Preliminary checks

Before you go further into looking for the cause of the problem, run through the following preliminary checks. These checks might highlight a simple cause or, at least, narrow the range of possible causes.

About this task

As you go through the questions, make a note of anything that might be relevant to the problem. Even if the observations you record do not at first suggest a cause, they might be useful to you later if you must carry out systematic problem determination.

Procedure

  1. Has the CICS® system run successfully before? If the CICS system has not run successfully before, it is possible that you have not yet set it up correctly. You can check that CICS installed correctly by running batch or online verification procedures. For more information, see Verifying the CICS installation. If CICS did install successfully, check the appropriate Upgrading information set for any possible impacts to your system. If you are currently upgrading to, CICS Transaction Server for z/OS®, Version 5 Release 4 ensure that you are aware of all the changes that were made for this release. For details, see What's New and Upgrading for the release from which you are upgrading.
  2. Are there any messages explaining the failure? If a transaction abends, and the task terminates abnormally, CICS sends a message reporting the fact to the CSMT log (or your site replacement). If you find a message there, it might immediately suggest a reason for the failure. Were there any unusual messages associated with CICS startup, or while the system was running before the error occurred? These messages might indicate some system problem that prevented your transaction from running successfully. If you see any messages that you do not understand, use the CICS messages transaction, CMAC, for online message information. If you do not have access to a CICS system to run the CMAC transaction, look in CICS messages for an explanation. A suggested course of action that you can take to resolve the problem might also be included with the explanation.
  3. Can you reproduce the error?
    1. Can you identify any application that is always in the system when the problem occurs?
      • Check for application coding errors.
      • Check that you sufficient resources are defined for the application, such as VSAM file strings. Typically, if the resources defined are not sufficient, you would find that the problem is related to the number of users of the application.
    2. Are you using exit programming interface (XPI) calls? If so, be sure to observe the XPI protocols and restrictions exactly. For programming information about the XPI, see The user exit programming interface (XPI). The exit programming interface is used to start a domain and enter its environment directly; using it incorrectly can cause severe CICS system problems. Here are some particular points for your attention:
      • Are the input parameters correct? If their format is not valid, they are rejected by the called domain, and an exception trace is made. If their values are acceptable to the domain but inappropriate for the system, they can cause unpredictable effects.
      • You cannot use some XPI calls within some of the user exits. If you do, the results can be unpredictable, and can cause CICS to stall or abend. See Global user exit programs for details on which exits can use XPI calls and which cannot.
    3. Consider your CICS system definition parameters if the problem is not related to any particular application. Poorly defined parameters can be the cause of problems in your system. You can find guidance about setting up your CICS system in, Specifying CICS system initialization parameters
    4. Does the problem seem to be related to system loading? If so, the system might be running near its maximum capacity, or it might be in need of tuning. For guidance about dealing with this problem, see Improving the performance of a CICS system.
  4. Does the failure occur at specific times of day? If the failure occurs at specific times of day, it can be dependent on system loading. Typically, peak system loading is at mid-morning and mid-afternoon, so those times are when load-dependent failures are most likely to happen. If your CICS network extends across more than one time zone, peak system loading might seem to you to occur at some other time of day.
  5. Is the failure intermittent? If an error is intermittent, particularly if it does not show the same symptoms, the problem might be more difficult to resolve. An intermittent failure can be caused by a “random” storage overlay. Furthermore, the transaction that caused the error might be deleted from the system long before the symptoms are seen. A method you can use to investigate random overlays is described in. Dealing with storage violations
  6. Have you made any service changes since the last successful run?
    1. Have you applied a PTF to CICS?
    2. Did it install successfully or did you get an error message during installation? If you installed it successfully, check with IBM® for any PTF error.
    3. Have any patches applied to any other program affected the way CICS interfaces with the program?
  7. Have you made any hardware changes since the last successful run?
  8. Have you made any software changes since the last successful run? If you installed a new or modified application, check for error messages in the output from the:
    • Translator
    • Compiler
    • Assembler
    • Linkage editor
  9. Have you made any administrative changes since the last successful run?
    1. Have you changed your initialization procedure, for example by JCL, CICS system initialization or override parameters, or z/OS Communications Server CONFIG/LIST?
    2. Has CICS generated any error messages during initialization?
    3. Have you installed any resource definitions defined using CEDA? If the definitions were made but not installed when CICS was last terminated, they might not be preserved over the termination and subsequent startup. In general, changes made to the CSD but not installed are not visible when the CICS system is warm started. However, if the change was in a group in the GRPLIST specified on a cold start, it is effectively installed during startup. (Changes which were installed are not visible after a cold start unless they were made to a group in the GRPLIST.)If START=AUTO was specified in the system initialization table, or as an override, you must examine the job log to find out how the CICS system last came up.
    4. Have you changed the configuration or the status of installed resources in CICS?

