RE: DataGuard

From: Boris Dali <boris_dali_at_yahoo.ca>
Date: Wed, 17 Jan 2007 14:53:27 -0500 (EST)
Message-ID: <367013.32204.qm@web32414.mail.mud.yahoo.com>

Laimutis,

For applications running on 10gR2 we switched from the homegrown scripts to a broker managed configuration. And yes one of the main reasons is that it is RAC aware. The other is the FSFO (fast-start failover). It's new, there are one-offs, but overall so far it works for us.

'Split-brain' scenario of having more than one database accepting connections in a broker configuration is prevented by the rule that two out of three participants in the broker config should agree before failover is attempted. There’s a brand-new paper on the MAA site about it:

http://www.oracle.com/technology/deploy/availability/pdf/MAA_WP_10gR2_FastStartFailoverBestPractices.pdf

Thanks,
Boris Dali.

Laimutis Nedzinskas <Laimutis.Nedzinskas_at_landsbanki.is> wrote:

> Once on 10gR1 I got my hands burned a little by DG
> process leaking RAM.
> Can this thing be trusted or is it better to rely on
> scripts?
>
> Next question: can a "Fast-Start Failover" be
> trusted? Did anyone try it?
> Running Observer on a separate machine raises a
> great lot of questions to me.
> What does happen if this observer starts standby
> database but primary is still online? It can be
> quite a mess if some clients do get connection to
> the primary and some to the standby. Or can the
> observer do kind of "fencing"(I´d say guaranteed
> killing) of the primary server? But how?
>
> Thank you in advance,
> Laimis N.

Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com

--
http://www.freelists.org/webpage/oracle-l

Received on Wed Jan 17 2007 - 13:53:27 CST