Understand Failure and Recovery In Java Batch Processor

This topic only applies to self-managed Intelligent Advisor edition

Intelligent Advisor Java Batch Processor uses parallel processors to progress through cases quickly and efficiently. Each processor is allocated a subset of the overall case workload. Any errors that occur in one of the processors may prevent completion of the remaining cases only for that processor. For more information, see Error Handling and Avoid Out of Memory Errors.

For example, consider a batch operation using four processors that are allocated the work for cases 1-10, 11 to 20, 21 to 30 and 31 to 40, respectively. Suppose processor 2 encounters a fatal error while processing case 16, but successfully wrote the results for cases 11 to 15. All other processors completed successfully. In this case, only cases 16 to 20 need to be restarted.

It is important to design your batch processing jobs to clearly identify which cases were successfully processed. This allows only the remaining cases to be processed upon restarting the processing job. For more information, see:

• Identify Processed and Unprocessed Cases
• Present Only Unprocessed Cases