Re: EMC Timefinder consistent split and Oracle --

From: sathish balasubramaniam <sat0789_at_gmail.com>
Date: Tue, 29 Mar 2011 16:04:25 +0400
Message-ID: <AANLkTike7f3S+=Z9-5fXHMYPcWpcWd8fnFrwBOpmOKVT_at_mail.gmail.com>



But i am not using a backup control file. I split the BCV copy using timefinder consistent split.
I do not put the database in the hot backup mode, split and then end backup. This is a recoverable copy.
In my case once the sync is done, i split the data DG. This is a restartartable copy.
 Technically speaking i just have to do a startup on the BCV side after the DG are mounted to bring up the database since instance recovery will take place.
But because there are few more archive logs due to sync happening at the flashback DG level, can i do a
RECOVER database until cancel after startup mount. Will it work on the restartable copy.

On Tue, Mar 29, 2011 at 3:50 PM, Harel Safra <harel.safra_at_gmail.com> wrote:

> Startup mount
> Recover database using backup controlfile until cancel;
> Alter database open resetlogs;
>
> It wont know that there're new archives, you'll have to tell it in the
> recover command (it'll prompt for archive names)
>
> Harel Safra
> Sent from my phone.
> On Mar 29, 2011 1:22 PM, "sathish balasubramaniam" <sat0789_at_gmail.com>
> wrote:
> > Agreed....but this is case where because of timefinder split, the clone
> > copy is equivalent to as if instance has been aborted.
> > No recovery is required other than reading online redo.
> > But becaue of sync is ongoing between primary and BCV archive DG, we have
> > few more archvie logs on the BCV side which are beyond the time when the
> > split happened.
> > I understand i don't have to do recover database , but a startup alone
> will
> > be enough to do instance recovery -- but that will not apply the new
> archive
> > logs.
> > My question is if i do a startup mount and do a recover database until
> > cancel, will my new archive logs that got generated after the split be
> > applied.
> > If the answer is yes, how does recover process know that there are new
> > archive logs present that we can apply.
> >
> >
> > Thanks,
> >
> >
> > On Tue, Mar 29, 2011 at 3:10 PM, Bobak, Mark <Mark.Bobak_at_proquest.com
> >wrote:
> >
> >> Seems to me, you are confusing instance recovery and database recovery.
> >> If instance crashes, and there's mo physical corruption, you'll get
> >> automatic instance recovery on next startup. If that happens, Oracle
> will
> >> automatically roll forward to time of crash, using active and current
> redo
> >> log. Archive logs will not be involved.
> >>
> >> If you issue 'recover database until...', now you're talking database
> >> recovery after restore from backup, and archive logs will need to be
> >> applied.
> >>
> >> Sent from my iPhone
> >>
> >> On Mar 29, 2011, at 6:55 AM, "sathish balasubramaniam" <
> sat0789_at_gmail.com>
> >> wrote:
> >>
> >> Basically what i am asking is i have had a database abort.
> >> And in this case a few more archive logs are generated after that due to
> >> timfinder sync.
> >> When i mount the database, control file is read and the locations and
> other
> >> info are verified.
> >> Then whe i issue recover database until command.
> >> I am assuming here that the recover database portion of the script will
> >> say and only instance recovery is required by looking at stop/start SCN
> and
> >> checkpoint SCN.
> >>
> >> So how will it know that there are few more new archive logs that need
> to
> >> be applied.
> >> Thanks,
> >> On Tue, Mar 29, 2011 at 1:05 PM, Kumar Madduri <ksmadduri_at_gmail.com
> >wrote:
> >>
> >>> Yes I think so and that should work.
> >>>
> >>>
> >>> On Tue, Mar 29, 2011 at 1:51 AM, sathish balasubramaniam <
> >>> sat0789_at_gmail.com> wrote:
> >>>
> >>>> How would i do that...Do i do a startup mount and say recover database
> >>>> until cancel and start applying archive logs generated between T1 and
> T2
> >>>> with or without cataloging in control file ??..Technically there is
> nothing
> >>>> to recover isn't it. It is an instance recovery situation.
> >>>> In that situation can we mount and say recover database until cancel
> and
> >>>> apply additional archive logs ??
> >>>>
> >>>> Thanks
> >>>>
> >>>> On Tue, Mar 29, 2011 at 12:19 PM, Kumar Madduri <ksmadduri_at_gmail.com
> >wrote:
> >>>>
> >>>>> Not sure why you want to catalog the archive logs in the control
> file.
> >>>>> It is possible to apply the archive logs between t1 and t2 and open
> the
> >>>>> database. If you are using bcv split then you are probably not using
> rman to
> >>>>> restore and recover. Then I dont think you need to let the control
> file know
> >>>>> (even with rman) about what archive logs you want to apply.
> >>>>>
> >>>>>
> >>>>> - kumar
> >>>>>
> >>>>> On Mon, Mar 28, 2011 at 11:59 PM, sathish balasubramaniam <
> >>>>> sat0789_at_gmail.com> wrote:
> >>>>>
> >>>>>> Hello All,
> >>>>>>
> >>>>>>
> >>>>>>
> >>>>>> I have question with regards to EMC timefinder consistent split and
> >>>>>> Oracle.
> >>>>>>
> >>>>>>
> >>>>>>
> >>>>>> I have 2 device groups
> >>>>>>
> >>>>>>
> >>>>>>
> >>>>>> Data which has datafile, controlfle and online redo log AND
> >>>>>>
> >>>>>> Flashback which has flashback and archive.
> >>>>>>
> >>>>>>
> >>>>>>
> >>>>>> Now I establish and sync up primary and BCV devices, then just
> before
> >>>>>> doing a split, I switch the logfile and do an timefinder consistent
> split
> >>>>>> only data device group.
> >>>>>>
> >>>>>> My flashback device group is still syncing up and accumulating new
> >>>>>> flashback and archive logs on the BCV side.
> >>>>>>
> >>>>>>
> >>>>>>
> >>>>>> Now assume that we have 4 new archive logs that has been generated
> >>>>>> after the split of data DG.
> >>>>>>
> >>>>>> Now I do an timefinder consistent split for the flashback DG.
> >>>>>>
> >>>>>>
> >>>>>>
> >>>>>> Data and redo device groups are of point in time T1 while
> >>>>>>
> >>>>>> Archive log and flashback logs are of point of time T2.
> >>>>>>
> >>>>>>
> >>>>>>
> >>>>>> I mount the BCV volumes and do a startup. At time point oracle will
> do
> >>>>>> a crash recovery and bring up the database as of point T1. – That is
> clear.
> >>>>>>
> >>>>>>
> >>>>>>
> >>>>>> My aim is to also apply the archive logs generated between T1 and
> T2.
> >>>>>>
> >>>>>> Now assume that instead of doing a startup, I do a startup mount and
> >>>>>> catalog the new archive logs in the control file and then I do a
> recover
> >>>>>> database.
> >>>>>> My question is will the new archive logs be applied by the recover
> >>>>>> process and can I bring up the database as of time T2 and open it up
> with
> >>>>>> resetlogs.
> >>>>>>
> >>>>>> Is this possbile..
> >>>>>>
> >>>>>> Appriciate your help in this regard...
> >>>>>>
> >>>>>> Thanks,
> >>>>>>
> >>>>>> Sat
> >>>>>>
> >>>>>
> >>>>>
> >>>>
> >>>
> >>
>

--
http://www.freelists.org/webpage/oracle-l
Received on Tue Mar 29 2011 - 07:04:25 CDT

Original text of this message