ASM on 11.1.0.7

From: Kumar Madduri <ksmadduri@xxxxxxxxx>
To: oracle Freelists <Oracle-L@xxxxxxxxxxxxx>
Date: Mon, 16 Nov 2009 21:36:15 -0800
Hi
We have an issue with upgrading to 11.1.0.7
Current environment is 2 node rac on 10.2.0.4 and ASM running on clusterware
11.1.0.7

As part of upgrading the ASM and RDBMS to 11.1.0.7, I move the init.ora of
ASM from  10.2.0.4 to 11.1.0.7 ORACLE HOME and try to bring up the asm
instance and it fails with ORA-29702.
When you truss sqlplus session that is causing this issue you see this
message in oclskd.log

2009-11-16 21:00:43.514: [  OCLSKD][1]clsskd_main: Kill daemon registered
with CSS.  Waiting for work.

Also in the trace directory I see a trc file generated for lmon which is
like this.

*** 2009-11-16 03:27:33.074
*** SESSION ID:(218.1) 2009-11-16 03:27:33.074
*** CLIENT ID:() 2009-11-16 03:27:33.074
*** SERVICE NAME:() 2009-11-16 03:27:33.074
*** MODULE NAME:() 2009-11-16 03:27:33.074
*** ACTION NAME:() 2009-11-16 03:27:33.074

GES resources 64 pool 1
GES enqueues 128
GES IPC: Receivers 1 Senders 1
GES IPC: Buffers Receive 1000 Send (i:100 b:100) Reserve 100
GES IPC: Msg Size Regular 416 Batch 8192
Batching factor: enqueue replay 201, ack 224
Batching factor: cache replay 126 size per lock 64

*** 2009-11-16 03:27:34.448
kjxggin: CGS tickets = 1000
kjxgrdmpcpu: CPU Total 16 Core 16 Socket 16 OCPU 16
kjxgrdmpcpu: High load threshold 20480

*** 2009-11-16 03:27:34.540
kjxgmrcfg: Reconfiguration started, type 1
kjxgmcs: Setting state to 0 0.

*** 2009-11-16 03:27:34.544
Name Service frozen
kjxgmcs: Setting state to 0 1.
kjxgmmeminfo: node(0) info has a wrong size (112 120)
ksirValidateModuleInfo: action = 10 startup = 0
Name Service Mode: multi (0x21)
kjxgmjoin: substate 1 action fails
2009-11-16 03:27:34.545977 : IMR recording device closed, terminating IMR

*** 2009-11-16 03:27:34.589
Global Enqueue Service Shutdown
If I relink the oracle home libraries with rac off option then the asm
instance comes up fine.

Also in the upgrade doc 429825.1, one of the pre upgrade steps is to set
cluster_database to false for the rdbms instance. We did the same for ASM
instance too. It seems to me that if cluster_database is set to false then
you have to relink with rac off option (in the sense they are directly
proportional). Is this the case?
My concern in bringing up the asm instance with off option is that , I can
do the upgrade but when I try to rac enable after the upgrade, I will hit
the same isssue because if you look at it- there is nothing that changes
between now and the upgrade as far as the 11.1.0.7 oracle binaries are
concerned.

I found one related link
http://forums.oracle.com/forums/thread.jspa?messageID=3715198 here but that
is on linux and we are on solaris 64. From Metalink, I cannot view patch
7687540 for Solaris or Linux.

Any inputs ?

Thank you
kumar
ASM on 11.1.0.7

Other related posts:
