When performing a File Transfer (JOBF) to a z/OS target, the transfer fails when the destination dataset runs out of space (B37 abend). The z/OS Agent log shows thousands of duplicate error entries within a few milliseconds:
The Agent continues to read the input file until the end and repeatedly attempts to write, rather than terminating the transfer immediately upon the first fatal error.
This has been confirmed to be a defect in the z/OS Agent that causes it to incorrectly retry fwrite() operations after a non-recoverable z/OS I/O error (SB37).
This issue is tracked via DE209291. We will target a fix in a future release. Expected future releases are as follows:
To be notified of updates to this issue, subscribe to this article by following the steps in Subscribe to a Broadcom knowledge article by article or product