Hi Markus First, here is my index:
I get no record with the following query (explain plan: unique scan, uses the Index myindex):
But I know the record is there. So I try this (explain plan: full table scan):
And I suddenly get my record. So by either using or not using the index I get two different results. I think I know the fix for this: index rebuild. But what I don't get is how it is even possible to end up with such a corrupt index? And even more important: how can I prevent such a behavior in the future? (DBMS is Oracle 11g EE 11.2.0.2.0 64bit) PS: very nice book of yours by the way :) Cheers asked Nov 19 '13 at 11:14 BaseBallBatBoy |
Hi! As per definition, this behavior is nor your fault but a database bug. The reason it took so long to respond is because I thought I might be able to show a case how to circumvent this general rule by creative use of partitions swaps and/or use of NOVALIDATE. However, I couldn't build a showcase for your observation. It's possible to build other showcases that should never happen using these techniques but not a different result when using an index or not. Contact Oracle support to prevent this problem in the future. answered Nov 26 '13 at 11:33 Markus Winand ♦♦ |