[Gluster-users] File has different size on different bricks

[Nux!]

Hi chaps,

The setup:
- 2 bricks (2 different servers), replica 2
- server A acts as a brick and also as a client with the volume mounted 
in /mnt/glusterfs1
- server B is just a brick

The test:
On server A:
- I created a 100GB file in /mnt/glusterfs/rawfile.img using truncate 
(so it's sparse)
- I formatted the raw file ext4 and loop-mounted it in 
/mnt/loopmounttest/ then started to download a 10GB file using wget
On server B:
- when the 10GB transfer was midway I restarted server B using `reboot 
-f`

The problem:
While server B was rebooting the wget transfer in server A froze for a 
few seconds then resumed. When server B came back online it started to 
update /brick/rawfile.img from server A, but it's now hours since 
everything has finished and on server A rawfile.img has 28GB, but on 
server B it has 26GB; so a 2GB difference.

`gluster volume heal glusterfs1 info split-brain` doesn't indicate any 
split brains.
I even tried dd-ing /mnt/glusterfs/rawfile.img to /dev/null in the hope 
it will force an update on server B, but no changes.

Suggestions?

-- 
Sent from the Delta quadrant using Borg technology!

Nux!
www.nux.ro