Question & Answer
Why would CICS abend with message ICH409I 878-010 during FRACINT processing followed by DFHXS0001 abend 878 AKEX in DFHXSAD? I am running out of MVS storage below the line. Here are some error messages from the console log:
ICH409I 878-010 ABEND DURING FRACINT PROCESSING IEA794I SVC DUMP HAS CAPTURED DUMP TITLE=CICS SYSTEM DUMP - REQUESTED BY DFHXSS DFHXS0001 An abend (code 878/AKEX) has occurred at offset X'FFFF' in module DFHXSAD
In this case, the problem started happening after upgrading to a new release of z/OS but that might just be a coincidence.
Reviewing a dump from the 878, I see the problem involves below the 16M line storage. Subpool 130 Key 0 ( SP130 K0) contains over 2M of storage. Most of the storage is in x'1000' byte chunks, with various FREE areas associated with it.
If you have a dump of the abend 878, you can look at the DQEs and FQEs in storage at the time of the failure by entering IPCS command verbx vsmdata 'nog summ'. Here are some examples of the DQEs and FQEs from looking at storage:
DQE: Addr 00659000 Size 2000 FQE: Addr 00659000 Size 8 FQE: Addr 00659570 Size 8 FQE: Addr 006597B8 Size 8 FQE: Addr 00659918 Size 10 FQE: Addr 00659AE8 Size 10 DQE: Addr 00885000 Size 1000 FQE: Addr 00885000 Size 110
Browsing the storage addresses, you will see security related control blocks with eye catchers: ACEE CGRP ACEX.
Turns out a RACF exit ICHDEX01 was linked to run in RMODE 24, which caused these control blocks to be getmained below the line.
In order for the changes to the ICHDEX01 exit to go into effect, an IPL needs to be done. You can read more about addressability of RACF exits in section Extended addressing for exits of the z/OS Security Server RACF System Programmer's Guide.
CICS/TS CICSTS CICS TS CICS Transaction Server
03 October 2016