[Gluster-users] Question

[Christopher Hawkins]

Bryan, 

Each server should be able to survive the failure after what's called the "ping-timeout" has expired (usually takes 30 - 45 seconds after a failure, in total). Can you post your configs?  Mine are like this, which are working in a replicate set up very similar to yours (or were, I just upgraded to 3.04 and have not re-tested failover yet).


Client vol file:
#### Add client feature and attach to remote subvolume
volume master1
  type protocol/client
  option transport-type tcp
  option remote-host 192.168.1.91
  option transport.socket.nodelay on
  option transport.remote-port 3399
  option remote-subvolume threads2
end-volume

volume master2
type protocol/client
  option transport-type tcp
  option remote-host 192.168.1.92
  option transport.socket.nodelay on
  option transport.remote-port 3399
  option remote-subvolume threads2
end-volume

### Add replication mirror
volume mirror1
    type cluster/replicate
    option read-subvolume master1
    subvolumes master1 master2
end-volume

Server:

volume db
  type storage/posix
  option directory /cluster/shadow/db
end-volume

volume locks2
type features/locks
    subvolumes db
end-volume

volume threads2
    type performance/io-threads
    option thread-count 8
    subvolumes locks2
end-volume

volume server-tcp
    type protocol/server
    option transport-type tcp
    option auth.addr.threads2.allow 192.168.1.*
    option transport.socket.listen-port 3399
    option transport.socket.nodelay on
    subvolumes threads2
end-volume


Chris

----- "Bryan McGuire" <bmcguire at newnet66.org> wrote:

> Hello,
> 
> I want to use the replication ability of  GulsterFS as a storage for 
> 
> our mail servers and have the following.
> 
> Two CentOS boxes - Gluster 3.0.4 installed on both. Each box is a  
> server and a client so we can load balance the email traffic between 
> 
> the two servers.
> 
> Everything works fine except when one server goes down. The other  
> seems to loose connectivity to the shared storage. I do an ls /mnt/ 
> mail and it just hangs and never responds.
> 
> Is this by design or am I missing something?
> 
> Bryan McGuire
> 
> 
> _______________________________________________
> Gluster-users mailing list
> Gluster-users at gluster.org
> http://gluster.org/cgi-bin/mailman/listinfo/gluster-users