This section lists special requirements for Sun Cluster Support for Oracle Parallel Server/Real Application Clusters.
Before you decide which architecture to use for the Oracle UDLM and the Oracle relational database management system (RDBMS), note the following points.
The architecture of both Oracle components must match. For example, if you have 64-bit architecture for your Oracle UDLM, you must have 64-bit architecture for your RDBMS.
If you have 32-bit architecture for your Oracle components, you can boot the node on which the components reside in either 32-bit mode or 64-bit mode. However, if you have 64-bit architecture for your Oracle components, you must boot the node on which the components reside in 64-bit mode.
You must use the same architecture when you boot all of the nodes. For example, if you boot one node to use 32-bit architecture, you must boot all of the nodes to use 32-bit architecture.
The following list shows the locations of the data service log files.
Current log: /var/cluster/ucmm/ucmm_reconf.log
Previous logs: /var/cluster/ucmm/ucmm_reconf.log.0 (0,1,...) – This location is dependent on the Oracle UDLM package.
Oracle UDLM logs: /var/cluster/ucmm/dlm_nodename/logs – If you cannot find the Oracle log files at this location, contact Oracle support.
Oracle UDLM core files: /var/cluster/ucmm/dlm_nodename/cores – If you cannot find the Oracle log files at this location, contact Oracle support.
In an Oracle Parallel Server/Real Application Clusters environment, multiple Oracle instances cooperate to provide access to the same shared database. The Oracle clients can use any of the instances to access the database. Thus, if one or more instances have failed, clients can connect to a surviving instance and continue to access the database.
If a node fails, boot the node into maintenance mode to correct the problem. After you have corrected the problem, reboot the node. See the Sun Cluster System Administration Guide for Solaris OS for more information.
When you install this data service, ensure that you complete all steps of all procedures that precede installing the Oracle RDBMS software and creating your Oracle database before you reboot the nodes. Otherwise, the nodes will panic. If the nodes panic, you must boot into maintenance mode to correct the problem. After you have corrected the problem, you must reboot the nodes. The procedures that you must complete are listed in Table 2–1.
If a cluster node that is running an instance of Oracle Parallel Server/Real Application Clusters fails, an operation that a client application attempted might be required to time out before the operation is attempted again on another instance. If the Transmission Control Protocol/Internet Protocol (TCP/IP) network timeout is high, the client application might require a significant length of time to detect the failure. Typically, client applications require between three and nine minutes to detect such failures.
In such situations, client applications can use the Sun Cluster LogicalHostname resource for connecting to an Oracle Parallel Server/Real Application Clusters database that is running on Sun Cluster. You can configure the LogicalHostname resource in a separate resource group that is mastered on the nodes on which Oracle Parallel Server/Real Application Clusters is running. If a node fails, the LogicalHostname resource fails over to another surviving node on which Oracle Parallel Server/Real Application Clusters is running. The failover of the LogicalHostname resource enables new connections to be directed to the other instance of Oracle Parallel Server/Real Application Clusters.
Before using the LogicalHostname resource for this purpose, consider the effect on existing user connections of failover or failback of the LogicalHostname resource.
For information about the installation, administration, and operation of the Oracle Parallel Fail Safe/Real Application Clusters Guard option, see the Oracle documentation. If you plan to use this product option with Sun Cluster 3.1, note the points in the subsections that follow before you install Sun Cluster 3.1.
If you use the Oracle Parallel Fail Safe/Real Application Clusters Guard option with Sun Cluster 3.1, the following restrictions apply to hostnames that you use in your cluster:
Hostnames cannot contain special characters.
You cannot change the hostnames after you install Sun Cluster 3.1.
For more information about these restrictions and any other requirements, see the Oracle documentation.
If you use the Oracle Parallel Fail Safe/Real Application Clusters Guard option with Sun Cluster 3.1, do not use Sun Cluster commands to perform the following operations:
Manipulating the state of resources that Oracle Parallel Fail Safe/Real Application Clusters Guard installs. Using Sun Cluster commands for this purpose might cause failures.
Querying the state of the resources that Oracle Parallel Fail Safe/Real Application Clusters Guard installs. This state might not reflect the actual state. To check the state of the Oracle Parallel Fail Safe/Real Application Clusters Guard, use the commands that Oracle supplies.