Re: Flashback isn't, er, flashing.

I was playing around with flashback and list partitioning on 9.2.0.1 Windows and had problems with both (obvious bugs). Patching to 9.2.0.4 corrected both problems.

When flashing back, try using the "as of scn" instead and see if that works correctly. You can get the current scn by querying dbms_flashback.get_system_change_number().

Hope this helps .../Dom.

"Howard J. Rogers" <hjr_at_dizwell.com> wrote in message news:<40b4117e$0$31674$afc38c87_at_news.optusnet.com.au>...
> "Domenic" <domenicg_at_hotmail.com> wrote in message
> news:c7e08a19.0405251643.4f2b2c_at_posting.google.com...
> > The difference is in the patch level. Patch Windows to the same level
> > as Linux and it will work.
> >
> > Dom.
>
> Thanks for the advice, Domenic.
>
> However, on what authority do you make it? I mean, where have you noted that
> this is a known bug that is fixed by applying a particular patch?
>
> I suspect you can't answer that (but would be happy to be proved wrong),
> because sometimes it works on the 9.2.0.1 database, and sometimes it
> doesn't! Which rather suggests its not the patch level at all :-(
>
> Here's a test which, as far as I can tell, is almost identical to that which
> I included in my original post:
>
> SQL> select * from v$version;
>
> BANNER
> ----------------------------------------------------------------
> Oracle9i Enterprise Edition Release 9.2.0.1.0 - Production
> PL/SQL Release 9.2.0.1.0 - Production
> CORE 9.2.0.1.0 Production
> TNS for 32-bit Windows: Version 9.2.0.1.0 - Production
> NLSRTL Version 9.2.0.1.0 - Production
>
> SQL> select sum(sal) from emp;
>
> SUM(SAL)
> ----------
> 3976.02
>
> (The current salaries)
>
> SQL> exec
> dbms_flashback.enable_at_time(to_timestamp('26-05-2004:13:25:00','DD-M
> M-YYYY:HH24:MI:SS'))
>
> PL/SQL procedure successfully completed.
>
> SQL> select sum(sal) from emp;
>
> SUM(SAL)
> ----------
> 29025
>
> (The original salaries are visible. Now for the acid test):
>
> SQL> select sum(sal) from emp as of timestamp
> 2 (to_timestamp('26-05-2004:13:25:00','DD-MM-YYYY:HH24:MI:SS'));
>
> SUM(SAL)
> ----------
> 29025
>
> Ooo-errr. So now table-level flashback is producing exactly the same result
> as the session-level version, whereas yesterday it wasn't.
>
> Now maybe it's my timing, my typing, or the wind-speed direction, and if
> anyone can spot the obvious howler I'm making that's causing the error, I'd
> be grateful if you'd point it out. But the one thing this isn't, I venture
> to suggest, is the patch level of the database. Neither can it be my other
> suggestion from the original post: the operating system.
>
> I dare say I would have offered exactly the same advice to a customer who
> just wants the feature to work. But it's not quite what someone who wants to
> actually understand why the problem arises in the first place needs to hear.
>
> But thanks anyway
> HJR
Received on Wed May 26 2004 - 21:34:50 CDT