[Gluster-users] Gluster 3.1.4 - Replica out-of-sync

[Jorge Pascual]

Hi Guys!

I'm trying this amazing tool and my conclusion is that It's the best open
source approach to elastic storage that I've seen.

However, I have a problem with the sync of the replica.

I've set-up a configuration with 2 servers / 2 bricks / one disk each one in
replica 2 and one native client.

When all servers are up and running, the system works like a charm. I
shutdown one machine (Brick2) and do 2 operations in the client side:

a) Create a new file
b) Modify an old file (very small... only 7 bytes)

The new file is created on the server side (Brick1) and the modified one
it's also correct on the server side (Brick1).
I start the Brick2 and trigger a self-heal. In the client I do that:
find /opt/shared -print0 | xargs --null stat >/dev/null

I look what has happened in the server side (Brick2) and see that:

a) The new file is created in Brick2 <-- that's right!
b) The old file continues with the old content. It hasn't synced <--- that's
bad!

So, I have the modified archive that is ok in client and Brick1 but it's out
of sync in Brick2.

Anyone knows what could it be happening?

The servers set-up is:

gluster> volume create myvol replica 2 transport tcp
192.168.1.136:/opt/storage1
192.168.1.134:/opt/storage1
gluster> volume set myvol auth.allow 192.168.1.*
gluster> volume start myvol
gluster> volume info myvol
Volume Name: myvol
Type: Replicate
Status: Started
Number of Bricks: 2
Transport-type: tcp
Bricks:
Brick1: 192.168.1.136:/opt/storage1
Brick2: 192.168.1.134:/opt/storage1
Options Reconfigured:
auth.allow: 192.168.1.*

The client set-up is:
mount -t glusterfs 192.168.1.136:/myvol /opt/shared

Regards!!

Jorge
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://supercolony.gluster.org/pipermail/gluster-users/attachments/20110419/356b0c91/attachment.html>