Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: 9iR2 export issue

Re: 9iR2 export issue

From: Joel Garry <joel-garry_at_home.com>
Date: 29 Dec 2003 14:17:43 -0800
Message-ID: <91884734.0312291417.76925ea2@posting.google.com>


"Goran" <goran99_remove_at_yahoo.com> wrote in message news:<bspaf5$o06$1_at_fegnews.vip.hr>...
> After (test) upgrade an 81740 db on HP-UX to 92040, we encountered an
> interesting behaviour.
> Standard backup/export procedure that we use on 8i production for export of
> several schemas doesn't work properly after upgrade... 9i export
> consistently encounters:
> "EXP-00056: ORACLE error 1466 encountered
> ORA-01466: unable to read data - table definition has changed"
> while attempting to export certain tables; export lasts for several hours
> and produces ~30GB dump file...
>
> Asuming that the tables really change their structure, as the error implies,
> we performed an investigation regarding the activities in the database; we
> learned from our users that structures of those several tables don't
> actually get changed (if they did, we would probably encounter the error in
> 8i production), but that users' procedures periodically trucate (few,
> several, many?) tables being exported.
>
> So we performed a little reproduction test on both versions: exporting a
> schema with two tables and truncating the second table while export of the
> first table is in progress We noted that 8i export actually never reports
> this error, but 9i export (from 9i test db) always reports the error.
>
> Since this is a complex and huge database and the export lasts for quite a
> long, we can't move the export time window. We can neither predict the
> truncate times, since there is user procedures activitiy all the time.
>
> Is this behaviour expected/normal/documented in 9i db compared to 8i?
> Any ideas how to workaround it?

Maybe
select to_char(created,'dd-mm-yyyy hh24:mi:ss') "Creation Time" object_name, object_type, object_id
from dba_objects
where created > sysdate;

Workaround would be CONSISTENT=N or don't let users truncate.

Also see metalink note 204334.1, maybe that can help.

jg

--
@home.com is bogus.
'The text on a home-brewing kit advised, "Let's Beer Great. Any time,
any where, all friends come together and drink beer with fun. It's
excellent to drink beer with all the people."' - Peter Rowe
Received on Mon Dec 29 2003 - 16:17:43 CST

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US