techblog 

Automatic Storage Management (ASM) is a special-purpose Oracle filesystem cum volume manager. It first appeared with 10g R1. Diskgroups are managed through an additional instance (similiar to a normal database instance), called the ASM instance. This ASM instance has an associated SID, background parameter and spfile associated with it, any database instances that have their datafiles stored via ASM, register themselves with the ASM instance, so if ASM is not running the database will not be running either.

The real advantage with ASM is when it comes to RAC, instead of pure RAW files, you can use ASM to store your clustered database files on, without having to use a 3rd party Cluster Filesystem. I have found a nice little feature that may bite the unaware if using ASM with RAC.

Using ASMLib I mark a disk for ASM usage:

(oracle$) /etc/init.d/oracleasm createdisk VOL1 /dev/sdb1

Now create a diskgroup from this volume:

sql> create diskgroup DATA1 external redundancy disk 'ORCL:VOL1';

You can now check that your diskgroup is ready to use:

SQL>select name, state, type, total_mb, free_mb from v$asm_diskgroup;

NAME     STATE       TYPE     TOTAL_MB    FREE_MB
DATA1    MOUNTED     EXTERN     153597     153503

However, if you are using RAC, look on the other node, it has automatically picked up the fact there is a new diskgroup created but you’ll find the state is DISMOUNTED. I tried doing a shutdown/startup on the ASM instance on the other node, but this gave an oracle error, ORA-15110. Looking this error up, is not very instructive, and the diskgroup while still visible was in the DISMOUNTED state.

What you have to do on the 2nd node, is to mount it manually the first time, then on subsequent reboots it will get mounted automatically:

sql>  alter diskgroup DATA1 mount;

Now all is well on node 2. So in a RAC cluster, the ASM instance on the 2nd node will not automatically mount a newly created diskgroup.