Using MySQL Toolkit in a HP Serviceguard Cluster Revision README: B.06.00

* (shared config) by putting the configuration files in a shared
filesystem
* (local config) by putting the configuration files on a single node,
then replicating the files to all other nodes in the cluster
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Local Configuration
If you opt to store the configuration files on a local disk, the
configuration MUST be replicated to local disks on all nodes
configured to run the package. If any change is made to the
configuration, the file must be copied to all nodes. As it is up to
the user to ensure the systems remain synchronized, the recommendation
is to put all configuration and data files on shared storage.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Shared Configuration
This is the recommended configuration. Here, the configuration and
database files are on shared disks, visible to all nodes. Since the
storage is shared, there is no additional work to ensure all nodes
have the same configuration at any point in time.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
To run MySQL in a HP Serviceguard environment:
* Each node must have the same version of the MySQL Database Server
software installed.
* Each node that will be configured to run the package must have
access to the configuration files. The most efficient way to ensure
all nodes have access to the configuration files is to put the
configuration files on shared storage that is accessible to all
nodes.
* Multiple databases configured in the same cluster will require
unique configuration per database. Each DB instance will require
its own volume group, set of configuration files, and copy of the
toolkit customized for the unique instance.
* All data files must be accessible to all nodes configured to run the
package. This means that all data files will need to reside on
shared storage that is accessible to all nodes.
* Each package is assigned an IP address called the relocatable (or
package) IP address.
When the package fails over from one node to another, the following
actions occur:
Primary Node
------------
1. The package is halted on the node where it is currently running.