[Linux-cluster] help on configuring a shared gfs volume in a load balanced http cluster

[Alex]

Hello cluster experts,

I'm new here and new to cluster world too... I need some help, in order to 
setup a cluster in our organization.

Shortly, our schema is:

2 routers for HA and load balancing
- ar (active router)
- br (backup router)

3 http servers located internaly acting as real web servers (rs1, rs2, rs3) 
behind 
ar and br routers.
rs1=192.168.113.3/24
rs2=192.168.113.4/24
rs3=192.168.113.5/24

2 shared data servers (shd1, shd2)
shd1=192.168.113.6/24
shd1=192.168.113.7/24

1 server for cluster management (rhclm)
rhclm=192.168.113.8/24

I've configured ar and br routers for high availability and load banacing and 
everything is ok. Active router (ar) are forwarding http requests to VIP 
(floating) external ip address to internaly ip addresses of rs1, rs2, rs3 
webservers.

Now, i don't know how to:
- configure and group some hard disks on our shd1 and sdh2 servers to 
form a shared volume for our rs1, rs2, rs3 real servers (i suppose that the 
correct topic should be shared volume using GFS...)
- make usable this volume and act as DOCUMENT ROOT on our rs1, rs2 and rs3 
webservers.

All our servers are running centos 5.2 and has all updates installed.

On rhclm (192.168.113.8) i installed cana and created a cluster with 2 nodes: 
shd1 and shd2.

Cana, generated the following cluster.conf on shd1 and shd2 servers:

[root at shd1 ~]# cat /etc/cluster/cluster.conf
<?xml version="1.0"?>
<cluster alias="wwwdata" config_version="2" name="wwwdata">
        <fence_daemon clean_start="0" post_fail_delay="0"
post_join_delay="3"/>
        <clusternodes>
                <clusternode name="192.168.113.7" nodeid="1" votes="1">
                        <fence/>
                </clusternode>
                <clusternode name="192.168.113.6" nodeid="2" votes="1">
                        <fence/>
                </clusternode>
        </clusternodes>
        <cman expected_votes="1" two_node="1"/>
        <fencedevices/>
        <rm>
                <failoverdomains/>
                <resources/>
        </rm>
        <totem consensus="4800" join="60" token="10000"
token_retransmits_before_loss_const="20"/>
</cluster>

Now, on shd1 i am using hda for centos OS and hdb (1,2) i want to make it 
available to be used on shared volume:

[root at shd1 ~]# cat /proc/partitions
major minor  #blocks  name
   3    64   39082680 hdb
   3    65   19541056 hdb1
   3    66   19541592 hdb2
[root at shd1 ~]#

on shd2 i have hda for centos and hdc (1,2) i want it available to be used on 
shared volume:
[root at shd2 ~]# cat /proc/partitions
major minor  #blocks  name
  22     0   78150744 hdc
  22     1   39075088 hdc1
  22     2   39075624 hdc2
[root at shd2 ~]#

Using cana, i couldn't find a way to create a volume, grouping hdb1 (from 
shd1) together with hdc1 (from sdh2) in one volume. I want to do this for 2 
reasons:
- i want that volume to be mounted as document root on rs1, rs2, rs3 real 
webservers
- i want that volume to be easy to extend adding new hdd on the fly of other 
computers to this volume (new hdd slices of other new computers).

Can anybody tell me how can i do it?

I don't know that for this design if correct to have:
- all 5 servers (rs1, rs2, rs3, shd1, shd2) to be configured as nodes in 
the same cluster
or
- rs1, rs2, rs3 to be part of one cluster and shd1 and shd2 to form another 
cluster

I read section: A.2. Configuring Shared Storage in this document
http://www.centos.org/docs/5/html/Cluster_Administration/ap-httpd-service-CA.html
but is not what i want.

Can anybody help me. A link pointing me to the correct direction or a howto 
will be appreciated.

Regards,
Alx