When reviewing the output from the system logger reports produced by IXGRPT1,
IXGRPT1J, and IXGRPT1L, look at the following key fields for CICS® system logs:
- The number of bytes deleted from primary storage should be close to the
number of bytes written
- The number of bytes deleted from the system log after writing to offload data sets should be very low:
- If this number is high, overhead is being incurred to move data to the
offload data set only to be later deleted.
- This is a key indicator that log tail deletion is not working as effectively
as it should.
- Check the MVS™ system log for any DFHRM0205 and DFHLG0743 messages from the affected CICS region.
- Look for long running tasks (using CICS monitoring data or a system monitoring
package), or check if AKPFREQ is too high.
- In general offloads are acceptable, but offloads triggered by NTRY FULL
indicators are not a good sign:
- NTRY FULL indicates that the entry to element ratio is too high
- It is probably the result of having unlike logstreams defined in the same
structure
- The offloads are being triggered by all the entries being used rather
than triggered by the HIGHOFFLOAD value.
- TYPE3 I/O counts should not appear in the statistics for coupling facility
log streams, because these indicate that I/O is being initiated when over
90% of the elements for the log stream are in use.
- Average buffer size is important because:
- If over 4K, the writes are asynchronous, with a greater overhead, rather
than synchronous
- Buffer size is used to determine the entry to element ratio.
- If MAXBUFSIZE specified on the log stream definition is less than 65532 bytes, the element size is 256 bytes.
- If MAXBUFSIZE is 65532 bytes, the element size is 512 bytes.
- The entry to element ratio is calculated as (average-buffer-size plus
4 divided by 4)
[[ Contents Previous Page | Next Page Index ]]