      The CICS configuration might have changed. Installed resources might be disabled or closed. For example, a file or program might be disabled. These changes are made using CEMT transactions, the Web User Interface, CICS explorer, or CICS applications. CICS provides a function to audit these changes by writing a message to the job log when these commands are issued. The message DFHAP1900 contains the date, time, transaction id, netname, user ID, and detail of the command. A system administrator, or anyone wanting to review audit records, can read the audit messages in the CICS job log to find any changes that were made to this CICS region. For commands using generic parameters, each individual command is audited, so a search for a specific file name, for example, would succeed in finding that file.

  10. Are specific parts of the network affected by the problem?
    1. Can you identify specific parts of the network that the problem affects? If you can, look for any explanatory message from the access method. Even if no message was sent to the console, you might find one in the CSNE log.
    2. Have you made any network-related changes?
    3. If the problem affects a single terminal, are your terminal definitions correct? Consider both the TERMINAL definition, and the TYPETERM definition it uses.
    4. If the problem affects a number of terminals, can you identify a factor that is common to all of them? For example:
      • Do they use the same TYPETERM definition? If so, it is likely that there is an error in that TYPETERM definition.
      • Is the whole network affected? If so, CICS might have stalled. See What to do if CICS has stalled for advice about dealing with CICS system stalls.
  11. Has the application run successfully before?
    1. Have any changes been made to the application since it last ran successfully? Examine the new or modified part of the application.
    2. Have you used RDO to create or alter a transaction, program, or map set? You must install these definitions before the resources are available to the running CICS region.
    3. If you changed any maps, have you created both a new phase (TYPE=MAP) and a new DSECT (TYPE=DSECT), and then recompiled every program using the new DSECT? Use the CEMT commands:
            CEMT SET PROGRAM(mapset) NEWCOPY
            CEMT SET PROGRAM(all programs) NEWCOPY
      .
    4. Have all the functions of the application been fully exercised before? Establish what the program was doing when the error occurred, and check the source code in that part of the program. If a program ran successfully on many previous occasions, examine the contents of any records, screen data, and files that were being processed when the error occurred. They might contain some unusual data value that causes a rarely used path in the program to be invoked.
    5. Check that the application successfully retrieved the records that it required at the time of the error.
    6. Check that all fields within the records at the time of the error contain data in a format acceptable to the program. Use CICS dump to do this. If you can reproduce the problem in a test system, you can use programming language debug tools and the CEDF transaction to check the data and solve the problem.
  12. The application has not run successfully before If your application has not yet run successfully, examine it carefully for any errors.
    1. Check the output from the translator, the compiler, and the linkage editor, for any reported errors. If your application fails to translate, compile or assemble, or link-edit cleanly into the correct phase library, it will also fail to run if you attempt to invoke it.
    2. Check the coding logic of the application. Do the symptoms of the failure indicate the function that is failing and, therefore, the piece of code in error?
    3. The following is a list of some programming errors commonly found in applications:
      • CICS areas are addressed incorrectly.
      • The rules for quasi-reentrancy are not followed.
      • Transient data is managed incorrectly.
      • File resources are not released.
      • Storage is corrupted by the program.
      • Return codes from CICS requests are ignored.


dfhs117.html | Timestamp icon Last updated: Monday, 5 June 2017