Multi-CPU environments

Consider the following when running Control-M/Restart in a multi-CPU environment:

• When the original job runs on the same CPU as the Control-M monitor, the DAARCH data set contains a set of pointers to the archived SYSDATA.

  If the restart job is also submitted on the same CPU, there are no special concerns. However, if the restart job can be submitted on a different CPU, certain concerns must be addressed if the restart is to be successful:

  • The SYSDATA of the previous runs must be archived on a disk that is shared between CPUs. (Read access is sufficient since the Control-M/Restart step only requires read access to the SYSDATA.)
  • The SYSDATA must also be in a catalog that can be accessed from all CPUs on which the job can run.

Note: If different catalogs are used in each CPU, or if JOBCAT or STEPCAT JCL statements are used in the job, the restart job is run in the same CPU in which it originally executed. If the restart job runs in a different CPU than the original one, and different catalogs are used, the results of the restart are unpredictable.

• When the original job runs on a different (remote) node than the Control-M monitor, the DAARCH data set contains the actual archived SYSDATA. Therefore, access of the SYSDATA during restart is not a problem.

  However, the considerations concerning catalogs remains. If all systems at the execution node do not share a common catalog, ensure that the restart executes on a system from which the required catalogs can be accessed.