We are not seeing all transactions that should appear in SYSVIEW's CTRANLOG display. We do however see a large number of Discards on the CICSLOGR display.
This article describes what occurs when a transaction record is sent to the CICSLOGR, potential causes of lost records, and how to prevent record loss.
When a CICS transaction ends a transaction SMF record from the CICS region is created. This CICS region, along with other regions, can send records to SYSVIEW's CICSLOGR to be written to a log stream. The transaction record is sent to the CICSLOGR via a buffered queue.
There are two potential causes of lost records at the CICSLOGR level:
1. SYSVIEW's service class/dispatch priority is lower than that of CICS. The CICSLOGR task in the SYSVIEW STC is unable to keep up with the CICS regions that are queuing records to it. The CICS job log will show a message similar to the following:
GSVC065E (SDCS) - SYSVIEW CICSLOGR job SYSVIEW 242 is less than CICS job CICSPROD 244
2. A single CICSLOGR can only buffer a limited number of records. If a large number of high volume CICS regions are feeding a single CICSLOGR, the logger may not be able to keep up and the buffer may fill. If the buffer of the CICSLOGR fills, new transaction records cannot be queued in the buffer and will be discarded.
1. If SYSVIEW's service class/dispatch priority is lower than that of CICS, set SYSVIEW's service class equal to or higher than CICS. The best practice is to assign SYSVIEW to the WLM service class SYSSTC.
2. If using a single CICSLOGR with many busy CICS regions connected to it, and transaction records are being discarded, additional CICSLOGRs should be created so groups of CICS regions can be defined to the separate CICSLOGRs.
Link to the CICSLOGR section of the R14.1 wiki:
Release: FAQSO.00200-14.1-SYSVIEW-Performance Management