Gluster users - Aug 2013 - GlusterFS procedures in case of failures?

Hi, I'm deploying a Gluster Cluster, composed of 2 server nodes, 4 
bricks (for now), with a replica 2 distributed setup. So, server1 is 
identical to server2, brick1 on server1 replicates brick1 on server2, 
and brick2 on server1 replicates brick2 on server2. Everything is 
working perfectly.

My main concern is never losing any data and I need to know what to do 
in case of:

1. A whole server crashes.
2. One brick in a server crashes.

In case of 1 happening, lets say that server2 crashes, I should be able 
to get a 3rd server, put 2 empty bricks there and have it replicating 
server1. What is the command I should do to get this working?

In case of 2 happening, lets say that brick1 on server2 crashes, I 
should be able to simply add another brick on server2. What is the 
command I should do to get this working?

Thank you very much for any answers.

It would be helpful if there was a guide for these types of things.
There is lots of information on setup but little on day to day
maintenance.

For example, what do I do if:
1.  A node in the cluster must be rebooted and I am running
    a. distributed
    b. distributed / replicated
2.  What do I do if all the nodes need to be rebooted? Should I repeat
step 1 in serial for all the missing nodes?

For a distributed filesystem, it would be really nice if any
operations on files on a missing node would block until the node
re-appeared.  Then I could reboot a node and just slow my clients
down.

Thanks,

Joel



On Wed, Aug 21, 2013 at 8:04 AM, Daniel Ferreira Jorge
<daniel.jorge at sticorp.com.br> wrote:
> Hi, I'm deploying a Gluster Cluster, composed of 2 server nodes, 4
bricks
> (for now), with a replica 2 distributed setup. So, server1 is identical to
> server2, brick1 on server1 replicates brick1 on server2, and brick2 on
> server1 replicates brick2 on server2. Everything is working perfectly.
>
> My main concern is never losing any data and I need to know what to do in
> case of:
>
> 1. A whole server crashes.
> 2. One brick in a server crashes.
>
> In case of 1 happening, lets say that server2 crashes, I should be able to
> get a 3rd server, put 2 empty bricks there and have it replicating server1.
> What is the command I should do to get this working?
>
> In case of 2 happening, lets say that brick1 on server2 crashes, I should
be
> able to simply add another brick on server2. What is the command I should
do
> to get this working?
>
> Thank you very much for any answers.
> _______________________________________________
> Gluster-users mailing list
> Gluster-users at gluster.org
> http://supercolony.gluster.org/mailman/listinfo/gluster-users