Re: moving datafiles

>
> "craig" <someone_at_nowhere.com> wrote in message
> news:gBaXb.72411$fD.38968_at_fed1read02...
> > I'm no pro to take it all with a grain of salt. As far as I know,
making
> a
> > tablespace read only just means no more transactions can be made against
> the
> > database.
> >
> > Existing transactions could still be pending against that Tablespace?
>
> No, because if there's a pending transaction, then you can't actually make
> it read-only. And if there are dirty buffers floating around from finished
> transactions, then those are flushed to disk before the thing becomes
> read-only (ie, making something read-only causes a checkpoint).
>

Thanks for the info, I am trying to put a lot of stuff together in my head as a neophyte.

Not sure if I had the terminology all right but my meaning was that the tablespace goes into a transitional read-only mode (or so my 9i docs say) where pending transactions can still commit/rollback but no new ones can be started. Else, it would be hard to go into read-only mode in a db with activity. Guess thiss all happens before SQL returns the prompt to the user. Not sure how it could apply to this case anyhow.

What's strange is his doc that says to do it with read only. It does seem odd that there is no way to do this transition online. Guess in the world of high end hardware striping/lvm/etc moving data files isn't common enough to get Oracle's attention?

I guess the question is the affect of taking a highly utilized tablespace down. If timed right it wouldn't be out of commision too long? Received on Fri Feb 13 2004 - 16:37:58 CST