+

Search Tips | Advanced Search

IBM MQ for z/OS recovery actions

Use this topic to understand some of the recovery actions for user detected and queue manager detected errors.

IBM MQ for z/OS can recover from program checks caused by incorrect user data. A completion and reason code are issued to the caller. These codes are documented in IBM MQ for z/OS messages, completion, and reason codes.


Program errors

Program errors might be associated with user application program code or IBM MQ code, and fall into two categories:


User detected errors

User detected errors are detected by the user (or a user-written application program) when the results of a service request are not as expected (for example, a nonzero completion code). The collection of problem determination data cannot be automated because detection occurs after the IBM MQ function has completed. Rerunning the application with the IBM MQ user parameter trace facility activated can provide the data needed to analyze the problem. The output from this trace is directed to the generalized trace facility (GTF).

We can turn the trace on and off using an operator command. See Use trace for problem determination on z/OS for more information.


Queue manager detected errors

The queue manager detects errors such as:

  • A program check
  • A data set filling up
  • An internal consistency error

IBM MQ analyzes the error and takes the following actions:

  • If the problem was caused by a user or application error (such as an invalid address being used), the error is reflected back to the application by completion and reason codes.
  • If the problem was not caused by a user or application error (for example, all available DASD has been used, or the system detected an internal inconsistency), IBM MQ recovers if possible, either by sending completion and reason codes to the application, or if this is not possible, by stopping the application.
  • If IBM MQ cannot recover, it terminates with a specific reason code. An SVC dump is typically taken recording information in the system diagnostic work area (SDWA) and variable recording area (VRA) portions of the dump, and an entry is made in SYS1.LOGREC.

Parent topic: Problem determination on z/OS

Last updated: 2020-10-04