Re: export / import problem

Kalle wrote:

> Hi all,
>
> I have a strange problem:
>
> We have a production and development environments on different servers.
> I need to do every now and then copies from prod to another environment,
> this is done with exp/imp.
>
> Those servers are both behind a firewall so I need to copy a dump using
> an ftp....
>
> Environments are on HP-UX and both have same oracle version (9.2.0.4)
> and of course same character settings, both export and import will be
> done with UTF8 characterset. But reason or another there will happen
> some problems during import. I cannot validate or create unique indexes
> because of similar rows...
>
> How is this possible because there are no multiple rows on source
> database...
>
> Where does the character conversion happen or what is happening here? I
> am confused, well, this is not the first time, but still :)
>
> Any ideas where to find the reason or solution...
>
> Kalle
>
>

You need to identify the duplicate key values in the copy. Something like "SELECT id, count(*) FROM table GROUP BY id HAVING count(*) > 1".

It could be a charset conversion, or something else such as the constraint issue already posted. You'll know once you find the offending rows and check them against the source database. Both IMP and EXP need a consistent and correct NLS_LANG environment variable, explicitly set, to avoid conversion problems. Both utilities usually give warnings if there are charset conversion issues.

-Mark Bole Received on Fri Feb 25 2005 - 09:34:58 CST