RE: Conditional failover rules across nodes in service

I would try server callouts to deal with these. Tying this down without too much thinking, I'd create a service for OLTP on nodes 1 and 2, both preferred. Then, if you get a node down event I'd check if both nodes are affected, and if so modify the service definition of OLTP to make nodes 3 and 4 preferred.

However, you need to think about what happens if the nodes come back ... A reverse process to this?

Just a quick thought...

Martin Bach
Martin Bach Consulting Ltd
http://www.linkedin.com/in/martincarstenbach http://martincarstenbach.wordpress.com

> -----Original Message-----
> From: oracle-l-bounce_at_freelists.org [mailto:oracle-l-bounce_at_freelists.org]
> On Behalf Of Herring Dave - dherri
> Sent: 27 May 2011 15:47
> To: ORACLE-L
> Subject: Conditional failover rules across nodes in service
>
> Folks,
>
> I've got a client request for conditional failover of connections within a
> service definition that I don't believe is possible but thought I'd double
check
> with this group. The db in question is on 10.2.0.2, RHEL, and a 4-node
RAC.
> Nodes 1 and 2 are for OLTP and nodes 3 and 4 are for Batch. This
distinction is
> managed by services:
>
> % srvctl config service -d $db_unique_name -a oltpapp PREF: node1 node2
> AVAIL: TAF: NONE batchapp PREF: node3 node4 AVAIL: TAF: NONE
>
> What they want to happen is if ONLY node1 or node2 goes down, connetions
> failover to the other oltp node. But, if BOTH node1 and node2 go down,
then
> service oltpapp should failover to nodes 3 and 4. I can obviously modify
> service definitions when these failures happen to meet their needs but
they
> want these failover rules to happen automatically.
>
> I'm pretty sure what they want can't be done. If I define nodes 3 and 4
as
> "Available", then a 1-to-1 relationship gets set so that if node1 goes
down:
>
> % srvctl remove service -d $db_unique_name -s oltpapp % srvctl add service
> -d $db_unique_name -s oltpapp -r "node1,node2" -a "node3,node4"
>
> ... oltpapp connections failover to node3. The same for node2 to node4.
>
> Is there any way of doing what the client is requesting?
>
> DAVID HERRING
> DBA
> Acxiom Corporation
> EML   dave.herring_at_acxiom.com
> TEL    630.944.4762
> MBL   630.430.5988
> 1501 Opus Pl, Downers Grove, IL 60515, USA WWW.ACXIOM.COM
>
> The information contained in this communication is confidential, is
intended
> only for the use of the recipient named above, and may be legally
privileged.
> If the reader of this message is not the intended recipient, you are
hereby
> notified that any dissemination, distribution or copying of this
communication
> is strictly prohibited. If you have received this communication in error,
please
> resend this communication to the sender and delete the original message or
> any copy of it from your computer system. Thank you.
> --
> http://www.freelists.org/webpage/oracle-l
>

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