[Linux-cluster] shared storage clustering

[Daniel Fernanduz]

Hello

           I am having two shared volumes (GFS) one for postgres and 
other to save files, so whenever i upload some files
           Informations related to the files will be updated in the 
postgres volume and actual file will be cpied to the Storage  
           volumes. Now database and file storage will be accessible 
thorough a shared mechanism.

           I had configured the failover cluster  in such a way that  
either node1 or node2 takes control when one of
           the node fails (Active /Passive clustering).

           My resource and service section of cluster.conf
                
                          <resources>
                        <script file="/etc/init.d/postgresql" 
name="Postgresql"/>
                        <clusterfs device="/dev/pgsqlvg/datalv" 
fstype="gfs" mountpoint="/pgsql" name="Postgres" 
options=""/>                        <script file="/etc/init.d/vsftpd" 
name="Vsftpd"/>
                        <clusterfs device="/dev/pgsqlvg/storage" 
fstype="gfs" mountpoint="/home/ftpuser" name="Storage" options=""/>
                        <ip address="151.8.18.147" monitor_link="1"/>
                </resources>
                <service autostart="1" domain="postgres" name="Postgres">
                        <script ref="Postgresql"/>
                        <clusterfs ref="Postgres"/>
                        <script ref="Vsftpd"/>
                        <clusterfs ref="Storage"/>
                        <ip ref="151.8.18.147"/>
                </service>

            Two shared volumes:
                *device="/dev/pgsqlvg/datalv" for postgres (<script 
file="/etc/init.d/postgresql" name="Postgresql"/>  for
                              database)
                device="/dev/pgsqlvg/storage" for storing files (  
<script file="/etc/init.d/vsftpd" name="Vsftpd"/>) for
                              loading files to storage.*

            When ever a node takes control two volumes had to be mounted 
and appropriate service for corresponding
            volumes had to be  started and  this cluster had to be 
accesible by an virtual ip ( <*ip address="151.8.18.147"
               monitor_link="1"/>.*

             For such situation whether my current conf is enough( 
*adding all resources to a single service*) or is there any
             actual configurations or appropriate procedures to achieve 
these types of settings inorder to avoid single point of
             failure.
            
             Thanks for the help in advance