X

Alejandro Vargas' Blog

  • February 28, 2007

ORA-15080 ORA-15063 Starting ASM instance on Backup Server

Alejandro Vargas
Technical Leader, ACS Global Delivery, Infrastructure & BigData

A couple of interesting tests today.

  1. Got a call reporting ORA-15080 ORA-15063 when trying to start an ASM srdf copy on a backup server. ASMlib listdisks and querydisk returned the same output as in production.

  2. Got a request to configure recovery  on a backup server for a new RAC database.
    The backup server already had configured automatic recovery of a 10.2.0.1 production database, with ASM based on raw devices.
    The database to add was on 10.2.0.3, and ASM disks managed by ASMLib.
I started to work on point 2.
The question that rose first was:

Can I use the same 10.2.0.1 ASM instance, to hold raw device based ASM disks from a 10.2.0.1 database, and ASMLib based ASM disks of a 10.2.0.3 database.

That was an interesting thing to check.

I started by preparing a separate init+ASM.ora to mount the new ASMLib based ASM disks.

I've brought the parameter asm_diskgroups from the RAC ASM instance init.ora, there was no asm_diskstring defined there.

When starting up the ASM instance I've got the same error I was asked about by the other customer (on point 1):

ORA-15032: not all alterations performed
ORA-15063: ASM discovered an

insufficient number of disks for diskgroup "DATADG"

I did check also the output of oracleasm listdisks and querydisk, and got same output as in production.
I've added asm_diskstring to init.ora and did try a couple of different strings, with the same result: ORA-1563
Then I did check  what was the path on the production database:

SQL> select NAME,PATH from  v$asm_disk;
NAME                           PATH
------------------------------ ------------------------------------------------------------------------
VOL1                            ORCL:VOL1

With this information I did set up the asm_diskstring parameter to 'orcl:vol*' and the diskgroup was mounted successfully.

I did mount to the 10.2.0.3 database, recovered it and opened it read only.

The next question was, can I mount also the raw devices belonging to the 10.2.0.1 database, mount and recover the database, and open it read only?

To do this I did add the 10.2.0.1 disk groups and discovery string to the ASM init.ora:

*.asm_diskgroups='DATADG','SPDATADG','SPARCHDG'
*.asm_diskstring='ORCL:VOL1','/dev/raw/*'

I did start the ASM instance, that was able to mount all disk groups:

SQL> startup
ASM instance started

Total System Global Area  130023424 bytes
Fixed Size                  2019032 bytes
Variable Size             102838568 bytes
ASM Cache                  25165824 bytes
ASM diskgroups mounted
SQL> exit

Disconnected from Oracle Database 10g Enterprise Edition Release 10.2.0.1.0 - 64bit Production
With the Partitioning, OLAP and Data Mining options

{oracle} /bksrv/app01/oracle/product/10gASM/dbs [bksrv] > asmcmd lsdg
State    Type    Rebal  Unbal  Sector  Block       AU  Total_MB  Free_MB  Req_mir_free_MB  Usable_file_MB  Offline_disks  Name
MOUNTED  EXTERN  N      N         512   4096  1048576    237366   225290                0          225290              0  SPARCHDG/
MOUNTED  EXTERN  N      N         512   4096  1048576    949464   550593                0          550593              0  SPDATADG/
MOUNTED  EXTERN  N      N         512   4096  1048576    118236    79056                0           79056              0  DATADG/

I did mount, recover and open readonly the 10.2.0.1 database.

At the end I had a 10.2.0.1 ASM Instance , serving 10.2.0.1 and 10.2.0.3  databases, with a mixed configuration of ASM disks on raw devices and ASMLib.

Not a configuration to repeat, but was interesting to try it.

Join the discussion

Comments ( 5 )
  • Jorge Suarez Wednesday, February 28, 2007
    sorry for an off topic question, but I'm stuck with a RAC situation, i already hace a db10g running for some time, now i installed crs and asm on 2 nodes (the old one plus a newcomer), I tried to run the rconfig script to turn my existing database to RAC, but it seems that because the db wasn't installed using crs (meaning, it didn't got installed simultaneously like asm) doesn't want to cooperate, the message is "Clusterware is not configured".
    theres is a way to cluster-ify a existing database home? everywhere on metalink says first crs, the install the clean db, with in my case I want the opposite.
    Any help will be greatly appreciated (BTE, I'm with a SR right now, but so far isn't helping)
    cheers
    JL
  • Mihail Thursday, March 1, 2007
    And what is 10g R3? Any more info on that release?
    <quote>The database to add was on 10g R3</quote>
  • Alejandro Vargas Thursday, March 1, 2007
    Thanks for the comment, there is no rdbms 10gR3 yet, I ment 10gR2 patch set 10.2.0.3.
  • Yossi Nixon Sunday, April 1, 2007
    you could have check the problem by running kfod asm_diskstring='ORCL:*' disks=all
    try it ;)
  • Murtuja Khokhar Saturday, April 12, 2008
    Hi,
    I have solved same error by setting appropriate permissions.I was getting error just after installing asm.Disk group was mounted on first node but it was not there in second node.
    I have executed bellow commands as a root user.
    chown oracle:dba /dev/rdsk/c3t1d0s3
    chown oracle:dba /dev/rdsk/c3t1d0s4
    chown oracle:dba /dev/rdsk/c3t1d0s5
    chown oracle:dba /dev/rdsk/c3t1d0s6
    chown oracle:dba /dev/rdsk/c3t1d0s7
    chmod 640 /dev/rdsk/c3t1d0s3
    chmod 640 /dev/rdsk/c3t1d0s4
    chmod 640 /dev/rdsk/c3t1d0s5
    chmod 640 /dev/rdsk/c3t1d0s6
    chmod 640 /dev/rdsk/c3t1d0s7
Please enter your name.Please provide a valid email address.Please enter a comment.CAPTCHA challenge response provided was incorrect. Please try again.