This is how I would approach the problem. 1) Do as was suggested and load the data without the index and find the dups in the 8i version. 2) If no dups are found in 8i, then the index build is running amok. Check your logs for some sign of trouble. 3) If dups are found in the 8i instance then look at your 9i database and find the same rows. If the rows are there and not reporting the duplicates, it's a database configuration difference (like the NLS conversation) or a bug. 4) If the duplicate rows are not in the 9i database then inspect your load log and see why the rows weren't loaded. On 7/4/07, Michael McMullen <ganstadba@xxxxxxxxxxx> wrote:
Hopefully this question won't devolve into a treatise on proper "troubleshooting". Troubleshooting is something I pride myself on. I knew I was taking a chance in posting a bare bones email and I paid the price. Oh well. But think of this. All rather puzzling. The export taken from an 8i database fails on import into the same (newly rebuilt 8i database) with primary key violation. This same export imports successfully into a 9i database on a different server. Same number of rows imported into the tables. There is no funny nls stuff. Primary key is two varchar columns, consisting of ipaddress and radius server name. So the dup on the 8i database doesn't show up on the 9i database. These are the types of oracle problems I find fascinating but since both the 8i server and database are desupported I'm really trying to push the client into upgrading and I am not going to spend alot of time troubleshooting it. Like I said, I was looking for a quick hit after spending days on recovery. Anyone guess what kind of mood I'm in? -- //www.freelists.org/webpage/oracle-l
-- Rumpi Gravenstein