IRLM failures

When an IRLM fails, IMS subsystems using the IRLM cannot continue normal operation. IMS terminates active programs using the IRLM with abend 3303 and puts their messages on a suspend queue.

For wait-for-input programs using the IRLM, IMS either terminates them with abend 3303 on their next database call or gives them a QC status code. IMS inhibits subsequent program scheduling until IMS reconnects with the IRLM.

Before IMS can back out the IRLM failure, every dependent region with an intent to use a database must terminate its threads. IMS cannot terminate or reconnect to the IRLM until all dependent regions disconnect from the IRLM. Therefore, you must abnormally terminate any region that has not terminated (such as those programs that are waiting in an for a timer expiration or a for a reply to a Write-to-Operator with Reply (WTOR) message.

Restart the IRLM using the z/OS® START command. Then, reconnect IMS subsystems to the restarted IRLM using the z/OS MODIFYjobnameRECONNECT command. IMS automatically dequeues messages on the suspend queue.