Create a NSS shared pool and volume on one of the nodes in the cluster, then specify the shared IP address where MySQL service will run.

By default, this shared volume is now mounted and available at the server where it was created at /media/nss/NSSVOL where NSSVOL is the shared volume.

Create an eDirectory™ user named mysql1 by executing the following command:

namuseradd -x o=user_context -g cn=admingroup,o=user_context mysql1

Prepare MySQL service to use the shared volume.

(Optional) If migrating datadir, skip this step and go to Step 5, otherwise, execute the following:

mysql_install_db --datadir=/media/nss/NSSVOL/mysql/data --user=mysql1

Go to iManager > Cluster > Cluster Config, then edit the load and unload script for the following cluster volume resource object.

The load script specifies the commands to start the MySQL service with the datadir on the shared volume. For example, the modified load script will be similar to the following sample:

#!/bin/bash
. /opt/novell/ncs/lib/ncsfuncs
# mount the NSS volume and add secondary IP address
exit_on_error nss /poolact=NSSPOOL
exit_on_error ncpcon mount NSSVOL=254
exit_on_error add_secondary_ipaddress a.b.c.d
exit_on_error ncpcon bind --ncpservername=TESTCLUST_IPOOL_SERVER --ipaddress=a.b.c.d
# start the service
echo "Starting MySQL server ..."
exit_on_error startproc -v /usr/bin/mysqld_safe --defaults-file=/media
/nss/NSSVOL/mysql/my.cnf --datadir=/media/nss/NSSVOL/mysql/data --bind-address=a.b.c.d --user=mysql1 > /dev/null 2>&1
echo "Sucessfully started MySQL server"
# return status
exit 0

The unload script specifies how the application or resource should terminate. For example, the modified unload script will be similar to the following sample

#!/bin/bash
. /opt/novell/ncs/lib/ncsfuncs
# stop service
ignore_error /usr/bin/mysqladmin shutdown --bind-address=a.b.c.d
-u root -ppassword
echo "Stopped MySQL server"
sleep 5
# delete the IP address and deactivate NSS volume
ignore_error ncpcon unbind --ncpservername=TESTCLUST_IPOOL_SERVER --ipaddress=a.b.c.d
ignore_error del_secondary_ipaddress a.b.c.d
ignore_error nss /pooldeact=IPOOL
# return status
exit 0

where a.b.c.d = clustered POOL IP address

Bring the resource offline and then bring it online.

Create an REISERFS file system partition of the required size. Do not configure the mount point; leave it blank.If you configure the mount point, an entry is created in fstab and the partition is mounted automatically after the system reboots.

Use fdisk -l to view the newly created partition. For example, it might be listed as /dev/sdc1.

Execute the following commands on the server where you created the partition:

mkdir /mnt/mysql

mount -t reiserfs /dev/sdc1 /mnt/mysql

cp /etc/my.cnf /mnt/mysql

mkdir /mnt/mysql/data

Edit the /mnt/mysql/my.cnf file and comment out the following entry:

datadir=

Execute the following commands:

chmod 555 /mnt/mysql/my.cnf

chown -R mysql /mnt/mysql

mysql_install_db --datadir=/mnt/mysql/data --user=mysql

chown -R mysql /mnt/mysql/data

Go to iManager > Cluster > Cluster Config and create a cluster resource.

Edit the load/unload script of the newly created resource object.

#!/bin/bash
. /opt/novell/ncs/lib/ncsfuncs
# mount the file system
ignore_error mkdir -p /mnt/mysql
exit_on_error mount -t reiserfs /dev/sdc1 /mnt/mysql
# add the IP address
exit_on_error add_secondary_ipaddress a.b.c.d
# start the service
chown -R mysql /mnt/mysql
echo "Starting MySQL server ..."
exit_on_error startproc -v /usr/bin/mysqld_safe --defaults-file=/mnt/mysql/my.cnf --datadir=/mnt/mysql/data --bind-address=a.b.c.d --socket=/mnt/mysql/mysql.sock --user=mysql > /dev/null 2>&1
echo "Sucessfully started MySQL server"
# return status
exit 0

#!/bin/bash
. /opt/novell/ncs/lib/ncsfuncs
# request service stop
ignore_error /usr/bin/mysqladmin shutdown --socket=/mnt/mysql/mysql.sock
echo "Stopped MySQL server"
sleep 5
# del the IP address
ignore_error del_secondary_ipaddress a.b.c.d
# umount the file system
exit_on_error umount /mnt/mysql
# return status
exit 0

MySQL for Linux is now configured to work in the active/passive clustering mode on Linux with its data directory present on a cluster-enabled traditional Linux file system.