A few weeks ago I posted a question about a query which was
intermittantly returning an ORA-01410 (invalid rowid) error. What
mystified me about this was that the query was not explicitely selecting
by rowid. I used analyze to check all indexes on the table referenced
in the error message and found no corrupted indexes.

My only clue was a google search on "ORA-01410" which yielded a response
from Jonathan Lewis to someone who was having a similar problem (
http://www.freelists.org/archives/oracle-l/08-2006/msg00075.html ). He
seemed to be saying that you could get this error if an index was being
rebuilt while the query was running. I had some discussions with our
application people and eventually learned that the software which runs
every hour to update this database, a data warehouse, from the
production database was dropping indexes before the update and
recreating them after the update. I just finished testing this on our
QA database ( start the query, drop and recreate the index) and got the
same error which the user had reported.

This is Oracle 8.1.7.0 on SunOS 5.9.

Thanks Jonathan.

Hope this will help someone else.

Peter Schauss

--
http://www.freelists.org/webpage/oracle-l