume Media
end-volume

volume writebehindMedia
    type performance/write-behind
    option cache-size 4MB
    subvolumes serverMedia
end-volume

volume readaheadMedia
    type performance/read-ahead
    option page-count 4
    subvolumes writebehindMedia
end-volume

volume Media
  type performance/io-cache
  option cache-size 512MB
  subvolumes readaheadMedia
end-volume

Ian Rogers wrote:
>
> Hi Jeti,
>
> Could you use LVM or mdadm on the server to create one filesystem on 
> the four disks and then export that?
i dont have much expirience on LVM, but already used mdadm (a bit)
either way, one thing i like about gluster is that it doesnt introduce a 
new filesystem OR handling of the filesystem underneath gluster
its not that i dont want to learn something new, but when it comes to 
safety of my data
i appreciate tools that i already know and in case of some failure i 
understand whats
happening and the problems that can further arise from such a failure

as a worst case i think i will export the disks via unix domain sockets, 
mount this directory
and reexport again using a second server vol file
at least it should work
>
> This won't fulfil your requirement of having whole files on one disk, 
> but does fulfil the other two (1 brink, add more disks later). Are you 
> intending to remove disks to transport files or something?
>
well, would be an interesting option, but not necessary i think ^^
but as stated above standard tools are easier and fail safer to use for 
me (currently)
> Ian
thx & regards