Re: Backups using storage system mirroring/splitting

"Jurjen Oskam" <joskam_at_quadpro.stupendous.org> wrote in message news:slrnc1hgvr.j8r.joskam_at_calvin.quadpro.stupendous.org...
> In article <40181b3d$0$4256$afc38c87_at_news.optusnet.com.au>, Howard J.
Rogers wrote:
> >>
> > Because how is system B's control file supposed to know that System A
has
> > generated 2 or 3 more archives?
>
> It doesn't, that's true. But it doesn't know about sequence #5005 and
#5006
> *at all*. At the time of the "alter database backup controlfile", sequence
> #5005 and #5006 were not even generated yet. Those sequences were
generated
> in the time between the creation of the copy of the controlfile, and the
> split-off of the database. This means that the controlfile-copy (which is
> copied to system B to start the database) doesn't know about sequence
#5005
> and #5006, but those sequences are in the redolog files. Applying
> them from the redolog file works, but applying them from the generated
> archived logs doesn't work.

But that's my exact point. How does a database ever know which archives have been produced? By checking the control file (which is what you can do yourself by looking at v$archived_log, which itself takes information out of the control file). But how does a database know about what's in the online redo logs? Because it knows where the online redo log are, and must be, and can go read them. You don't need to a control file to tell you what SCNs are sitting in the online logs, you just look. You do need one to tell you what archives have been produced, to where, and what their start SCNs were.

Online redo and archived redo are, in other words, "house-kept" in two completely different ways, with two completely different mechanisms. So it is not at all surprising that the online stuff works and the archived stuff doesn't.

Regards
HJR

-- 
------------------------------------
Oracle insights at www.dizwell.com
------------------------------------