Clustering Automated Process Schedulers

If you have a large or mission-critical implementation of Crystal Enterprise, you will probably want to run several APS machines together in an APS cluster. An APS cluster consists of two or more APS servers working together to maintain the system database. If a machine that is running one APS fails, a machine with another APS will continue to service Crystal Enterprise requests. This "failover" support helps to ensure that Crystal Enterprise users can still access information when there is equipment failure.

This section shows how to add a new APS cluster member to a production system that is already up and running. When you add a new APS to an existing cluster, you instruct the new APS to connect to the existing APS database and to share the processing workload with any existing APS machines. For information about your current APS and APS cluster, go to the Settings management area of the CMC and click the Cluster tab.

Before clustering APS machines, note the following clustering requirements:

• Run each APS cluster member on the same operating system.
• Configure each machine similarly:
  • Install the same operating system service packs and patches.
  • Install the same version of Crystal Enterprise (including patches, if applicable).
  • Ensure that each APS connects to the APS database in the same manner: whether you use native or ODBC drivers, ensure that the drivers are the same on each machine.
  • Check that each APS uses the same database user account and password to connect to the APS database.
  • Run each APS service/daemon under the same account. (On Windows, the default is the "LocalSystem" account.)
• Install each and every APS cluster member on the same subnet.
• If the database server is multihomed, add one of its valid IP addresses to the hosts file on each APS machine. This ensures that each cluster member communicates with the database on the same IP address, without relying upon a WINS server for name to IP resolution.

Tip:    By default, an APS cluster name reflects the name of the first APS that you install, but the cluster name is prefixed by the @ symbol. For instance, if your existing APS is called CRYSTALAPS, then the default cluster name is @CRYSTALAPS. To modify the default name, see Changing the name of an APS cluster.

There are two ways to add a new APS cluster member. Follow the appropriate procedure, depending upon whether or not you have already installed a second APS:

• Installing a new APS and adding it to a cluster

  See this section if you have not already installed the new APS on its own machine.

• Adding an installed APS to a cluster

  Follow this procedure if you have already installed a second, independent APS on its own machine. While testing various server configurations, for instance, you might have set up an independent Crystal Enterprise system with its own APS. Follow this procedure when you want to incorporate this independent APS into your production system.

Note:    Back up your current APS database before making any changes. If necessary, contact your database administrator.