FIXED: ASM Upgrade from 10.2 to 11.2 erroring with ORA-01034: ORACLE not available

  • From: "Tyfanie Wineriter" <tyfaniew@xxxxxxxxxxx>
  • To: "'oracle-l'" <oracle-l@xxxxxxxxxxxxx>
  • Date: Fri, 9 Jul 2010 08:19:42 -0700

All - 

I wanted to follow up with what I believe has fixed this problem, in case
another user runs into it in the future & is searching through the archives.

The problem ended up being related to Metalink  [ID 398374.1].  It seems
that when the consultant set up our ASM, he created it in crs as "srvctl add
asm -n node1 -I ASM1 -o $ORACLE_HOME"

The problem with this is that our instances are named "+ASM1" and "+ASM2".  

We haven't actually had any problems with ASM except an error in the ASM
Logfile when it boots up, but it seemed to be a harmless error, as we were
able to upgrade ASM and run it without error otherwise.  It seems the 11gR2
OUI tries to be entirely too smart & pull things out of CRS that CRS is
actually ok with (why would CRS allow the asm instance to be added without
the "+" if the OUI would error out on it??).

Anyway.. I haven't finished the upgrade yet (I've re-registered ASM as +ASM
on only one node.. still have the other to do), but I have gotten past the
initial error and wanted to update the list.

Thank you to everyone for your ideas.  It was actually Peter that reminded
me to look at all the ASM logs, not just the ASMCA & alert_+ASM logs for
information.  That's when I looked at the
$ASM_HOME/log/$node/racg/...asm.log & found the PROC-4 error.

Thank you again!

~ Tyfanie Wineriter ~

Database Administrator
University of Oregon
1212 University of Oregon
Eugene, OR 97402-1212
(541) 346-1366

-----Original Message-----
From: oracle-l-bounce@xxxxxxxxxxxxx [mailto:oracle-l-bounce@xxxxxxxxxxxxx]
On Behalf Of Tyfanie Wineriter
Sent: Tuesday, July 06, 2010 5:35 AM
To: 'oracle-l'
Subject: RE: ASM Upgrade from 10.2 to 11.2 erroring with ORA-01034: ORACLE
not available

Peter:   I was hoping for something along the same lines, but there is
absolutely nothing in the ASM logs.  It doesn't startup, shutdown, or show
anything other than a continued existence.  The listener logs show nothing
either.  Of course, on further reflection, if Netca had done it's thing,
shouldn't I see SOMETHING in the listener logs?  Probably time to delve
further into what the NETCA part does.

Thomas:  I have not.  If Oracle has something for me today that fixes it, I
will update the list.

Tony: I'm not sure that could apply to a RAC ASM, as It's a 2-node cluster,
it's always in High Availability mode.  I have tried shutting both down, as
well as restarting them back up.  I could try just 1 instance open instead
of both.. yes, I think I'll try that when I get in today. (I need "Real
DBA's check their Oracle-L at 5am from home in their pajamas" on a t-shirt)

Mary Elizabeth:  Thanks, me too!

Thanks everyone for your time & thoughts.  I really appreciate having some
new ideas on what to try!


-----Original Message-----
From: oracle-l-bounce@xxxxxxxxxxxxx [mailto:oracle-l-bounce@xxxxxxxxxxxxx]
On Behalf Of Peter Hitchman
Sent: Monday, July 05, 2010 2:43 AM
To: oracle-l
Subject: Re: ASM Upgrade from 10.2 to 11.2 erroring with ORA-01034: ORACLE
not available

Hi Tyfanie,
I believe I have the same problem, but I am migrating between 11.1 and
11.2 on Linux 64 bit platfom. I tried a couple of time to migrate and
one time the asmca failed the other time it said it worked, but in
fact failed. In the ASM alert log I see the error:

ORA-00600: internal error code, arguments: [kfeInvokeLockCallback06],
[30], [], [], [], [], [], [], [], [], [], []
ORA-15240: attribute name SAGE_ONLY is invalid

Basically the ASM instance shutsdown, but the OUI does not recognzie
this and blindly stumbles on outputing "Oracle not available" in the

I have an SR open with Oracle but so far no response. Also I started a
thread on the list last week and a patch was suggested, but I thought
I should go via Oracle Support to be sure! I would be interested to
know if you see something similar in your ASM alert log.

One other thing, having the environment variable ORA_CRS_HOME set to
the original CRS home will also break the migration, because the wrong
crsctl programme gets called, assuming of course you are on a *nix




Other related posts:

  • » FIXED: ASM Upgrade from 10.2 to 11.2 erroring with ORA-01034: ORACLE not available - Tyfanie Wineriter