Upgrading ODA to (12c)

Today I upgraded our Oracle Database Appliance to

This is totally unsupported by Oracle, absolutely not recommended for a production environment and I’m going to have to perform a bare metal restore of the unit in order to get it back into a supported state. However, we’re just playing so that’s fine!

The first task was getting the software across to the ODA. Remember that GI actually runs as the “grid” user, so the GI zip files will need to be copied and extracted in /home/grid. The database software runs as “oracle” so that can go in /home/oracle.

Then you need to create the directory to be used for the GI software. Remember to do this on both nodes –

mkdir /u01/app/
chown grid:oinstall /u01/app/
chmod 755 /u01/app/

Then run cluvfy from the 12c software you extracted. I did it like this –

./runcluvfy.sh stage -pre crsinst -upgrade -n oda-nodea,oda-nodeb -src_crshome /u01/app/ -dest_crshome /u01/app/ -dest_version

This will actually fail on the “ASM disk size” check but this can be safely ignored. A manual check of the disk size shows that they’re at least all the same size within their own diskgroup –

SQL> select distinct group_number,total_mb from v$asm_disk;

------------ ----------
0 0
1 491520
2 75080
3 70004

After that, you can simply execute runInstaller to perform the upgrade which should run through cleanly. I chose to include both nodes in batch1 and provided a root password to automate the running of the root scripts. I also chose to configure the Grid Infrastructure Management Repository so that all of the 12c features would be available to me later.

Interestingly, this repository database gets created as follows –

[root@oda-nodea ~]# ps -ef|grep pmon
grid 17264 1 0 07:46 ? 00:00:00 mdb_pmon_-MGMTDB

[root@oda-nodea ~]# ls -l /proc/17264/fd/|grep u01
lrwx------ 1 grid oinstall 64 Sep 6 07:54 11 -> /u01/app/
lrwx------ 1 grid oinstall 64 Sep 6 07:54 14 -> /u01/app/
lr-x------ 1 grid oinstall 64 Sep 6 07:54 15 -> /u01/app/
lrwx------ 1 grid oinstall 64 Sep 6 07:54 3 -> /u01/app/
l-wx------ 1 grid oinstall 64 Sep 6 07:54 4 -> /u01/app/
lr-x------ 1 grid oinstall 64 Sep 6 07:54 8 -> /u01/app/

It is running out of the GI home as the grid user.

You can now modify the compatibility of the ASM diskgroups so that some of the new features can be used during database creation. I’m leaving database compatibility as because I want to leave our old databases running out of the 11g RDBMS home.

SQL> alter diskgroup data set attribute 'compatible.asm' = '';

Diskgroup altered.

SQL> alter diskgroup reco set attribute 'compatible.asm' = '';

Diskgroup altered.

SQL> alter diskgroup redo set attribute 'compatible.asm' = '';

Diskgroup altered.

I was then able to install the 12c database software into a new home, and create a new database running out of it. I chose to install the password file onto shared storage (a new feature) and ran the database across both nodes.

I don’t really expect OAKCLI to handle these changes very well, but some of the basic stuff does still seem to work.

[root@oda-nodeb ~]# oakcli show databases
Database Name Database Type Database HomeName Database HomeLocation Database Version
---------------- ----------- ---------------- --------------------------------------- ---------------------
w0l0sint RAC OraDb11203_home1 /u01/app/oracle/product/,16742216)
ODADB RAC OraDb11203_home1 /u01/app/oracle/product/,16742216)
_mgmtdb 0
cdb RAC OraDB12Home1 /u01/app/oracle/product/

Post a comment or leave a trackback: Trackback URL.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: