Because of the REUSE option, as observations are deleted from the data file, the empty pages are put on an available space chain. When the data set is re-populated, the pages are pulled off the available space chain in no particular order. The record id (RID), which is indexed, consists of the page number of the observation in the data file followed by some other fields.

In the test cases, the indexed variable only has one value, so the RIDs for that value are put into the index in a multiple occurrence chain, which is kept in ascending order. When an index is built initially for a REUSE table, or for any insertions for a non-REUSE table, the RIDs are given to the index in ascending order.

In the case of REUSE, since the data file pages may not in order, the RIDs must be inserted into the proper place in the multiple occurrence chain to keep it in ascending order. If the RIDs are not in order, the delete code may not be able to find the RID to delete, as in the test cases. There is a bug in the insertion code for multiple occurrences that is allowing RIDs to be placed into the chain out of order.

Select the Hot Fix tab in this note to access the hot fix for this problem.

Operating System and Release Information

A fix for SAS 9.1.3 (9.1 TS1M3) for this issue is available at:

A fix for SAS 9.1.3 (9.1 TS1M3) with Asian Language Support (DBCS) for this issue is available at:

A fix for this issue for Base SAS 9.21 is available. However, if you install SAS Maintenance after applying this hot fix, files delivered in the hot fix will not receive the proper Maintenance update. Please contact SAS Technical Support at the link shown below for information on accessing the hot fix.