RE: Clusterware and nodeid
- From: "Martin Bach" <development@xxxxxxxxxxxxxxxxx>
- To: <s.cislaghi@xxxxxxxxx>, "'Oracle L'" <oracle-l@xxxxxxxxxxxxx>
- Date: Fri, 24 Jun 2011 07:31:24 +0100
Hi Stefano,
The instance names are now part of the ASM instance's profile, I assume the
problem lies there. You can check the instance to node name mapping by
viewing the profile. On my setup-four node RAC cluster with nodes
rac11gr2drnode{1,2,3,4} I found this:
$ crsctl stat res ora.asm -p
NAME=ora.asm
...
DEFAULT_TEMPLATE=PROPERTY(RESOURCE_CLASS=asm) ELEMENT(INSTANCE_NAME=
%GEN_USR_OR
A_INST_NAME%)
...
DESCRIPTION=Oracle ASM resource
...
GEN_USR_ORA_INST_NAME=
GEN_USR_ORA_INST_NAME@SERVERNAME(rac11gr2drnode1)=+ASM1
GEN_USR_ORA_INST_NAME@SERVERNAME(rac11gr2drnode2)=+ASM2
GEN_USR_ORA_INST_NAME@SERVERNAME(rac11gr2drnode3)=+ASM3
GEN_USR_ORA_INST_NAME@SERVERNAME(rac11gr2drnode4)=+ASM4
...
The mapping is performed in these lines. I'm not sure if it's supported but
you could try (on a test system!!!) to change the attribute
GEN_USR_ORA_INST_NAME@SERVERNAME(<hostname>)=+ASMx for the hosts in
question. You should probably involve oracle support, prompt them to give
you their blessing and then go ahead after having taken manual OCR & OLR
backups.
Hope this helps,
Martin
Martin Bach
Martin Bach Consulting Ltd
http://www.linkedin.com/in/martincarstenbach
http://martincarstenbach.wordpress.com
> -----Original Message-----
...
> Subject: Clusterware and nodeid
>
> Hi all,
>
> I've few 3-nodes clusters with grid infrastructure 11.2.0.2 installed for
single
> instances (configured as active/passive).
> One cluster was created due to some upgrade work in a different order, as
>
> node01 - id1 - +ASM1
> node02 - id3 - +ASM3
> node03 - id2 - +ASM2
>
> I've tried to remove node02 and node03 with standard procedure and rejoin
[...]
--
//www.freelists.org/webpage/oracle-l
Other related posts:
