Howdy list.

We have an application which accesses Natural on CICS (v4.1) via Broker over TCPIP. It has VERY long running server transactions which set up access to Natural once only and then service requests inbound from Broker for the rest of the day. They execute mostly Natural programs, but occasionally have to call Cobol routines from Natural.

Surprisingly, abends are very rare on this system, with only the occasional ASRA (maybe once per day) occurring in Natural, and, even more rarely, the odd ASRA in a Cobol routine. Our problem is that the Cobol ASRA's cause the region to go SOS below the line every time without fail. It makes no difference which Cobol program abends - the result is always SOS. This same application area has another region which executes the same Natural and Cobol routines, only not via Broker (these requests come in over a VAN through LU6.2). They also have same ASRA abends at identical offsets in the Cobol routines (not to mention countless AE** abends which they won't fix), but this region does not have the SOS problem. These two regions have an identical setup apart from inclusion of TCPIP in the problematic one.

The area which goes SOS is the UDSA. We have examined the storage in question, and are unable to identify it. Working storage is above the line. The server task appears to recover (sort of....) from the abend and looks to have gone on to process other requests before finally grabbing the last bit of available storage in the UDSA some 2 mins after the ASRA.

Has anyone else seen this sort of thing happening? We are unable to recreate the problem on test.

Mike Lailvaux CICSTeam Standard Bank of S A Riverclub Computer Centre Tel +27 11 700 1445