[Gluster-users] replica disk usage differs.

[Gerald Brandt]

Hi,

I have a replica 2 setup that I use via Gluster NFS (3.3).  Both bricks show all files in sync (gluster volume heal <VOLNAME> info).  I forced a sync the new way (gluster volume heal <VOLNAME> all) and the old way (find /root/tmp/ -noleaf -print0 | xargs --null stat >/dev/null [gluster volume mounted at /root/tmp]).

Both volumes show a different disk used value.  Both are similar sized disks, with XFS under Ubuntu 12.04.

/dev/md0       5858437120 995558664 4862878456  17% /export/gluster/RAID-6     (NFS mount point)
/dev/md0       5858437120 907253448 4951183672  16% /export/gluster/RAID-6     (replica)


Yet, when I look at the gluster mounts, the disk usage is the same:


192.168.10.2:/NFS_RAID6_FO 5858437120 995558784 4862878336  17% /root/tmp      (NFS mount point)
192.168.10.1:/NFS_RAID6_FO 5858437120 995558784 4862878336  17% /root/tmp1     (replica)


Can someone explain the reason for the inconsistency to me?

Thanks,
Gerald