Is it fair/safe to say if both fixes are deployed simultaneously, there should be NO occurance of this issue?
Lead Sys DBA
IBM Global Technical Services on assignmemt to Humana Inc.
T 502.476.2538 or 502.407.7266
Roy,was puzzled by this statement:
"In particular there are rows with the same primary key while non key columns can have different values containing the tablespace rids."Normally there cannot be more than 1 rows with same primary key, are you implying the table ends up with duplicate rows, or something else? I suspect the use of word "primary" was not quite right.Maybe it should mean duplicate index entries are created in the partitioning index?
Nic,How's it going?I don't quite follow:
"We have some parted tables that have a 2nd index that is a DPSI (with very similar columns to the PI). Both PI and DPSI use the Parting column as the 1st column in the index definition."If an Index starts with the table partitioning column(s), then it is a partitioning index or PI. DPSIs are partitioned indexes but do not commence with the partitioning column(s), right? Or at least not all of them. That seems to be inconsistent with your DPSI, or maybe your wording is slightly off.I doubt it is possible to force something to be DPSI, when it really is not? Is it more likely that you have more than one "Partitioning Index" which is very possible since Table Partitioning came in.DPSI normally cannot provide a rows set in its key sequence without merging rows from each partition used, unless predicates limit query to a single partition scanned. DPSIs can be useful (rarely) but is not a viable alternative to a normal NPI. Its a special case only thing.Was so funny when Table Partitioning came in that people were talking about converting their NPIs to DPSIs. Converting them to be PIs if lead columns matched partitioning, was far more likely.
Picky Micky Hannan (Ha ha)
UPS6756, #1544403 Oleander Drive, Suite CWilmington, NC 28403 Phone: (910) 660-8649Fax: (910) 523-5504