[Gluster-users] Expand distributed replicated volume

Tue May 13 09:01:03 UTC 2014

Hi Lalatendu,

I've made the same test.

I've create a volume with 2 nodes sharing a 10G disk, without option "replica 2"

On the client, the share is 20Go.

The add a brick, the volume was 30Go.

This scenario works fine.

The problem occure when I want to have replication.

Running the same scenario with 6 nodes, and adding option "replica 2", finnaly my volume is 20go with 4 and adding 2 more nodes.

If I create the same volume with the 6 nodes from the begining it OK, but I can't expand it with nodes 7 & 8 to 40Go , still 30go.

For this test I run Fedora 20 with Gluster 3.5 from Fedora repository.

I do not have this problem with MooseFS, but this solution do not have NFS support.

Best regards

Hugues

-----Message initial-----
De:Lalatendu Mohanty <lmohanty at redhat.com>
Envoyé:lun. 05-05-2014 15:10
Sujet:Re: [Gluster-users] Expand distributed replicated volume
À:Hugues Lepesant <hugues at lepesant.com>; gluster-users at gluster.org; 

On 05/05/2014 03:08 PM, Hugues Lepesant wrote:

Hi all,

Does someone have already meet this ?

Best regards,

Hugues

 I tried the similar steps on two Fedora20 VMs (each with 3 100GB partition) + gluster 3.5.0, but did not hit this issue. Here are the steps I performed.

 1. Created a distribute replicate volume (2x2).
 2. mounted the volume using native glusterfs mount. "df -h" shows 200GB for the mount point.
 3. performed add brick with 2 partitions. The volume info shows 3x2)
 4.  df -h now shows 300GB for the mount point. 

 Thanks,
 Lala

 -----Message initial-----
 De: Hugues Lepesant <hugues at lepesant.com>
 Envoyé: mar. 29-04-2014 11:51
 Sujet: [Gluster-users] Expand distributed replicated volume
 À: gluster-users at gluster.org; 

Hi all,

I want to create a distributed replicated volume.
 And I want to be able to expand this volume.

 To test this, I have 6 nodes each have a 10G disk to share.

 First I create the initial volume.

 # gluster volume create testvol replica 2 transport tcp node-01:/export/sdb1/brick node-02:/export/sdb1/brick node-03:/export/sdb1/brick node-04:/export/sdb1/brick

 On the client, I mount the volume :

 # mount -t glusterfs node-01:/testvol /storage-pool
 # df -h /storage-pool/
 Filesystem        Size  Used Avail Use% Mounted on
 node-01:/testvol   20G  3.9G   17G  20% /storage-pool

 The volume size is 20G, what I expect.

 Now I want to expand this volume, adding brick.

 To do so, I do on node-01:
 # gluster volume add-brick testvol node-05:/export/sdb1/brick node-06:/export/sdb1/brick
 # gluster volume rebalance testvol start
 # gluster volume rebalance testvol status
 # gluster volume info testvol

 Volume Name: testvol
 Type: Distributed-Replicate
 Volume ID: cd24ec0f-3503-4d67-9032-1db1d0987f9c
 Status: Started
 Number of Bricks: 3 x 2 = 6
 Transport-type: tcp
 Bricks:
 Brick1: node-01:/export/sdb1/brick
 Brick2: node-02:/export/sdb1/brick
 Brick3: node-03:/export/sdb1/brick
 Brick4: node-04:/export/sdb1/brick
 Brick5: node-05:/export/sdb1/brick
 Brick6: node-06:/export/sdb1/brick

 The back on client :

 # df -h /storage-pool/
 Filesystem        Size  Used Avail Use% Mounted on
 node-01:/testvol   20G  2.9G   18G  15% /storage-pool

 Even If I umount/mount the volume, It's always 20G.
 Should it not display a size of 30GB ?
 Like if I create a distributed volume and the add-brick with replicate option

 # gluster volume create testvol transport tcp node-01:/export/sdb1/brick node-02:/export/sdb1/brick node-03:/export/sdb1/brick
 # gluster volume start testvol
 # gluster volume add-brick testvol replica 2 node-04:/export/sdb1/brick node-05:/export/sdb1/brick node-06:/export/sdb1/brick
 # gluster volume info testvol

 Volume Name: testvol
 Type: Distributed-Replicate
 Volume ID: 31159749-85fb-4006-8240-25b74a7eb537
 Status: Started
 Number of Bricks: 3 x 2 = 6
 Transport-type: tcp
 Bricks:
 Brick1: node-01:/export/sdb1/brick
 Brick2: node-04:/export/sdb1/brick
 Brick3: node-02:/export/sdb1/brick
 Brick4: node-05:/export/sdb1/brick
 Brick5: node-03:/export/sdb1/brick
 Brick6: node-06:/export/sdb1/brick

 On client : 
 # df -h /storage-pool/
 Filesystem        Size  Used Avail Use% Mounted on
 node-01:/testvol   30G   97M   30G   1% /storage-pool

 I'm using glusterfs-server 3.5.0 on Ubuntu 14.04 Trusty.

 Any help is welcome.
 Best regards,
 Hugues

_______________________________________________  Gluster-users mailing list  Gluster-users at gluster.org  http://supercolony.gluster.org/mailman/listinfo/gluster-users

_______________________________________________ Gluster-users mailing list Gluster-users at gluster.org http://supercolony.gluster.org/mailman/listinfo/gluster-users

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://supercolony.gluster.org/pipermail/gluster-users/attachments/20140513/4eadbe44/attachment.html>