Question & Answer
Question
I see IEE986E messages indicating that SMF buffer space usage went from 25 % to 100%. Then CICS returns message DFHMN0101 with a SMF return code of hexadecimal 28. At the same time, my colleague was using CEDF in CICS, and it was after entering the PF4 key to "SUPPRESS DISPLAYS" that the SMF buffer space became 100% full. I just upgraded the region from CICS Transaction Server for z/OS (CICS TS) V4.2 to V5.3.
What is the difference between CICS TS V5.3 and earlier CICS releases in terms of the RATE at which these Debug buffers (SMF records) are being written?
What is the difference between CICS TS 5.3 and earlier CICS releases in terms of the SIZE of the Debug buffers (SMF Records) that are being written?
These are the messages in my operlog:
16:22:19.66 *IEE986E SMF HAS USED 25% OF AVAILABLE BUFFER SPACE
16:22:20.59 *IEE986E SMF HAS USED 26% OF AVAILABLE BUFFER SPACE
16:22:21.51 *IEE986E SMF HAS USED 28% OF AVAILABLE BUFFER SPACE
16:22:22.42 *IEE986E SMF HAS USED 29% OF AVAILABLE BUFFER SPACE
16:22:23.34 *IEE986E SMF HAS USED 31% OF AVAILABLE BUFFER SPACE
:
16:22:58.76 *IEE986E SMF HAS USED 81% OF AVAILABLE BUFFER SPACE
16:22:59.69 *IEE986E SMF HAS USED 82% OF AVAILABLE BUFFER SPACE
16:23:01.58 *IEE986E SMF HAS USED 85% OF AVAILABLE BUFFER SPACE
16:23:03.42 *IEE986E SMF HAS USED 87% OF AVAILABLE BUFFER SPACE
16:23:04.32 *IEE986E SMF HAS USED 89% OF AVAILABLE BUFFER SPACE
16:23:05.27 *IEE986E SMF HAS USED 90% OF AVAILABLE BUFFER SPACE
16:23:07.11 *IEE986E SMF HAS USED 92% OF AVAILABLE BUFFER SPACE
16:23:07.98 *IEE986E SMF HAS USED 93% OF AVAILABLE BUFFER SPACE
16:23:08.93 *IEE986E SMF HAS USED 95% OF AVAILABLE BUFFER SPACE
16:23:09.88 *IEE986E SMF HAS USED 96% OF AVAILABLE BUFFER SPACE
16:23:10.82 *IEE986E SMF HAS USED 98% OF AVAILABLE BUFFER SPACE
16:23:11.74 *IEE986E SMF HAS USED 100% OF AVAILABLE BUFFER SPACE
16:23:12.93 *IEE979W SMF DATA LOST - NO BUFFER SPACE AVAILABLE TIME=16:23:12
16:23:12.93 +DFHMN0101 SMF error - SMF return code X'28'.
16:23:12.95 +DFHMN0101 SMF error - SMF return code X'28'.
16:23:12.95 +DFHMN0101 SMF error - SMF return code X'28'.
:
:
Answer
In terms of the rate, in both CICS TS V5.3 or lower release, CICS will attach 2 CEDF tasks (before and after) for each EXEC CICS command and each call to an external resource manager (for example, a DB2 SQL call). At the end of each CEDF, CICS will write the task SMF data to the buffer. In this case, even though PF4 was entered to suppress the DISPLAYs, many CEDF tasks were still attached and the SMF data was written to the buffer at the end of each task. This has not been changed between the old releases of CICS and CICS TS V5.3.
In terms of the SIZE, the size of SMF records for CICS TS V5.3 is x'D98' or 3480 in decimal. In CICS TS V4.2, the size of SMF records is much smaller, probably a couple hundreds bytes less.
There are 3 ways to fix this problem:
Specify DFHMCT TYPE=INITIAL macro COMPRESS=YES in your monitoring control table (MCT) to turn on data compression for monitoring records. If you specify the system initialization parameter MCT=NO, the default MCT built by CICS specifies COMPRESS=YES. If you do not want to compress monitoring records, you must specify COMPRESS=NO in your MCT.
Exclude specific data fields from the performance data record. For example, if you do not use FEPI, you can exclude it. See DFHMCT examples for more examples.
Specify MN=OFF in your system initialization table (SIT) to turn off monitoring if you won't be using the SMF data from the region.
Product Synonym
CICS/TS CICSTS CICS TS CICS Transaction Server
Was this topic helpful?
Document Information
Modified date:
27 October 2017
UID
dwa1407276