Hello guys, i'm back =)
This suggest works but, no fully.
Actual Scenario:
2 bricks (node1 and node2) - It contains all data (about 2.0Tb).
I added more 2 bricks (node3 and node4) with replica 4.
In a client server (client1) I already want the "/data" mounted, and executed the command "find . | xargs stat".
So, the nodes 1 and 2 starting to replicate data to new bircks, but after a week the replication stop.
Now i have 2.0Tb in node1 and node2, and 1.3Tb in node3 and node4.
How can i sync the ramainder data to this nodes ?
2013/11/26 shwetha <spandura@xxxxxxxxxx>
1.
Option: cluster.read-subvolume
Default Value: (null)
Description: inode-read fops happen only on one of the bricks in replicate. Afr will prefer the one specified using this option if it is not stale. Option value must be one of the xlator names of the children. Ex: <volname>-client-0 till <volname>-client-<number-of-bricks - 1>
For any volume option information execute "gluster volume set help"
2. " find . " will get fetch all the directories entries of the mount. When we execute stat on each of file and directory it gets the stat structure of the file.
In your case we set the "cluster.read-subvolume" to node-1 is because node-1 has all the data and the read has to happen from node-1 so that 1) GFID is assigned to the file/directory ( since your files/directories were created from backend and doesn't have any GFID's yet ) 2) when stat is performed on the file/directory the file/directory is self-healed to node-2.
When you use "cluster.read-subvolume" all reads will be going to node-1 . Hence suggested to you reset this option as soon as the self-heal of data is complete.
-Shwetha
On 11/25/2013 05:40 PM, Raphael Rabelo wrote:
Hi guys!
shwetha,
Just to understand your sugestion:
1. What meaning "cluster.read-subvolume"? I searching something about it, but didn't find anything...2. Why i need to exec "xarg stat" to force the gluster "read" the file?
My volume has about 1.2Tb of used space, i can't stop the read 'cause my application write/read in a real time...
Thanks!
2013/11/22 Raphael Rabelo <rabeloo@xxxxxxxxx>
Thank you!
I'll try this!
Best regards.
2013/11/22 shwetha <spandura@xxxxxxxxxx>
Hi Raphael,
Following are the steps to be followed to sync data from node1 to node2 assuming node1 has the data.
1. gluster peer probe node1
2. gluster volume create gv0 replica 2 node1:/data node2:/data
3. gluster volume start gv0
4. gluster volume set gv0 cluster.read-subvolume gv0-client-0
5. Create a fuse mount from the client node to the volume : mount -t glusterfs node1:/gv0 /mnt/gv0
6. From client node : cd /mnt/gv0 ; find . | xargs stat
This should self-heal all your data from node1 to node2.
Once the self-heal is complete reset the "cluster.read-subvolume" volume option.
7. gluster volume reset gv0 cluster.read-subvolume
8. Unmount the mount point and remount for using it again.
Regards
-Shwetha
On 11/21/2013 08:51 PM, Raphael Rabelo wrote:
I guys!
i have 2 servers in replicate mode, the node 1 has all data, and the cluster 2 is empty.
I created a volume (gv0) and start it.
Now, how can I synchronize all files on the node 1 by the node 2 ?
Steps that I followed:
gluster peer probe node1
gluster volume create gv0 replica 2 node1:/data node2:/data
gluster volume gvo start
thanks!
_______________________________________________ Gluster-users mailing list Gluster-users@xxxxxxxxxxx http://supercolony.gluster.org/mailman/listinfo/gluster-users
_______________________________________________ Gluster-users mailing list Gluster-users@xxxxxxxxxxx http://supercolony.gluster.org/mailman/listinfo/gluster-users
