Gluster users - Dec 2010 - simple 2 server replication question - how to mount for HA

We are setting up a redundant storage cluster with 2 servers with Gluster FS 3.1

 

 

Per the documentation here:

http://www.gluster.com/community/documentation/index.php/Gluster_3.1:_Configuring_Distributed_Replicated_Volumes

 

We would create a volume using this:

 

#gluster volume create test-volume replica 2 transport tcp server1:/exp1
server2:/exp2

 

 

Now my question is how do we mount the clients in a HA fashion so that if one of
the servers goes down things stay online?

 

If we use this mount option:

 

 

#mount -t glusterfs [-o <options>] <volumeserver>:<volumeid>
<mount?point>

 

 

with a single server then there is an issue if that server goes down,

 

 

 

I understand we could point to a volume file instead,  could that volume file
contain references for both servers?

 

 

#mount -t glusterfs [-o <options>] <path/to/volumefile>
<mountpoint>

 

 

 

Should/is the replication done on the server side or the client side?  It seems
it is being replicated by the server with this ?replica 2? volume, so the
question is how do we do the mount?

 

Is it possible that the volume information is synched between the 2 servers so
that we can point the clients in our environment to either Gluster servers?

 

 

Many thanks,

Ben,

If you're using Gluster 3.1.1 then you don't need to do anything with
vol
files including mounting using them.  When you mount using -t glusterfs as
an option, you can just point to the hostname:/VOLFILE and Gluster will
automatically retrieve vol file information and open parallel connections
to the Gluster storage nodes.  This also means you don't need to mount
different hostnames to different clients since the hostname you use is
only used for the initial configuration retrieval.  Also, when using -t
glusterfs to mount, your connection will already be highly available since
the Gluster native protocol has failover built in to it.  If you are
testing failover, just be aware that there is a ping-timeout setting
described in the documentation that handles how long Gluster will wait for
a server if it is unable to contact a server.

-Jacob

-----Original Message-----
From: gluster-users-bounces at gluster.org
[mailto:gluster-users-bounces at gluster.org] On Behalf Of Ben Blakley
Sent: Wednesday, December 01, 2010 10:49 PM
To: gluster-users at gluster.org
Subject: Re: [Gluster-users] simple 2 server replication question - how to
mount for HA

We are setting up a redundant storage cluster with 2 servers with Gluster
FS 3.1





Per the documentation here:

http://www.gluster.com/community/documentation/index.php/Gluster_3.1:_Conf
iguring_Distributed_Replicated_Volumes



We would create a volume using this:



#gluster volume create test-volume replica 2 transport tcp server1:/exp1
server2:/exp2





Now my question is how do we mount the clients in a HA fashion so that if
one of the servers goes down things stay online?



If we use this mount option:





#mount -t glusterfs [-o <options>] <volumeserver>:<volumeid>
<mount?
point>





with a single server then there is an issue if that server goes down,







I understand we could point to a volume file instead,  could that volume
file contain references for both servers?





#mount -t glusterfs [-o <options>] <path/to/volumefile>
<mountpoint>







Should/is the replication done on the server side or the client side?  It
seems it is being replicated by the server with this ?replica 2? volume,
so the question is how do we do the mount?



Is it possible that the volume information is synched between the 2
servers so that we can point the clients in our environment to either
Gluster servers?





Many thanks,