Wednesday 07 May, 2008 - 20:47
Julian Dyke points to Raw Devices on Linux (Note:224302.1). Here the ownership should be oracle:dba. Still no help.
Changed all the raw devices for ocr and voting disks to have ownership oracle:oinstall and permissions 660. To get past the voting disk screen, I created a cluster called richmond with only one (1) member (richmond1). The software was then installed on richmond1 successfully. The first root script ($ORACLE_BASE/oraInventory/orainstRoot.sh) run successfully, but the second one ($ORACLE_HOME/root.sh) failed:
# /u00/crs/oracle/product/10/app/root.sh
Checking to see if Oracle CRS stack is already configured
/etc/oracle does not exist. Creating it now.
Setting the permissions on OCR backup directory
Setting up NS directories
Oracle Cluster Registry configuration upgraded successfully
assigning default hostname richmond1 for node 1.
Successfully accumulated necessary OCR keys.
Using ports: CSS=49895 CRS=49896 EVMC=49898 and EVMR=49897.
node :
node 1: richmond1 richmond1-priv richmond1
Creating OCR keys for user 'root', privgrp 'root'..
Operation successful.
Now formatting voting device: /dev/raw/raw2
Failure -2 opening file handle for (raw2)
Failed to initialize Oracle Cluster Registry for cluster
I examined the script to see where these messages came out from, but could not see where they did. Tried running with sh -x. Only managed to get the image running at 100%. Powered off and on the image. Still got 100%. Deleted the image, and cloned richmond1 from RHEL 3.7 1GB again.
This experiment has been very interesting in that the problems with /dev/raw/raw2 occur when richmond2 is in the cluster definition. I will need to find out a better way of detecting the underlying error.