Steps for diagnosing problems on z/OS®

To diagnose a problem, follow these steps:

1. When the problem occurs, gather all the available diagnosis information for problem. Use the Problem diagnostic worksheet as a template for recording data. This might also include your internal problem report describing external symptoms, what might have triggered the problem, and what was done to recover, including the following types of diagnostic information:
   • Dumps
   • Traces
   • Error messages
   • SYS1.LOGREC entries
   • External symptoms
   • Hardware devices
   • Processor models
   • Any other information
   These topics can help you collect the data more effectively:
   • Gathering diagnosis data
   • Problem categories
2. After the problem type is identified, see these diagnosis procedures to identify the source and extract symptoms:
   • Diagnosing an abend
   • Diagnosing a system hang or wait state
   • Diagnosing a job or subsystem hang
   • Diagnosing a loop
   • Diagnosing an output problem
   • Diagnosing a performance problem
3. While using the procedure, build a search argument from the data collected. See Extracting problem symptoms and search arguments for more information.
4. Perform the search. Keep in mind that you might refine your search with more data from the problem. See Searching problem reporting databases for more information.
5. If the problem is not found in a database, report it as a new problem providing the documentation and information collected in Problem diagnostic worksheet. See Reporting problems to IBM for more information.

Tip: Sometimes information is found that is useful in routing the problem to the right place. For example, if the problem is an ABEND0C4 in module XYZ and your search shows multiple hits for ABEND0C4 and XYC and information about product from another company, contact that company or search that company's problem reporting databases.