Now that we have established that a "broken" record is really a "broken-up" (i.e., "split") record, there are approaches for understanding the split and dealing with the multiple records that resulted from the split. Use of the timestamp in the SMF record is not the way.
>From the RMF folks: Users who do not have 2.5 should use the fields in the Reassembly area of the RMF Product section mapped by ERBPRO to analyze the split records without reassembling them into a large SMF record. This can be done since the split records are self-containing, i.e. SMF header triplet (offset,length,number of data sections) are adjusted to reflect the right location and number of data sections. In other words, applications can analyze each single split SMF record and logically combine the extracted information in the application w/o reassembling multiple small records into one large record. The Reassembly area starts at offset '68'X of the RMF Product section and contains a 16-bytes header (mapped by RMFRHD) and multiple 4-bytes reassembly information blocks (mapped by RMFRBL). Header field RMFRBR can be read to understand how many split records have to be analyzed to get the "big picture". RMFRSQ tells the user the sequence number of the split SMF record. Use RMF Product section field RMFIST, RMFDAT, RMFINT and SMF header fields SMFRTY, SMFSTY, SMFSID (mapped by ERBHDR) to determine whether split records belong to the same original SMF record. z/OS 2.5 users can use the new GRBSMFR service as described on page 13ff in the z/OS Data Gatherer Programmer's Guide. https://www-01.ibm.com/servers/resourcelink/svc00100.nsf/pages/zOSV2R5gc315701?OpenDocument The service is 2.5 only and is not available on earlier releases Peter Relson z/OS Core Technology Design Peter rel...@us.ibm.com 1-845-435-8390 8+295-8390 ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
