Re: Put Oracle 8i standby in managed recovery

From: joel garry <joel-garry_at_home.com>
Date: Fri, 14 Aug 2009 08:52:10 -0700 (PDT)
Message-ID: <da8c7574-02bd-478b-84ed-5ea7e1cb8094_at_l35g2000pra.googlegroups.com>

On Aug 14, 6:52 am, "Fabrice" <emouc..._at_spaminfonietest.fr> wrote:
> Thanks for your help.
> I know for Oracle but we can't migrate under an another version for the
> moment.
>
> And for NET8 how to manage network or standby failures in the time. For
> example during the week end.
> I'm afraid of the stop of the production database if Oracle can not archive
> its redologs because the remote localisation is unreachable.
>
> Do a manual recover per day is so terrible or impossible ?
>
> thanks
> fabrice

Please don't top post (that means, either put your response at the end or interleaved with what you are responding to).

I disagree with Sybrand, my experience from 8i was that there were some rare but nasty bugs that could even blow back and halt your production db, from a simple network outage.

It is a relatively simple matter to do periodic recoveries on the standby, indeed, that was the standard way to do it before managed recovery became available. It would help if you stated specifically which version of Oracle, ie 8.1.7.4, and which exact version and platform you are using.

A korn shell script like this run periodically from cron as the oracle owner on the standby works fine:

. /usr/local/bin/somescripttosetupenvironment sqlplus -s /nolog << EOF
connect / as sysdba
recover automatic standby database;
EOF The tricky part is copying over complete log files, that's where you have to reinvent the wheel - you want to be notified if something doesn't work. Most remote copy programs on unix keep the file unreadable until it is completely over, so using the automatic recover doesn't get bothered by half-copied logs. It's just on the sending side you want to be sure and only send completely written files. If your network is the slightest bit slow, it becomes worthwhile to compress before sending and uncompress after. Google around for scripts, and test carefully.

Here is a mini-faq about this group: http://dbaoracle.net/readme-cdos.htm

>
> "sybrandb" <sybra..._at_gmail.com> a écrit dans le message de news:
> 3bb2b94e-fb26-4b58-b435-adfa10d7d..._at_f37g2000yqn.googlegroups.com...
>
> > On 14 aug, 13:09, "Fabrice" <emouc..._at_spaminfonietest.fr> wrote:
> >> (re) hello
>
> >> Sorry for this new question.
> >> I don't use Net 8 to transfer archived redologs from my production
> >> database
> >> to the standby.
> >> I don't want to block the base in case of network failure.
>
> >> So I copy by script,each hours, the archived redologs file from the
> >> production to the local destination directory of my standby (parameter
> >> STANDBY_ARCHIVE_DEST). And regurlaly I execute this script :
>
> >> set autorecovery on
> >> recover standby database
>
> >> Is there a way to realize a managed "recover standby" without use net8 to
> >> transfer the logs. So I would like that the recover is automatic when I
> >> copy
> >> a archived redo in the right place on the standby.
> >> It seems that the command "RECOVER MANAGED STANDBY DATABASE" works only
> >> with
> >> a network transfer by net8.
>
> >> Thanks for your help.
>
> > No there is no way to realize a 'managed recover standby' without
> > using  Net8, and your concerns about Net8 are likely unjustified.
> > Apart from that, Oracle 8i has been desupported a long time ago, and
> > the Standby database feature has become much more robust in 9i
> > (already desupported) and 10gR2.
>
> > -----------
> > Sybrand Bakker
> > Senior Oracle DBA

jg

--
_at_home.com is bogus.
“I don't like Mondays. This livens up the day.”
http://www3.signonsandiego.com/stories/2009/aug/13/bn13spencer-denied-parole/?metro

Received on Fri Aug 14 2009 - 10:52:10 CDT