[Gluster-users] Quota translator: df now shows 0% used

Dan Bretherton dab at mail.nerc-essc.ac.uk
Tue Apr 13 15:27:10 UTC 2010

Dear All,
I decided to try the new volume quota feature after a non-responding
server for a distributed volume caused the other server's brick to
fill up completely.  After reading in a previous posting that
features/quota only works on the server at the moment, I added it to
my existing two server vol files.  I am not seeing any errors on
server startup or on mounting the distributed volume on clients, but
"df -hl" shows the volume as 0% full even though it is actually ~90%
full.  Read and write access appear to be working normally but I
haven't tried filling it up to find out if the quota translator
actually works.  One other observation is that new files created after
the quota translator was added appear to be showing up in the %used
column; when I checked just now it had gone up to 1% used.  I tried
running "ls -lR" on a couple of the clients (of which there are
several, incidentally) even though I'm not sure this self heal
operation has any effect on a simple distributed volume.  It didn't
have any effect on the %used according to df.  The volume file for one
of the two bricks is shown below.  I tried adding features/quota at
the end as well (just before the server section) but it didn't make
any difference.  Any comments or suggestions would be much appreciated

-Dan B (new list member).
# ==================
# Server volume file with quota translator added
volume posix1
  type storage/posix
  option directory /local/glusterfs

volume quota1
    type features/quota
    option disk-usage-limit 2500GB
    subvolumes posix1

volume locks1
    type features/locks
    subvolumes quota1

volume io-cache1
  type performance/io-cache
  subvolumes locks1

volume brick1
    type performance/io-threads
    subvolumes io-cache1

volume server
    type protocol/server
    option transport-type tcp
    option auth.addr.brick1.allow *
    option listen-port 6996
    subvolumes brick1

More information about the Gluster-users mailing list