How SA z/OS Uses Error Thresholds

Error thresholds influence whether SA z/OS recovers from an error situation.

For applications, you can define a critical threshold for restarting these applications. This is a number of error conditions within a certain time interval, for example, five error conditions requiring restart within one hour. During a condition requiring restart, SA z/OS checks whether the number of occurrences of the condition reaches the critical threshold. If it is reached, SA z/OS does not attempt to restart a resource.

For z/OS components, such as dump data sets or log data sets, you can define thresholds to limit the frequency of how often they may be deleted after they have filled up without an action being taken or a notification being sent to the operator.

Error thresholds also determine when you should be alerted to problems. The primary use of error thresholds is to track subsystem abends and ensure that the abend and restart cycle does not become an infinite loop, but they may also be customized for other uses.

Refer to IBM Z® System Automation Defining Automation Policy for information on how to define error thresholds. The following sections describe how to obtain information about them.