[Gluster-users] Problem wit glusterfs-server 3.4

[Lalatendu Mohanty]

On 09/23/2013 04:12 AM, Marjan Grabowski wrote:
> Hi at all!
>
> i'm trying to use glusterfs for the first time and have the following problem:
> I want to have two nodes.
> On node1 I have a raid1-sytem running in /raid/storage
> Both nodes see the other and now I try to create a volume.
> While I create the first volume on a fresh system (node1) for the first time, gluster said:
> volume create: glustervolume: failed: /raid/storage/ or a prefix of it is already part of a volume
> How can it be? It is possible, that the mdadm-raid1-sytem is the reason?
>
>
> Regards,
> Tito

If you have used "/raid/storage/" in any volume create command (even if 
the actual command failed that time), you will get the error. This done 
to prevent users from using the same brick in multiple volumes which 
might cause data loss.

If you are sure that the partition is not used anywhere in gluster and 
you want to use it , you can explicitly remove the extended attributes 
associated with it. It would be a small hack.  I use a small 
script(below)  to do it for me.

#!/usr/bin/perl

if ($#ARGV < 1) {
     print("\n$0 <brick path> [brick path] ..\n");
     print("\nRun this script on each rhs node\n");
}
foreach my $file (@ARGV) {
     my $output = `getfattr -d -m . $file`;
     my @lines = split('\n', $output);
     foreach $line ( @lines ) {
             if ($line =~ /(.*)=.*==/){
                 my $attr = $1;
                 system("setfattr -x $attr $file");
         }
     }
}

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