Wednesday 07 May, 2008 - 23:35
Rebuilding richmond1
Used kudzu to configure eth0 and eth1 using the design from Richmond Cluster (02)
# vi /etc/sysconfig/network to change the host name to richmond1.yaocm.id.au
Created /etc/sysconfig/network-scripts/route-eth0 (see Richmond Cluster (04) for details). Copied to route-eth1
Modified the IP network names from Richmond Cluster (04) in accordance with the reasons given in Richmond Cluster (07)
# cat >/etc/hosts
192.168.100.65 richmond1 richmond1.yaocm.id.au
# Do not remove the following line, or various programs
# that require network functionality will fail.
127.0.0.1 localhost.localdomain localhost
192.168.100.33 richmond1-vip richmond1-vip.yaocm.id.au
192.168.100.34 richmond2-vip richmond2-vip.yaocm.id.au
192.168.100.66 richmond2 richmond2.yaocm.id.au
192.168.100.97 central central.yaocm.id.au
192.168.101.1 richmond1-priv richmond1-priv.yaocm.id.au
192.168.101.2 richmond2-priv richmond2-priv.yaocm.id.au
Created the oracle user and associated groups (see Richmond Cluster (04) for details).
Continued to used device name mappings for raw devices as explained in Richmond Cluster (09) instead of the design proposed in Richmond Cluster (02)
# cat >>/etc/sysconfig/rawdevices
/dev/raw/raw1 /dev/sdb1
/dev/raw/raw2 /dev/sdb2
/dev/raw/raw16 /dev/sdc1
/dev/raw/raw17 /dev/sdc2
/dev/raw/raw31 /dev/sdd1
/dev/raw/raw32 /dev/sdd2
Maintaining the same security as described in Richmond Cluster (10)
# chown oracle:oinstall /dev/raw/raw*
# chmod 660 /dev/raw/raw*
The major change is to the disk partitioning of the shared disk drives. Instead of using gpt, now using msdos. When I entered parted, I got the error message about the corrupted partition (see Richmond Cluster (09)) only on /dev/sdb and /dev/sdc but not /dev/sdd. The difference being, I imagine, that the ocr and ocr mirror were put onto the first two (2) devices but not the latter.
# parted /dev/sdb
(parted) mklabel msdos
(parted) mkpart primary 1 257
(parted) mkpart primary 257 513