Input/output errors

When a write error is found on the current JT log, the log that is in use is switched, and the NM subtask is posted to make a backup. The JL subtask is then posted to copy the JT log that had the output error. If a new output error occurs before the JL subtask has finished copying the first JT log, it is possible that there are no empty JT logs available. In this situation, the NM subtask is posted to stop. To reduce the risk of this occurring, you should always use a minimum of three JT log data sets.

If the dual-logging function is active, HCL Workload Automation for Z duplicates the JT records in the corresponding dual JT log. Dual logs are switched at the same time and in the same sequence as the JT logs. The number of dual JT data sets is determined by the number of normal JT data sets.

JT log sequencing requires that the log that is to be made current has previously been archived. If not, the NM subtask terminates. (Repeated write errors could cause this situation to occur if the JL subtask cannot copy the JT logs as fast as the normal-mode-manager subtask can switch JT logs.)

The subtask terminates if open errors are found on the JT archive data set. Also, the JL subtask terminates if it is unable to copy the JT log to the JT log archive data set because, for example, of an I/O error on the archive data set. If the JT log or archive data set cannot be opened, the subtask terminates. If the subtask terminates abnormally, the JT log data set is not made available for reuse.

HCL Workload Automation for Z automatically recovers from read errors on a JT log during restart. If the error occurs on the first record of the log, the Normal Mode Manager task regards the JT log as empty. A read error on a record other than the first is treated as an end-of-file on the file.