PDA

View Full Version : Installing Dbvisit Standby on a Linux/Unix cluster (non RAC) system


Arjen Visser
05-29-2008, 04:33 PM
A 2 node cluster (with heartbeat) is a system where one node is active and the other node is in a dormant mode.

It is also called an active/passive cluster.

If the dormant node detects that the other node is no longer active, then it will activate itself and become the active node.
The cluster will usually have a (shared) file system which is either mounted on node 1 or node 2 depending on which is the active node.
The cluster will also have a virtual node name (or logical hostname) and a virtual IP address. This virtual name and IP address will be associated with the active node and will swap over to the other node if that node becomes active.

To setup Dbvisit Standby on a two node cluster you can either choose to either install Dbvisit Standby on the shared file system or a local file system.
- If you install Dbvisit Standby on a shared file system, you only need to install Dbvisit Standby once and maintain one copy of Dbvisit Standby.
- If you install Dbvisit Standby on a local file system, you will need to install Dbvisit Standby twice and maintain 2 copies of Dbvisit Standby.

Assume your primary node clusters are called node 1 and node 2. Your logical hostname is called vnode

Installing Dbvisit Standby on a local file system:
1) Install Dbvisit Standby as per normal on the local file system on node 1.
2) Edit the DDC file and set SOURCE to the logical hostname of the cluster (SOURCE=vnode).
3) You then must also set HOSTNAME_CMD to return the name of the logical hostname of the cluster.
Example:
HOSTNAME_CMD = /usr/local/bin/hostname.sh
Where /usr/local/bin/hostname.sh is a script which returns the logical hostname of the cluster. This can be a simple echo "logical cluster"
command or a command which returns the logical cluster name.

In this case the script /usr/local/bin/hostname.sh will simply contain:
echo vnode

Ensure that this script is not present on the standby server!
4) Copy over the Dbvisit Standby install directory to node 2 including the DDC file.
5) Copy the /usr/local/bin/hostname.sh to node 2.
6) Schedule Dbvisit Standby in the cron on the active node. Also add Dbvisit Standby to the cron on the dormant node, but do not activate the cron. Ensure that the failover process to the other node includes activating the Dbvisit Standby cron.

Installing Dbvisit Standby on a shared file system:
1) Install Dbvisit Standby as per normal on the shared file system on node 1.
2) Edit the DDC file and set SOURCE to the logical hostname of the cluster (SOURCE=vnode).
3) You then must also set HOSTNAME_CMD to return the name of the logical hostname of the cluster.
Example:
HOSTNAME_CMD = /usr/local/bin/hostname.sh
Where /usr/local/bin/hostname.sh is a script which returns the logical hostname of the cluster. This can be a simple echo "logical cluster"
command or a command which returns the logical cluster name.

In this case the script /usr/local/bin/hostname.sh will simply contain:
echo vnode

Ensure that this script is not present on the standby server!
4) Schedule Dbvisit Standby in the cron on the active node. Also add Dbvisit Standby to the cron on the dormant node, but do not activate the cron. Ensure that the failover process to the other node includes activating the Dbvisit Standby cron.