I have a 4 node cluster in test production and this is quite the problem.

Linux Fedora 10/11 client Fuse 2.74 Gluster 2.0.3
Gentoo 2.6.27-gentoo-r8 server Gluster 2.0.3

When mounted native the filesystem does not complete the writing of
Point Cloud files. When mounted via CIFS (glusterfs exported via
Samba) it writes the files.

I started an strace of the process but it crashed. Too many steps
before it actually gets to the Point Cloud generation. Renderman
Studio runs via Maya so that is VERY slow on the strace as well.

Any ideas of how to solve this problem?

Everything thing else is working. Had a problem with Houdini Mantra in
Gluster 2.0.1 that was fixed with an upgrade to 2.0.3. I was hoping
2.0.4 would fix this problem but no dice.

What other Info would you like?

Client Config

volume remote01
  type protocol/client
  option transport-type tcp
  option remote-host slave14
  option remote-subvolume brick01
end-volume

volume remote02
  type protocol/client
  option transport-type tcp
  option remote-host slave15
  option remote-subvolume brick02
end-volume

volume remote03
  type protocol/client
  option transport-type tcp
  option remote-host slave16
  option remote-subvolume brick03
end-volume

volume remote04
  type protocol/client
  option transport-type tcp
  option remote-host slave20
  option remote-subvolume brick04
end-volume

volume distribute
  type cluster/distribute
  subvolumes remote01 remote02 remote03 remote04
end-volume

volume writebehind
  type performance/write-behind
  option aggregate-size 128KB
  option window-size 1MB
  subvolumes distribute
end-volume

volume cache
  type performance/io-cache
  option cache-size 128MB
  subvolumes writebehind
end-volume

Server Config

volume posix
  type storage/posix
  option directory /node
end-volume

volume locks
  type features/locks
  subvolumes posix
end-volume

volume brick01
  type performance/io-threads
  option thread-count 8
  subvolumes locks
end-volume

volume server
  type protocol/server
  option transport-type tcp
  option auth.addr.brick01.allow 127.0.0.1,192.168.1*
  subvolumes brick01
end-volume