Problem diagnosis/recovery

If a job causes problems when given normal ThruPut Manager processing, there are various approaches you can take to diagnose and/or circumvent the problem. The following sections describe some of the actions that can be taken.

Bypassing job analyzer processing

A job can be routed around ThruPut Manager by using $TJ to assign the job an Execution Class. Ensuring the job is not processed before the $TJ can take effect might require halting ($ZI) the Job Analyzer. For example:

Note: If you bypass the Job Analyzer, ThruPut Manager JECL statements are not processed. Also, jobs that would normally be assigned Binding or Limiting Agents in JAL will be affected. This could have undesirable effects.

Note: Jobs that have not yet executed can be reprocessed by the Job Analyzer. Simply assign the primary class to the particular job with a $TJ command.

Job analyzer problem diagnosis/recovery

ABEND situation

If a job causes a problem in the Job Analyzer that results in an abend, then:

• ThruPut Manager generates an SVC dump containing diagnostic data about the problem.

• Message DTM1000I is issued, indicating the name and number of the job that is causing the problem. For example:

  DTM1000I JOB 1234, Jobname, HAS CAUSED AN ERROR
                   IN THE JOB ANALYZER AND WILL BE HELD
  
• The job being processed is queued in the class in which it was being processed (either the primary class or a deferred class).

• The job is held by Multi-hold Services (MHS) to prevent it from causing repeated failures of the Job Analyzer. To release the job, use the command:

  /MHS RELEASE jobnumber

Loop or waiting situation

If the Job Analyzer appears to be in a loop or waiting condition, you can cancel it using the z/OS CANCEL command. (The most likely sign that this is happening is that the Job Analysis queue is building up.) If the cause of the problem is unknown, include the DUMP keyword. This is shown below:

or if you have more than one Job Analyzer in your system:

The DUMP keyword causes an SVC dump to be taken, and the job to be requeued in hold status. Use DUMP if you suspect a Job Analyzer problem.

Use the z/OS DISPLAY ACTIVE command to determine if the Job Analyzer is in a loop by observing its CPU consumption:

Using DISPLAY commands

Often, problems such as the failure of a job to run can be explained by examining the output from the appropriate ThruPut Manager DISPLAY command. This is particularly useful if the commands are issued immediately upon encountering a problem. Use:

• DCS DISPLAY for DCS problems.
• JBS DISPLAY (Binding), JBS DISPLAY (Incompatible), and JBS DISPLAY (Jobs) for JBS problems.
• JCS DISPLAY for JCS problems.
• JLS DISPLAY for JLS problems.
• JOB DISPLAY for general job-related problems.

If you are calling ThruPut Manager Customer Support to report a problem, the output from these commands is quite valuable.

ThruPut Manager Support services problem diagnosis/recovery

If an abend occurs in TMSS, an SVC dump of the problem is taken for diagnostic purposes.

If the error occurred in a component of TMSS (such as the CVL processor), TMSS attempts to reinstate the function, unless its error retry limit is exceeded. If the error limit is exceeded, TMSS asks the operator's permission to retry the function. A reply of

Note: Two functions of TMSS, the Support Services Interface and ThruPut Manager Operator Services, are critical for continued operation of TMSS. If these tasks are not reinstated, TMSS terminates processing immediately.