RAC database / disk persistence
I have built an Oracle 10g RAC database on two nodes and a shared storage connected using a FC switch and the system is up and running.
We have 16 disks on the storage and have used them for Oracle files (OCR, voting disk files, and database files). We are using ASM disk groups for database files and flashback recovery area. We added a line similar to the following in /etc/sysconfig/rawdevices file for each partition that was created:
/dev/raw/raw1 /dev/sdb1
Our /etc/sysconfig/rawdevices file looks like the following:
/dev/raw/raw1 /dev/sdb1
/dev/raw/raw2 /dev/sdb2
/dev/raw/raw3 /dev/sdc1
/dev/raw/raw4 /dev/sdc2
/dev/raw/raw5 /dev/sdc3
/dev/raw/raw6 /dev/sdd
/dev/raw/raw7 /dev/sde
/dev/raw/raw8 /dev/sdf
/dev/raw/raw9 /dev/sdg
/dev/raw/raw10 /dev/sdh
/dev/raw/raw11 /dev/sdi
/dev/raw/raw12 /dev/sdj
/dev/raw/raw13 /dev/sdk
/dev/raw/raw14 /dev/sdl
/dev/raw/raw15 /dev/sdm
/dev/raw/raw16 /dev/sdn
/dev/raw/raw17 /dev/sdo
/dev/raw/raw18 /dev/sdp
/dev/raw/raw19 /dev/sdq
The question I have is do we have to worry about the disk persistence? For example, I was told that if one disk fails, the device names could bump up, like if sdd fails, sde could become sdd, sdf could become sde, etc. Also I was told that if a failed disk is replaced by a good one, the OS could change the existing device names. Because of this device name changes, all raw1 through raw19 could point to disks that were not what they were configured to originally. Is this an issue? Or are the steps in the guide just to get Oracle software installed. After that, Oracle Clusterware and ASM know what each disk is for and take care of the device name changes if any?
Thank you,
Jiong
