WebLogic Server or Standalone Agent Crash Repository Database Failure

7-10 Oracle Fusion Middleware High Availability Guide the cluster. This address must be specified in the agent definition in the master repository. The scheduler singleton also routes all scheduled sessions startup requests to this address so that they are load balanced over the cluster. ■ Oracle Data Integrators master and work repositories database is configured with Oracle Real Application Clusters Oracle RAC to protect from database failures. Oracle Data Integrator components perform the appropriate reconnection and operations retries if database instance failure occurs.

7.3.2 Standalone Agent High Availability with OPMN

The standalone agent is a standalone Java process started from a command line interface. This agent is typically deployed locally on the source or target machines for optimal integration flow performances. You can use OPMN to start, stop and protect the standalone agent in this situation. To add a standalone agent to OPMN: 1. Edit the agentcreate.properties file contained in the ODI_ HOMEoraclediagentbin directory to match your agent and OPMN configuration. 2. Run the script that adds this agent to the OPMN configuration. ■ For UNIX: ODI_HOMEoraclediagentbinopmn_addagent.sh ■ For Windows: ODI_HOMEoraclediagentbinopmn_addagent.bat You can determine the status of Oracle HTTP Server using opmnctl: opmnctl status To start all agent components in an Oracle instance using opmnctl: opmnctl startproc process-type=odiagent To start a specific agent component, such as odiagent1, using opmnctl: opmnctl startproc ias-component=odiagent1 To stop all agent components in an Oracle instance using opmnctl: opmnctl stopproc process-type=odiagent To stop a specific agent component, such as odiagent1, using opmnctl: opmnctl stopproc ias-component=odiagent1

7.3.3 Oracle Data Integrator Protection from Failure and Expected Behavior

This section describes how an Oracle Data Integrator high availability cluster deployment and OPMN protects components from failure. This section also describes expected behavior in the event of component failure.

7.3.3.1 WebLogic Server or Standalone Agent Crash

If a WebLogic Server crashes, Node Manager attempts to restart it locally. If repeated restarts fail, the WebLogic Server infrastructure attempts to perform a server migration of the server to the other node in the cluster. While the failover takes place, the other WebLogic instance becomes the scheduler and is able to read, compute, and execute the schedule for all work repositories. A Coherence cache is used to handle the scheduler lifecycle. Locking guarantees the uniqueness of the scheduler, and event High Availability for Oracle Data Integrator 7-11 notification provides scheduler migration. Note that when an agent restarts and computes its schedule, it takes into account schedules in progress those in the middle of an execution cycle. These are automatically continued in their execution cycle beyond the server startup time. New sessions will be triggered as if the scheduler was never stopped. Stale sessions are moved to an error state and are treated as such when restarted. This session recoveryrestart is described in Section 7.2.1.1, Sessions Interruption and Section 7.2.1.2, Recovering Sessions. If a standalone agent crashes, OPMN restarts it locally. Existing sessions on the failing agent running become stale sessions, which are removed when the agent is restarted. Oracle Data Integrator agents may be down due to failure in accessing resources, or other issues unrelated to whether the managed server is running. Therefore, Oracle recommends that administrators monitor the managed server logs for cluster errors caused by the application. For information about log file locations, see Section 7.2.5.4, Oracle Data Integrator Log Locations and Configuration. The Oracle Data Integrator Console does not support HTTP session failover. The user must log into the Oracle Data Integrator Console again after a failure.

7.3.3.2 Repository Database Failure

The Oracle Data Integrator repositories are protected against failures in the database by using multi data sources. These multi data sources are typically configured during the initial set up of the system Oracle Fusion Middleware Configuration Wizard allows you to define these multi-pools directly at installation time and guarantee that when an Oracle RAC database instance that hosts a repository fails, the connections are re-established with available database instances. The multi data source allows you to configure connections to multiple instances in an Oracle RAC database. The Java EE agent uses WebLogic multi data sources that are configured during initial setup. The standalone agent uses the Oracle RAC JDBC connection string specified in odiparams.sh. For additional information about multi data source configuration with Oracle RAC, see Section 4.1.2, Using Multi Data Sources with Oracle RAC. Oracle Data Integrator implements a retry logic that allows in-flight sessions to proceed if a repository instance becomes unavailable and is restored at a later time. In an Oracle RAC enabled configuration, both in-flight and incoming session execution requests are served as long as an Oracle RAC node is available. This is supported in both the standalone and Java EE agents using the Retry Connection Count number and Connection Retry Delay time parameters. Users can configure these parameters when generating the WebLogic Server template for the Java EE agent and by editing odiparams.sh or odiparams.bat for the standalone agent. If Oracle Data Integrator Studio loses its connection to an Oracle RAC database, you will lose any Oracle Data Integrator Studio work performed since the last save operation. As a general practice, save your work on a regular basis when you use Oracle Data Integrator Studio.

7.3.3.3 Scheduler Node Failure