Let’s imagine you are a wise DBA that has to move some Oracle 10g databases lying on a RAC + ASM   to a new storage.

If you can use Google and Oracle support site you will not have problems : the online method suggested by BrianWood in oracle forums, based on ASM rebalancing works perfectly even in Oracle 10g (note that  the rebalance power ranges from 0 to 11 in Oracle 10g) and even for large disk groups (  I am talking about figures close to one terabyte  ).

To move OCR and Voting disks you can rely on Support note [ID 428681.1] … acually you could rely unless you are unlucky enough to have RAC installed on Win Server 2003 R2 64bit  (it was NOT my choice).

In this case, the unlucky DBA, after naming the OCR disk with guioracleobjmanager.exe and formatting with logpartformat will get the following error when invoking ocrconfig replace ocr \\.\<new_disk>    :

PROT-21: Invalid parameter

The problem here is that even when you commit and sync all nodes on guioracleobjmanager on one node, the other nodes are not fully aware of the new disk!

The only way to make them aware is restart the OracleObjectService (and so to stop and restart the crs ) on all the other nodes.

WATCH OUT: Do not  issue “ocrconfig replace ocr \\.\<new_disk>”  when any of the nodes is down  (I mean crs is down) otherwhise it will not be able to join the cluster anymore! ( I did it and I had to revert back to the old ocr disks to fix the problem)

Advertisements