[Gluster-users] how to recover a accidentally delete brick directory?

[shwetha]

Since the "trusted.glusterfs.volume-id" is not set on the bricks , the 
volume start will fail.

1) Execute : "getfattr -e hex -n trusted.glusterfs.volume-id 
/opt/gluster_data/eccp_glance"  from any of the node which the brick 
process running.

You will get the hex value of the "trusted.glusterfs.volume-id" extended 
attribute . Set this value on the bricks created (On the nodes where you 
deleted the bricks)

2) setfattr -n trusted.glusterfs.volume-id -v 
<hex_value_of_the_trusted_glusterfs_volume-id> /opt/gluster_data/eccp_glance

3) gluster volume start <volume_name> force  : To restart the brick process

4) gluster volume status <volume_name> : Check all the brick process are 
started.

5) gluster volume heal <volume_name> full : To trigger self-heal on to 
the removed bricks.

Shwetha
On 11/28/2013 02:59 PM, weiyuanke wrote:
> hi Shwetha,
>
>
>  command "gluster volume start eccp_glance force" on the other node 
> gives following:
> Pasted Graphic.tiff
>
> with cli.log
> Pasted Graphic 1.tiff
>
> on the damaged node，gluster volume start eccp_glance force gives:
> Pasted Graphic 2.tiff
>
>
>
>
> ---------------------------
> 韦远科
> 010 5881 3749
> 中国科学院 计算机网络信息中心
>
>
>
>
>
> On 2013年11月28日, at 下午4:50, shwetha <spandura at redhat.com 
> <mailto:spandura at redhat.com>> wrote:
>
>> 1) create the brick directory "/opt/gluster_data/eccp_glance" on the 
>> nodes where you deleted the directories.
>>
>> 2) From any of the storage node execute :
>>
>>  1. gluster volume start <volume_name> force  : To restart the brick
>>     process
>>  2. gluster volume status <volume_name> : Check all the brick process
>>     are started.
>>  3. gluster volume heal <volume_name> full : To trigger self-heal on
>>     to the removed bricks.
>>
>> -Shwetha
>>
>> On 11/28/2013 02:09 PM, 韦远科 wrote:
>>> hi all,
>>>
>>> I accidentally removed the brick directory of a volume on one node, 
>>> the replica for this volume is 2.
>>>
>>> now the situation is , there is no corresponding glusterfsd process  
>>> on this node, and 'glusterfs volume status' shows that the brick is 
>>> offline, like this:
>>> Brick 192.168.64.11:/opt/gluster_data/eccp_glance  N/A    Y    2513
>>> Brick 192.168.64.12:/opt/gluster_data/eccp_glance  49161    Y    2542
>>> Brick 192.168.64.17:/opt/gluster_data/eccp_glance  49164    Y    2537
>>> Brick 192.168.64.18:/opt/gluster_data/eccp_glance  49154    Y    4978
>>> Brick 192.168.64.29:/opt/gluster_data/eccp_glance  N/A    N    N/A
>>> Brick 192.168.64.30:/opt/gluster_data/eccp_glance  49154    Y    4072
>>> Brick 192.168.64.25:/opt/gluster_data/eccp_glance  49155    Y    11975
>>> Brick 192.168.64.26:/opt/gluster_data/eccp_glance  49155    Y    17947
>>> Brick 192.168.64.13:/opt/gluster_data/eccp_glance  49154    Y    26045
>>> Brick 192.168.64.14:/opt/gluster_data/eccp_glance  49154    Y    22143
>>>
>>>
>>> so are there ways to bring this brick back to normal?
>>>
>>> thanks!
>>>
>>>
>>> -----------------------------------------------------------------
>>> 韦远科
>>> 中国科学院 计算机网络信息中心
>>>
>>>
>>>
>>> _______________________________________________
>>> 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/20131128/b54406e9/attachment.html>