Just after midnight, we run an SMF EXTRACT, extracting 'YESTERDAY''s data. It was noticed that the extract job was mounting 7 input tapes instead of only the 1 it should have. After research, it was discovered that there were 19 DUMP datasets residing on 6 unique tape volumes that had various future dates in the Standard SMF Header, instead of the date that the record was created, or as defined 'moved to the SMF buffer'. All were SMF TYPE255 (Sysview) and Subtype27 (CICS transaction detail record). I located and ran the Utility to clean up the SMF Control DataSet (SCDS). An SMF LISTH command, and fed that output to PGM SMFIGADX, to generate DELETEX and ADDX control cards, cards were edited and submitted. SCDS is now clean. The question is how did SYSVIEW create SMF records with future dates. This only happened on one of our DEV machines, and our suspicion is that IBM's HOURGLASS product caused it.
Release : 15.0
Component : SYSVIEW
Hourglass intercepts SVC 11 (TIME macro).
SYSVIEW executes the TIME Macro when the SMF header is updated.
Other dates within a record have Store Clock, STCK, values which would be inline with the current date and time.
In the past it appears customers have coded exclusions for SYSVIEW modules to avoid this type of problem.
The exclusion would be similar to GSV*